molecular phylogenetics of fouquieriaceae: evidence from nuclear ...
molecular phylogenetics of fouquieriaceae: evidence from nuclear ...
molecular phylogenetics of fouquieriaceae: evidence from nuclear ...
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
American Journal <strong>of</strong> Botany 86(4): 578–589. 1999.<br />
MOLECULAR PHYLOGENETICS OF FOUQUIERIACEAE:<br />
EVIDENCE FROM NUCLEAR RDNA ITS STUDIES 1<br />
LISA M. SCHULTHEIS 2 AND BRUCE G. BALDWIN<br />
Department <strong>of</strong> Integrative Biology and Jepson Herbarium, University <strong>of</strong> California, Berkeley, California 94720<br />
Molecular sequence data <strong>from</strong> the 18S–26S rDNA internal transcribed spacer (ITS) region support the monophyly <strong>of</strong><br />
Fouquieria sensu lato (Fouquieriaceae) and the three subgenera (subg. Fouquieria, subg. Bronnia, subg. Idria) previously<br />
recognized within it. Resolution within subg. Fouquieria differs somewhat between parsimony and maximum likelihood<br />
(ML) trees. Section Fouquieria and sect. Ocotilla within subg. Fouquieria are not well supported as monophyletic groups.<br />
Uncertainty regarding placement <strong>of</strong> the root within Fouquieriaceae makes discussion <strong>of</strong> character evolution within the family<br />
difficult. Three root positions are consistent with rate-constant evolution <strong>of</strong> ITS sequences: (1) along the branch to subg.<br />
Idria, (2) along the branch to subg. Bronnia, and (3) along the branch to subg. Fouquieria. The first root position listed is<br />
equivalent to an outgroup rooting. The third root position listed is equivalent to a midpoint rooting. Of the three root<br />
positions above, only the third is along a branch that may be sufficiently long to act as a long-branch attractor. The first<br />
two root positions would result in character reconstruction suggesting that succulent growth forms and white floral pigmentation<br />
are ancestral within the family, with shifts to woody growth forms and to red floral pigmentation. The third root<br />
position results in equivocal reconstruction <strong>of</strong> the ancestral growth form, equivocal reconstruction <strong>of</strong> ancestral floral pigmentation<br />
in parsimony trees, and a suggestion <strong>of</strong> white floral pigmentation as ancestral in ML trees. Two previous hypotheses<br />
<strong>of</strong> polyploid origins are compatible with the <strong>molecular</strong> data presented here: (1) origin <strong>of</strong> the tetraploid F. diguetii<br />
<strong>from</strong> F. macdougalii, and (2) allopolyploid origin <strong>of</strong> the hexaploid F. burragei <strong>from</strong> the tetraploid F. diguetii and a diploid<br />
species similar to F. splendens. Direct descent <strong>of</strong> the hexaploid F. columnaris <strong>from</strong> the subg. Bronnia lineage is not supported<br />
by our data. An amphiploid origin <strong>of</strong> F. columnaris involving a member <strong>of</strong> the subg. Bronnia lineage and an extinct taxon<br />
outside subg. Bronnia, however, cannot be ruled out.<br />
Key words: boojum; Fouquieria; Fouquieriaceae; ITS; long-branch attraction; ocotillo; rooting.<br />
Fouquieriaceae is a small and well-diagnosed family<br />
<strong>of</strong> flowering plants distributed throughout the warm deserts<br />
<strong>of</strong> Mexico and the southwestern United States.<br />
Unique features <strong>of</strong> Fouquieriaceae include placentation<br />
that changes <strong>from</strong> parietal to axile during fruit development,<br />
decurrent spines formed <strong>from</strong> the petioles <strong>of</strong> primary<br />
leaves, and an anastomosing network <strong>of</strong> cortical<br />
water storage tissue (Scott, 1932; Henrickson, 1969a,<br />
1972). Relationships within the family have been examined<br />
most thoroughly by Henrickson (1972) using morphological,<br />
cytological, and ecological data. From this<br />
work, Henrickson proposed the recognition <strong>of</strong> one genus,<br />
Fouquieria (including Idria), three subgenera (Fouquieria,<br />
Bronnia, and Idria), and 11 species in Fouquieriaceae<br />
(see Table 1). We generated phylogenetic hypotheses for<br />
1 Manuscript received 10 March 1998; revision accepted 10 September<br />
1998.<br />
The authors thank Sarah Vetault and Dr. Michael Donoghue (Harvard<br />
University) for providing DNA samples; Dr. Mark Porter (Rancho Santa<br />
Ana Botanic Garden), Staci Markos (U. C. Berkeley), and Lena Hileman<br />
(Harvard University) for providing DNA sequences; and Bob Perill<br />
(Boojum Unlimited, Tucson, AZ), Dr. Richard Felger (Drylands Institute,<br />
Tucson, AZ), and Cathy Babcock (Desert Botanic Garden, Phoenix,<br />
AZ) for additional plant material. We thank Dr. John Huelsenbeck<br />
(University <strong>of</strong> Rochester) for his extensive help and advice regarding<br />
<strong>molecular</strong> sequence simulations and use <strong>of</strong> his Siminator program, and<br />
Dr. David Sw<strong>of</strong>ford (Smithsonian Institution) for granting us permission<br />
to use and publish results <strong>from</strong> a test version <strong>of</strong> PAUP* 4.0. We also<br />
thank Staci Markos, Dr. Mark Porter, and two anonymous reviewers for<br />
helpful comments on the manuscript. This work was supported in part<br />
by funding <strong>from</strong> the A. W. Mellon Foundation through the Botany Department<br />
at Duke University and the Lawrence R. Heckard Fund <strong>of</strong> the<br />
Jepson Herbarium at U. C. Berkeley.<br />
2 Author for correspondence.<br />
578<br />
Fouquieriaceae based on <strong>molecular</strong> data <strong>from</strong> the 18S–<br />
26S rDNA internal transcribed spacer (ITS) region. These<br />
trees were used to evaluate the current taxonomy <strong>of</strong> the<br />
group, as proposed by Henrickson and to examine hypotheses<br />
<strong>of</strong> character evolution.<br />
Henrickson’s (1972) subgenera <strong>of</strong> Fouquieriaceae<br />
roughly correspond to growth forms. Subgenus Idria and<br />
subg. Bronnia comprise the succulent taxa; subg. Fouquieria<br />
contains the woody taxa. The succulent species<br />
achieve their growth form by an increase in parenchymatous<br />
water storage tissue within the secondary xylem<br />
and, in F. columnaris, the pith (Humphrey, 1935; Henrickson,<br />
1969c). Subgenus Idria contains F. columnaris<br />
(boojum tree), a columnar plant with a succulent trunk<br />
reaching a height <strong>of</strong> up to 40 feet (Humphrey, 1935).<br />
Subgenus Bronnia contains F. fasciculata and F. purpusii,<br />
both <strong>of</strong> which have basally succulent trunks. Subgenus<br />
Fouquieria contains eight woody (or dendroid)<br />
species, split into sect. Fouquieria and sect. Ocotilla. The<br />
six species <strong>of</strong> sect. Fouquieria bear distinct trunks and<br />
branching stems. The two species <strong>of</strong> sect. Ocotilla have<br />
a reduced trunk and vertical unbranched stems. In his<br />
comparative anatomical study <strong>of</strong> F. columnaris and F.<br />
splendens, Humphrey (1935) suggested that members <strong>of</strong><br />
Fouquieriaceae cannot be readily classified as shrubs or<br />
as stem-succulents, the two predominant types <strong>of</strong> xerophytic<br />
plants he recognized within the southwestern<br />
North American deserts. The shrub forms <strong>of</strong> subg. Fouquieria<br />
possess a degree <strong>of</strong> cortical succulence, and the<br />
succulent forms are not succulent throughout, even in F.<br />
columnaris, which possesses nonsucculent branches.<br />
Growth forms in the family are intermediate between typical<br />
xeromorphic types.
April 1999] SCHULTHEIS AND BALDWIN—MOLECULAR PHYLOGENETICS OF FOUQUIERIACEAE<br />
579<br />
TABLE 1. Classification <strong>of</strong> Fouquieriaceae, following Henrickson<br />
(1972).<br />
Fouquieriaceae A.P. de Candolle<br />
Fouquieria H.B.K.<br />
Subgenus Fouquieria<br />
Section Fouquieria<br />
F. leonilae F. Miranda<br />
F. ochoterenae F. Miranda<br />
F. macdougalii Nash<br />
F. diguetii (van Tiegh.) I.M. Johnston<br />
F. burragei Rose<br />
F. formosa H.B.K.<br />
Subgenus Fouquieria<br />
Section Ocotilla Henrickson<br />
F. splendens Engelm. in Wisliz.<br />
ssp. splendens<br />
ssp. campanulata (Nash) Henrickson<br />
var. campanulata<br />
var. albiflora Henrickson<br />
ssp. breviflora Henrickson<br />
F. shrevei I.M. Johnston<br />
Subgenus Bronnia (H.B.K.) Henrickson<br />
F. fasciculata (Willd. ex Roem. et Schult.) Nash<br />
F. purpusii T.S. Brandegee<br />
Subgenus Idria (Kellogg) Henrickson<br />
F. columnaris (Kellogg) Kellogg ex Curran<br />
The base chromosome number in Fouquieriaceae is x<br />
12 (Henrickson, 1972). Both Fouquieria columnaris<br />
and F. burragei are hexaploids (n 36), F. diguetii is a<br />
tetraploid (n 24), and the remaining species are diploid<br />
(n 12) (chromosome counts for F. formosa are lacking).<br />
Henrickson (1972) suggested that F. burragei is part<br />
<strong>of</strong> a polyploid series including F. macdougalii and the<br />
tetraploid F. diguetii. He suggested that F. diguetii may<br />
have descended <strong>from</strong> ‘‘F. macdougalii stock’’ and that<br />
F. burragei may have been <strong>of</strong> amphiploid origin with<br />
possible parents including F. diguetii and ‘‘some presumably<br />
now extinct diploid species having numerous stamens,<br />
possibly short, white corollas and spicate or racemose<br />
inflorescences’’ (Henrickson, 1972, p. 499). Similarly,<br />
a species <strong>from</strong> the F. purpusii and F. fasciculata<br />
lineage (subg. Bronnia) may have served as a parent in<br />
an amphiploid derivation <strong>of</strong> F. columnaris (subg. Idria)<br />
(Henrickson, 1972).<br />
With regard to floral characteristics, just over half <strong>of</strong><br />
Fouquieriaceae species possess stereotypical hummingbird-pollinated<br />
flowers, with tubular red corollas containing<br />
nectar at the base. All red-flowered species are in<br />
subg. Fouquieria. Also in subg. Fouquieria are F. burragei<br />
and F. shrevei, both <strong>of</strong> which have corollas that are<br />
pink in bud and white to pink at maturity (Henrickson,<br />
1972). Members <strong>of</strong> subg. Bronnia and subg. Idria have<br />
corollas that are cream yellow to white throughout their<br />
development (Henrickson, 1972). Members <strong>of</strong> these two<br />
subgenera, as well as F. shrevei, lack floral anthocyanins<br />
(Scogin, 1977). In addition to floral pigmentation, flowers<br />
<strong>of</strong> Fouquieria taxa differ in the number <strong>of</strong> stamens, the<br />
degree <strong>of</strong> corolla limb reflexion, and their pattern <strong>of</strong> arrangement<br />
in inflorescences.<br />
Fouquieriaceae has been considered closely related to<br />
various families and orders since its establishment by De<br />
Candolle in 1828. De Candolle suggested an affinity with<br />
Portulacaceae, as did Humboldt, Bonpland, and Kunth<br />
when they first described the genus Fouquieria in 1823<br />
(for expanded discussion and references see Henrickson<br />
[1967, 1972]). The hypothesized affinity <strong>of</strong> Fouquieriaceae<br />
with Portulacaceae is untenable because Caryophyllales<br />
is diagnosed by numerous characteristics (Mabry,<br />
1977) lacking in Fouquieriaceae. The type specimen<br />
for Fouquieria was originally placed in Cantua (Polemoniaceae)<br />
by Roemer and Schultes (1819). An affinity<br />
with Polemoniaceae was also suggested by Nash (1903),<br />
in the first treatment <strong>of</strong> Fouquieriaceae, and by Henrickson<br />
(1967, based on pollen comparisons), the author <strong>of</strong><br />
the most recent treatment. Additional suggestions for the<br />
position <strong>of</strong> Fouquieriaceae have included alignment with<br />
Tamaricales (e.g., Hutchinson, 1926; Takhtajan, 1969),<br />
Violales (e.g., Cronquist, 1981), Ebenales (e.g., Bessey,<br />
1915; Morton et al., 1996), Solanales (e.g., Thorne, 1969;<br />
Scogin, 1977; both included Polemoniaceae within Solanales),<br />
and Ericales (e.g., Dahlgren, Jensen, and Nielsen,<br />
1976; Jensen and Nielson, 1982; Hufford, 1992;<br />
Olmstead et al., 1993). The Ericales clade in which Fouquieriaceae<br />
has been placed in <strong>molecular</strong> trees is broadly<br />
circumscribed, including Ebenales taxa among others<br />
(Olmstead et al., 1993). Anderberg (1992) placed Fouquieriaceae<br />
at the base <strong>of</strong> Asteridae but noted that the<br />
family could shift closer to the Ericales clade with increased<br />
taxon sampling. The general pattern that has<br />
emerged <strong>from</strong> various phylogenetic analyses is placement<br />
<strong>of</strong> Fouquieriaceae at the base <strong>of</strong> Asteridae, perhaps within<br />
a broadly circumscribed ericalean grouping.<br />
MATERIALS AND METHODS<br />
Total DNA was isolated <strong>from</strong> 24 specimens (Table 2), following a<br />
minor modification <strong>of</strong> the CTAB protocol <strong>of</strong> Doyle and Doyle (1987),<br />
and purified on CsCl 2 gradients. Two chlor<strong>of</strong>orm/isoamyl alcohol extractions<br />
and two ethanol precipitations (following isopropanol precipitation)<br />
were added to Doyle and Doyle’s method. Leaf material was<br />
dried and stored in silica gel desiccant prior to DNA isolation. Most<br />
DNAs <strong>from</strong> Fouquieriaceae taxa were generously provided by Sarah<br />
Vetault (University <strong>of</strong> Arizona) and Michael Donoghue (Harvard University),<br />
with subsequent collections obtained to provide additional representation<br />
within each species, when possible. Outgroup taxa were chosen<br />
to represent the various orders with which Fouquieriaceae have been<br />
aligned. One ingroup sequence was provided by J. Mark Porter (Rancho<br />
Santa Ana Botanic Garden), and a total <strong>of</strong> nine outgroup sequences<br />
were provided by J. Mark Porter, Staci Markos (U. C. Berkeley), and<br />
Lena Hileman (Harvard University), as noted in Table 2.<br />
Single-stranded DNAs <strong>of</strong> ITS 1 and ITS 2 were generated, purified,<br />
and sequenced following Baldwin (1992). For some taxa, double-stranded<br />
DNAs were generated using a GeneAmp 9600 with the following<br />
conditions: initial denaturation (97C, 1 min), followed by 40 cycles <strong>of</strong><br />
denaturation (97C, 10 s), annealing (48C, 30 s), and extension (72C,<br />
20 s increasing 4 s with each cycle), and concluding with a final extension<br />
(72C, 7 min). The polymerase chain reactions (PCR) contained<br />
the following components: 5.0 L 10X PCR buffer II (Perkin Elmer,<br />
Foster City, California), 2.5 mmol/L MgCl 2, 1.0 mmol/L dNTPs, 2.5<br />
L glycerol, 0.5 mol/L ITS-I primer (5’-GTCCACTGAACCTTAT-<br />
CATTTAG-3’; designed by L. E. Urbatsch, Louisiana State University),<br />
0.5 mol/L ITS4 primer (White et al., 1990), 1.0 unit AmpliTaq DNA<br />
polymerase (Perkin Elmer, Foster City, California), and 1–10 ng DNA,<br />
to a total volume <strong>of</strong> 50 L. PCR products were cleaned using either<br />
Ultrafree-MC 100000 NMWL polysulfone membrane filter units (Millipore<br />
Corporation, Bedford, Massachusetts), or Wizard PCR Preps<br />
DNA Purification System (Promega, Madison, Wisconsin), following<br />
manufacturer’s instructions. Sequencing reactions were performed with
580 AMERICAN JOURNAL OF BOTANY<br />
[Vol. 86<br />
TABLE 2. Source <strong>of</strong> taxa used. a<br />
Taxon Source<br />
Family Cyrillaceae<br />
Cyrilla racemiflora L. Schultheis 2–94; DUKE<br />
Family Ericaceae<br />
Arctostaphylos nummularia A. Gray<br />
A. uva-ursi (L.) Sprengel<br />
Arbutus menziesii Pursh<br />
Lyonia lucida (Lam.) K. Koch<br />
Vaccinium fuscatum Ait.<br />
Family Fouquieriaceae<br />
Fouquieria burragei Rose (1) e<br />
F. burragei (2)<br />
F. columnaris (Kellogg) Kellogg ex Curran (1) e<br />
F. columnaris (2)<br />
F. diguetti (van Teigh.) I.M. Johnston (1) e<br />
F. diguetti (2)<br />
F. fasciculata (Willd. ex Roem. et Schult.) Nash (1)<br />
F. fasciculata (2) e<br />
F. fasciculata (3)<br />
F. formosa H.B.K. e<br />
F. leonilae F. Miranda (1) e<br />
F. leonilae (2)<br />
F. macdougalii Nash (1) e<br />
F. macdougalii (2)<br />
F. ochoterenae F. Miranda e<br />
F. purpusii T.S. Brandegee e<br />
F. shrevei I.M. Johnston (1) e<br />
F. shrevei (2)<br />
F. splendens Engelm. in Wisliz. (1) e<br />
F. splendens (2)<br />
Family Polemoniaceae<br />
Acanthogilia gloriosa (Brandeg.) Day & Moran<br />
Gen et sp. nov. (Porter, in prep)<br />
Bonplandia geminiflora Cav.<br />
Cantua quericifolia Cav.<br />
Cobaea baiurita Standley<br />
Loeselia glandulosa (Cav.) Don.<br />
V. T. Parker & M. Vasey 0398; SFSU b<br />
V. T. Parker & M. Vasey 0046; SFSU b<br />
University <strong>of</strong> California, Berkeley Botanic Garden c<br />
Schultheis 1–94; DUKE<br />
Schultheis, n.v. d<br />
Perill f , p.v. g<br />
DBG h 1993-0948 (Henrickson 9113)<br />
C. T. Mason 136413 i<br />
DBG 1939–0072 (G. Lindsey & R. Hoard), p.v.<br />
Perill, p.v.<br />
DBG 1939–0074 (G. Lindsey sn), p.v.<br />
DBG 1977–1444, p.v.<br />
Perill, p.v.<br />
DBG 1974–0221, p.v.<br />
MMBG h 66–035 (Henrickson 2108)<br />
MMBG 66–036 (Henrickson 2164; RSA)<br />
DBG 1993001501 (Henrickson 4236; RSA)<br />
Perill, p.v.<br />
DBG 1965–7944, p.v.<br />
Perill, p.v.<br />
Perill, p.v.<br />
R. Felger j , p.v.<br />
DBG 1974–0154 (R. Engard sn), p.v.<br />
n.v.<br />
J.M. Porter sn; SJNM k<br />
J.M. Porter & K.D. Heil 7987; SJNM k<br />
J.M. Porter & K. Heil 7991; SJNM k (Gilia scabra Brandegee;<br />
see Porter, 1996)<br />
R. Patterson sn; RSA k<br />
R. Patterson sn; RSA k<br />
O. Clarke 293; AZ k<br />
J.M. Porter & C. Campbell 9231; AZ, SJNM k<br />
Family Symplocaceae<br />
Symplocus tinctoria L Hérit Schultheis 7–94; DUKE<br />
Family Tamaricaceae<br />
Tamarix L. sp. Schultheis 19–94; DUKE<br />
a Sequences <strong>from</strong> all samples are available in GenBank (accession numbers GBANAF084314–GBANAF084361 [the prefix GBAN links the online<br />
version <strong>of</strong> AJB with GenBank and is not part <strong>of</strong> the actual GenBank accession number]) with the exception <strong>of</strong> sequences provided by others,<br />
as noted. The sequence alignment is available <strong>from</strong> L. Schultheis upon request.<br />
b Sequence provided by S. Markos, University <strong>of</strong> California at Berkeley (see Markos, 1995).<br />
c Sequence provided by L. Hileman, Harvard University (GenBank accession number GBANAF086828).<br />
d n.v. no voucher<br />
e Genomic DNA received <strong>from</strong> S. Vetault (University <strong>of</strong> Arizona) and M. Donoghue (Harvard University), who received material <strong>from</strong> the source<br />
indicated.<br />
f Perill Material received <strong>from</strong> Bob Perill, Boojum Unlimited, Tucson, AZ 85743.<br />
g p.v. photo voucher, deposited at UC.<br />
h The botanic garden accession number is given, with a collector and/or collection number and place <strong>of</strong> deposition following in parentheses if<br />
that information is available. DBG Desert Botanical Gardens <strong>of</strong> Arizona, Phoenix, AZ. MMBG Mildred Mathias Botanical Garden, Los<br />
Angeles, CA.<br />
i Material taken <strong>from</strong> the University <strong>of</strong> Arizona cactus garden. The cited collection, deposited at ARIZ, was also taken <strong>from</strong> the cactus garden.<br />
j Material received <strong>from</strong> Dr. Richard Felger, Drylands Institute, Tucson, AZ<br />
k Sequence provided by J.M. Porter, Rancho Santa Ana Botanic Garden (see Porter, 1996).<br />
an Applied Biosystems, Inc. (ABI) PRISM Dye Terminator Cycle Sequencing<br />
Ready Reaction Kit (Foster City, California). Sequencing<br />
products were cleaned with Centri-sep spin columns (Princeton Separations,<br />
Adelphia, New Jersey), and electrophoresed on 4% polyacrylamide<br />
gels using an ABI 377 automated sequencer. Sequences were<br />
visualized using ABI Sequence Navigator s<strong>of</strong>tware.<br />
Alignment <strong>of</strong> ITS 1 and ITS 2 sequences was conducted separately.<br />
Sequences <strong>from</strong> the 5.8S region were not included. Visual alignment <strong>of</strong><br />
sequences within Fouquieriaceae was achieved readily. Sequence alignment<br />
with the outgroups was attempted using Clustal V (Higgins, 1994)<br />
with fixed and floating gap penalties <strong>of</strong> ten. Sequence divergence between<br />
Fouquieriaceae and potential outgroups was substantial, ranging
April 1999] SCHULTHEIS AND BALDWIN—MOLECULAR PHYLOGENETICS OF FOUQUIERIACEAE<br />
581<br />
TABLE 3. Results <strong>of</strong> test for rate constancy <strong>of</strong> ITS sequence evolution by nucleotide substitution in Fouquieria with placement <strong>of</strong> the root in all<br />
possible positions. a<br />
FC f<br />
FC,FFA,FP<br />
FFA,FP<br />
FP<br />
FFA<br />
FF<br />
FO<br />
FL<br />
FB(2)<br />
FS(2)<br />
FSH<br />
FM<br />
FB<br />
FD<br />
FF,FP,FFA,FC<br />
Root b With clock c Without clock 2lnLR d Reject clock e<br />
FC,FFA,FP,FO,FL,FF<br />
FL,FO<br />
FS(2),FSH<br />
FD,FB<br />
1121.91562<br />
1121.59848<br />
1121.91924<br />
1139.11882<br />
1139.25322<br />
1131.54298<br />
1139.42992<br />
1139.41329<br />
1141.49784<br />
1142.50684<br />
1142.54318<br />
1140.37033<br />
1156.27012<br />
1156.24244<br />
1131.54288<br />
1136.29718<br />
1136.28829<br />
1141.50706<br />
1140.33975<br />
1118.27553 7.28<br />
6.65<br />
7.29<br />
41.69<br />
a The tree topology used was derived <strong>from</strong> a branch-and-bound search <strong>of</strong> the listed samples with the parsimony criterion (analysis 5 <strong>of</strong> Materials<br />
and Methods).<br />
b Root the root is placed along the branch between the sample(s) listed and the remaining samples.<br />
c With and without clock ln likelihood values with and without enforcement <strong>of</strong> a <strong>molecular</strong> clock.<br />
d LR likelihood ratio; the difference between ln likelihood values with and without enforcement <strong>of</strong> a clock.<br />
e Rejection <strong>of</strong> the clock is based on chi-square values (df 10, P 0.05, CV 18.31)<br />
f FC Fouquieria columnaris; FFAF. fasciculata; FPF. purpusii; FFF. formosa; FO F. ochoterenae; FL F. leonilae; FS F.<br />
splendens; FSH F. shrevei; FMF. macdougalii; FBF. burragei; FDF. diguetii. All sequences were <strong>from</strong> sample (1) as listed in Table<br />
2, unless otherwise indicated.<br />
<strong>from</strong> 33.8 to 50.8%, as was divergence between potential outgroups.<br />
Because <strong>of</strong> these high divergences and alignment difficulties, the full<br />
data set was not used. Instead, a data file was created retaining all<br />
sequence information within Fouquieriaceae but with all ambiguously<br />
aligned regions within the outgroup sequences coded as missing data<br />
(‘‘?’’). This is similar to the approach taken by Bruns et al. (1992) to<br />
enable retention <strong>of</strong> sequence information for those sequences with unambiguous<br />
alignments in otherwise problematic regions.<br />
Phylogenetic analyses were carried out using PAUP version 3.1<br />
(Sw<strong>of</strong>ford, 1993) and test versions d55-d59 <strong>of</strong> PAUP* 4.0 (with permission,<br />
D. L. Sw<strong>of</strong>ford, Smithsonian Institution, personal communication).<br />
Multiple analyses were undertaken, both with and without inclusion<br />
<strong>of</strong> outgroup taxa.<br />
Analysis 1 <strong>of</strong> Fouquieriaceae taxa (without outgroup sequences) employed<br />
a branch-and-bound search with the parsimony criterion to find<br />
all minimum-length (unrooted) trees. Clade support was assessed with<br />
both bootstrap and decay analyses. The bootstrap employed 100 replicates<br />
with branch-and-bound searches. Decay analysis employed a<br />
branch-and-bound search saving all trees up to six steps longer than the<br />
most parsimonious tree. The trees were filtered for progressively shorter<br />
tree lengths and a strict consensus tree was calculated at each tree<br />
length.<br />
Analysis 2 included all taxa but with partial sequence data, the ambiguously<br />
aligned regions in outgroup sequences being coded as missing<br />
data. Complete heuristic searches were conducted with 100 replicates<br />
<strong>of</strong> random taxon addition, TBR branch swapping, and MULPARS<br />
in effect to find all minimum-length trees and to obtain an outgroup<br />
rooting <strong>of</strong> Fouquieriaceae. Clade support was assessed using the ‘‘fast<br />
heuristic’’ bootstrap algorithm available in PAUP*. This algorithm employs<br />
less thorough searches than the standard bootstrap algorithm, but<br />
provides a conservative estimate <strong>of</strong> bootstrap values (M. J. Sanderson,<br />
U. C. Davis, personal communication).<br />
Analysis 3 employed maximum-likelihood (ML) analysis <strong>of</strong> ingroup<br />
taxa under the HKY85 model <strong>of</strong> sequence evolution, both with and<br />
41.95<br />
26.53<br />
42.31<br />
42.28<br />
46.44<br />
48.46<br />
48.53<br />
44.19<br />
75.99<br />
75.93<br />
26.53<br />
36.04<br />
36.02<br />
46.46<br />
44.13<br />
No<br />
No<br />
No<br />
Yes<br />
Yes<br />
Yes<br />
Yes<br />
Yes<br />
Yes<br />
Yes<br />
Yes<br />
Yes<br />
Yes<br />
Yes<br />
Yes<br />
Yes<br />
Yes<br />
Yes<br />
Yes<br />
without enforcement <strong>of</strong> a <strong>molecular</strong> clock to determine whether similar<br />
topologies would be obtained (consistent with rate constancy <strong>of</strong> ITS<br />
evolution) and to compare with results <strong>of</strong> the parsimony analyses. The<br />
12 trees <strong>from</strong> analysis 1 were used as the starting trees for a heuristic<br />
search with TBR branch swapping. Base frequencies were empirically<br />
assessed. The proportion <strong>of</strong> invariant sites was left at the default setting<br />
<strong>of</strong> zero. Transition: transversion () parameters and shape () parameters<br />
were estimated on tree number 1 <strong>from</strong> analysis 1 (6.390220 and<br />
0.193432, respectively), with rate heterogeneity across sites following<br />
a gamma distribution with five rate categories.<br />
Analysis 4 was conducted with ingroup taxa using a parsimony criterion,<br />
but with various weighting schemes, in consideration <strong>of</strong> the high<br />
transition: transversion () ratio estimated on tree number 1 <strong>from</strong> analysis<br />
1. Heuristic searches with 20 replicates <strong>of</strong> random taxon addition,<br />
TBR branch swapping, and MULPARS in effect were conducted. The<br />
transversion:transition weighting schemes employed were 18:10, 2:1,<br />
25:10, 27:10, and 3:1. A weighting scheme <strong>of</strong> 18:10 is recommended<br />
by Albert and Mishler (1992) for an estimated transition:transversion<br />
() ratio <strong>of</strong> 6.4 and lambda value <strong>of</strong> 0.1.<br />
Analysis 5 determined the likelihood values both with and without<br />
enforcement <strong>of</strong> a <strong>molecular</strong> clock for one <strong>of</strong> two trees generated with<br />
a branch-and-bound search <strong>of</strong> ingroup taxa under the parsimony criterion<br />
where the root placement within the ingroup was forced to fall at<br />
each <strong>of</strong> all possible branches. Fewer ingroup sequences were included<br />
in this analysis in order to minimize the number <strong>of</strong> necessary calculations<br />
[FC(1), FFA(1), FP(1), FL(1), FO(1), FM(1), FD(1), FB(1), FB(2),<br />
FS(2), FSH(1), FF; see Table 3 for abbreviations]. Maximum-likelihood<br />
estimates <strong>of</strong> the transition:transversion () ratio (5.756055) and the<br />
shape parameter (0.147754) were generated <strong>from</strong> the same unrooted<br />
topology, with rate heterogeneity following a gamma distribution with<br />
five rate categories. Base frequencies were empirically assessed. Evolutionary<br />
rate heterogeneity across lineages was assessed using a global<br />
likelihood-ratio (LR) test (Felsenstein, 1988; Huelsenbeck and Rannala,<br />
1997).
582 AMERICAN JOURNAL OF BOTANY<br />
[Vol. 86<br />
Analysis 6 used simulated data sets to examine the possibility that<br />
root placement within Fouquieriaceae was due to long-branch attraction<br />
(Huelsenbeck, 1997; see Fig. 4 for a phylogram). One hundred data sets<br />
for each <strong>of</strong> 20 possible root placements within Fouquieriaceae were<br />
simulated using the Siminator program provided by J. P. Huelsenbeck<br />
(University <strong>of</strong> Rochester). The ingroup topology <strong>of</strong> input trees was<br />
equivalent to that <strong>of</strong> tree number 1 in the parsimony analysis <strong>of</strong> only<br />
ingroup taxa (analysis 1). Nucleotide frequencies and size <strong>of</strong> each simulated<br />
data set were kept identical to those found in the data set <strong>from</strong><br />
analysis 2. Transition:transversion () ratios, shape parameters, and<br />
branch lengths were estimated in PAUP* using a maximum-likelihood<br />
criterion for each <strong>of</strong> the input trees on which the simulated data sets<br />
were modeled. Heuristic searches <strong>of</strong> all simulated data sets were conducted<br />
with simple addition <strong>of</strong> sequences and tree-bisection-reconnection<br />
(TBR) branch swapping under a parsimony criterion. Root placements<br />
within Fouquieriaceae in the resulting trees were tallied by filtering<br />
trees with backbone constraints corresponding to each <strong>of</strong> 20 possible<br />
root placements.<br />
MacClade version 3.0 was used to explore character evolution (Maddison<br />
and Maddison, 1992). All characters and character state changes<br />
were equally weighted, and all most parsimonious states were reconstructed<br />
at each node.<br />
RESULTS<br />
Analysis 1, <strong>of</strong> Fouquieriaceae taxa only, resulted in 12<br />
minimum-length trees <strong>of</strong> 85 steps (CI 0.82; 0.78 without<br />
uninformative characters). Fifty-two out <strong>of</strong> 487 characters<br />
(postalignment) are potentially informative for parsimony<br />
analysis (26/260 in ITS 1, 26/227 in ITS 2). Absolute<br />
lengths for Fouquieriaceae sequences range <strong>from</strong><br />
254 to 260 base pairs (bp) in ITS1 and <strong>from</strong> 225 to 226<br />
in ITS2. Bootstrap and decay values show strong support<br />
along the branches separating each <strong>of</strong> the three groups<br />
corresponding to Henrickson’s three subgenera (subg. Idria,<br />
subg. Bronnia, and subg. Fouquieria) (Fig. 1). All<br />
but one set <strong>of</strong> intraspecific samples also form groups with<br />
strong support. Fouquieria burragei is the only exception,<br />
with one sample falling in the F. diguetii clade, with<br />
robust support, and the other sample falling with weak<br />
support in the sect. Ocotilla clade, with F. splendens and<br />
F. shrevei. Relationships within subg. Fouquieria are not<br />
well resolved. Fouquieria formosa falls sister to the remaining<br />
members <strong>of</strong> subg. Fouquieria, within which F.<br />
leonilae, F. ochoterenae, and a clade comprising F. diguetii,<br />
F. burragei, F. macdougalii, F.splendens, and F.<br />
shrevei form a trichotomy.<br />
Analysis 2, including the alignable portions <strong>of</strong> the outgroup<br />
sequences, yielded 240 minimum-length trees <strong>of</strong><br />
504 steps (CI 0.71; 0.62 without uninformative characters).<br />
Out <strong>of</strong> 529 characters (postalignment), 168 are<br />
potentially informative for parsimony analysis (84 <strong>from</strong><br />
ITS1; 84 <strong>from</strong> ITS2). The root for Fouquieriaceae falls<br />
between the two succulent subgenera, with subg. Idria<br />
sister to subg. Bronnia plus subg. Fouquieria. The strict<br />
consensus <strong>of</strong> all minimum-length trees (Fig. 2) has less<br />
resolution than does the strict consensus tree <strong>from</strong> analysis<br />
<strong>of</strong> ingroup taxa alone (Fig. 1). ‘‘Fast heuristic’’ bootstrap<br />
values indicated strong support for subg. Idria and<br />
subg. Bronnia (96 and 92%, respectively) and modest<br />
support for subg. Fouquieria (64%). Aside <strong>from</strong> intraspecific<br />
groupings, all but one <strong>of</strong> which are well supported,<br />
there is no resolution within subg. Fouquieria.<br />
Samples <strong>of</strong> F. burragei again are the exception to intra-<br />
specific groupings, with one sample placed in the F. diguetii<br />
clade and the other placed in an unresolved position.<br />
Only 54 unique ingroup topologies are present<br />
among the 240 trees. Of these 54 topologies, 12 are compatible<br />
with the strict consensus <strong>of</strong> the 12 parsimony trees<br />
<strong>from</strong> analysis 1 (Fig. 1) and six are compatible with the<br />
strict consensus <strong>of</strong> 54 maximum-likelihood trees <strong>from</strong><br />
analysis 3 (Fig. 3). The remaining 36 trees have a clade<br />
or grade at the base <strong>of</strong> subg. Fouquieria consisting either<br />
<strong>of</strong> F. formosa, F. leonilae, and F. ochoterenae (12 trees)<br />
or <strong>of</strong> F. burragei (2), F. splendens, and F. shrevei (24<br />
trees). The length <strong>of</strong> these ingroup topologies after pruning<br />
outgroup taxa is 85 steps for the 12 trees compatible<br />
with the results <strong>of</strong> analysis 1, and 86 steps for the remaining<br />
42 trees.<br />
Analysis 3 produced 54 maximum-likelihood (ML)<br />
trees without enforcement <strong>of</strong> a <strong>molecular</strong> clock (-ln likelihood<br />
1151.36, Fig. 3) and two trees with enforcement<br />
<strong>of</strong> a <strong>molecular</strong> clock (-ln likelihood 1156.04). The two<br />
trees obtained <strong>from</strong> ML analysis <strong>of</strong> the data set with enforced<br />
constancy <strong>of</strong> evolutionary rate are among the 54<br />
trees obtained <strong>from</strong> ML analysis without a clock constraint.<br />
If the rooting indicated in the two clock-enforced<br />
trees is accepted (Fig. 4), then a <strong>molecular</strong> clock cannot<br />
be rejected at the conventional alpha 0.05 level<br />
(2lnLR 9.36, df 18, P 0.95). The ingroup topology<br />
<strong>of</strong> the strict consensus <strong>of</strong> all 54 trees (Fig. 3)<br />
differs <strong>from</strong> that found in the parsimony-based analysis<br />
<strong>of</strong> ingroup taxa alone (Fig. 1) but is compatible with a<br />
subset <strong>of</strong> ingroup topologies found in the parsimonybased<br />
analysis with the inclusion <strong>of</strong> outgroups (Fig. 2).<br />
The positions <strong>of</strong> the group composed <strong>of</strong> F. formosa, F.<br />
leonilae, and F. ochoterenae and the group comprising<br />
F. splendens, F. shrevei and one specimen <strong>of</strong> F. burragei<br />
are reversed relative to the parsimony-based analysis <strong>of</strong><br />
ingroup taxa alone. Using the strict consensus <strong>of</strong> the 54<br />
ML trees as a backbone constraint for a branch-andbound<br />
parsimony analysis reveals that the ML trees are<br />
only one step longer than the maximum parsimony trees<br />
(86 vs. 85 steps, respectively). This length difference is<br />
insignificant as assessed by the ‘‘compare-2’’ T-PTP test<br />
with 100 replicates <strong>of</strong> branch-and-bound searches, implemented<br />
in PAUP* (P 0.4). (When described under a<br />
likelihood criterion, the maximum parsimony trees have<br />
-ln likelihood values ranging <strong>from</strong> 1155.07 to 1156.04.)<br />
In analysis 4, three <strong>of</strong> the five transversion:transition<br />
weighting schemes employed resulted in the same 12 tree<br />
topologies. The strict consensus <strong>of</strong> these 12 trees (length<br />
1040 for 25:10 weighting, length 1064 for 27:10 weighting,<br />
and length 110 for 3:1 weighting; all are length 86<br />
under unweighted parsimony) is equivalent to the strict<br />
consensus <strong>of</strong> the 54 ML trees <strong>from</strong> analysis 3 (Fig. 3),<br />
except that the weighted-parsimony tree contains less resolution<br />
in the Fouquieria formosa, F. leonilae, and F.<br />
ochoterenae clade, with the three taxa forming a trichotomy.<br />
The search with a 2:1 weighting scheme resulted in<br />
24 trees <strong>of</strong> length 98, the strict consensus <strong>of</strong> which lacked<br />
resolution within subg. Fouquieria beyond intraspecific<br />
groupings and the placement <strong>of</strong> F. burragei (1) with F.<br />
diguetii sequences. Twelve <strong>of</strong> these 24 trees (length 85<br />
under unweighted parsimony) are compatible with the<br />
strict consensus <strong>of</strong> trees <strong>from</strong> the unweighted parsimony<br />
analysis (Fig. 1), six (length 86 under unweighted par-
April 1999] SCHULTHEIS AND BALDWIN—MOLECULAR PHYLOGENETICS OF FOUQUIERIACEAE<br />
583<br />
Fig. 1. The strict consensus <strong>of</strong> 12 minimum-length unrooted trees <strong>of</strong> 85 steps (CI 0.82; 0.78 without uninformative characters), resulting<br />
<strong>from</strong> a branch-and-bound search <strong>of</strong> ingroup taxa only with the parsimony criterion (analysis 1 in Materials and Methods). Numbers above the<br />
branches indicate the percentage <strong>of</strong> trees in which the clade appears in 100 bootstrap replicates followed by decay indices (in parentheses). Subgenera<br />
indicated are those recognized by Henrickson (1972).<br />
simony) are compatible with the strict consensus <strong>of</strong> the<br />
ML trees (Fig. 3), and the remaining six (length 86 under<br />
unweighted parsimony) are compatible with the strict<br />
consensus <strong>of</strong> the ML trees except that the positions <strong>of</strong><br />
Fouquieria formosa and F. ochoterenae are reversed. The<br />
search with an 18:10 weighting scheme resulted in 12<br />
trees <strong>of</strong> length 954 (length 85 under unweighted parsimony),<br />
the strict consensus <strong>of</strong> which is equivalent to that<br />
<strong>of</strong> trees <strong>from</strong> the unweighted parsimony search (Fig. 1).<br />
Analysis 5 results (Table 3) indicate that clock-like<br />
evolution can be rejected for all root placements within<br />
Fouquieriaceae except (1) between the succulent and<br />
woody taxa (equivalent to the midpoint rooting and to<br />
the rooting seen in Fig. 4), (2) between subg. Idria and<br />
the remaining taxa (equivalent to the outgroup rooting<br />
seen in Fig. 2), or (3) between subg. Bronnia and the<br />
remaining taxa (Table 3).<br />
Analysis 6 suggests that neither the branch between<br />
subg. Idria and remaining taxa, nor between subg. Bronnia<br />
and remaining taxa is sufficiently long to act as a<br />
long-branch attractor. Of the 31 604 trees generated by<br />
analysis <strong>of</strong> 2000 simulated data sets (100 for each <strong>of</strong> 20<br />
possible root placements), only in 5% was the root placed<br />
between subg. Idria and the remaining taxa, and in only<br />
6% between subg. Bronnia and the remaining taxa (Table<br />
4). The root was placed between the succulent and woody<br />
taxa in 15% <strong>of</strong> the trees, indicating that the branch separating<br />
the two groups may indeed act as a long-branch<br />
attractor.<br />
Ingroup ITS topologies in Fouquieriaceae differ de-
584 AMERICAN JOURNAL OF BOTANY<br />
[Vol. 86<br />
Fig. 2. The strict consensus <strong>of</strong> 240 minimum-length trees <strong>of</strong> 504 steps (CI 0.71; 0.62 without uninformative characters), resulting <strong>from</strong> a<br />
heuristic search with 100 replicates <strong>of</strong> random sequence addition and with the parsimony criterion (analysis 2 in Materials and Methods). Sequences<br />
<strong>of</strong> ingroup taxa (Fouquieria ssp.) are included in their entirety, while only the alignable portions <strong>of</strong> outgroup taxa are included, with the unalignable<br />
sites coded as missing data. Numbers above branches indicate the percentage <strong>of</strong> trees in which the clade appears in a ‘‘fast heuristic’’ bootstrap,<br />
as implemented in PAUP*. Subgenera indicated are those recognized by Henrickson (1972).<br />
pending on the type <strong>of</strong> analysis conducted (parsimony,<br />
weighted parsimony, maximum likelihood), and on<br />
whether outgroup taxa are included. The trees produced<br />
by weighted parsimony with 25:10, 27:10, and 3:1 transversion:<br />
transition weighting and maximum-likelihood<br />
(ML) analyses are compatible (Fig. 3); future reference<br />
to ML trees in the Discussion also includes these weighted<br />
parsimony trees, while reference to parsimony trees<br />
refers to the unweighted parsimony trees. The trees produced<br />
by parsimony analysis including outgroup taxa<br />
(Fig. 2) contain trees compatible with both parsimony<br />
(Fig. 1) and ML (Fig. 3) analyses <strong>of</strong> ingroup taxa alone.<br />
When described under a parsimony criterion, the ML<br />
trees are only one step longer than the most parsimonious<br />
trees (86 vs. 85 steps), an insignificant difference (P <br />
0.4, ‘‘compare-2’’ T-TPT test). Differences between the<br />
parsimony and ML trees involve clades with weak support,<br />
as estimated by bootstrap values and decay indices<br />
(on the parsimony trees).<br />
DISCUSSION<br />
Monophyly <strong>of</strong> subgenera—The three subgenera recognized<br />
by Henrickson (1972) are supported as mono-<br />
phyletic groups based on the ITS data, irrespective <strong>of</strong> tree<br />
reconstruction criterion (parsimony or ML) or rooting<br />
method (outgroup, midpoint, or clock-consistent approaches).<br />
Within subg. Fouquieria, neither sect. Fouquieria<br />
nor sect. Ocotilla are well-supported monophyletic<br />
groups in the <strong>molecular</strong> trees. Those taxa corresponding<br />
to sect. Fouquieria are nested within those that<br />
correspond to sect. Ocotilla in the ML trees (Figs. 3–4).<br />
The reverse is the case in the maximum parsimony topologies<br />
(Fig. 1).<br />
Uncertainty in placement <strong>of</strong> the root in Fouquieriaceae<br />
dictates that if the genus Idria, containing I. (Fouquieria)<br />
columnaris, is recognized (e.g., Wiggins, 1980) then the<br />
genus Bronnia should also be reinstated for F. purpusii<br />
and F. fasciculata to ensure monophyly <strong>of</strong> Fouquieria<br />
sensu stricto (s.s.). The issue <strong>of</strong> whether to recognize one<br />
or three genera in Fouquieriaceae is, however, purely a<br />
subjective question <strong>of</strong> rank (subgenus vs. genus) (see<br />
Cantino, Olmstead, and Wagstaff, 1997).<br />
In his 1972 treatment <strong>of</strong> Fouquieriaceae, Henrickson<br />
presented a ‘‘putatively phylogenetic system <strong>of</strong> relationship’’<br />
in which he identified four main groups <strong>of</strong> taxa,<br />
corresponding to the subgenera and sections in his tax-
April 1999] SCHULTHEIS AND BALDWIN—MOLECULAR PHYLOGENETICS OF FOUQUIERIACEAE<br />
585<br />
Figs. 3–4. Trees resulting <strong>from</strong> maximum-likelihood analyses <strong>of</strong> ingroup taxa. 3. (Top) The strict consensus <strong>of</strong> 54 unrooted trees <strong>of</strong> -ln<br />
likelihood 1151.36, resulting <strong>from</strong> a heuristic search <strong>of</strong> ingroup taxa with the maximum-likelihood criterion and without enforcement <strong>of</strong> a<br />
<strong>molecular</strong> clock (analysis 3 <strong>of</strong> Materials and Methods). Subgenera are those recognized by Henrickson (1972). This tree is equivalent to the strict<br />
consensus <strong>of</strong> trees produced under parsimony analysis <strong>of</strong> ingroup taxa with certain differential weightings <strong>of</strong> transitions vs. transversions, except<br />
that Fouquieria formosa, F. leonilae, and F. ochoterenae form a trichotomy in the parsimony consensus tree (analysis four <strong>of</strong> Materials and<br />
Methods). 4. (Bottom) A phylogram depiction <strong>of</strong> the first <strong>of</strong> two trees <strong>of</strong> -ln likelihood 1156.04, resulting <strong>from</strong> a heuristic search <strong>of</strong> ingroup<br />
taxa with the maximum-likelihood criterion and enforcement <strong>of</strong> a <strong>molecular</strong> clock. The two trees produced in this search are among the 54 produced<br />
in the search without enforcement <strong>of</strong> a <strong>molecular</strong> clock. The root depicted is equivalent to a midpoint rooting.<br />
onomic scheme (Table 1). The four groups Henrickson<br />
identified were based on cytological data, anatomical<br />
data, and a phenetic analysis <strong>of</strong> 71 ecological, vegetative,<br />
and reproductive characters. He identified Fouquieria<br />
leonilae and F. ochoterenae as most representative <strong>of</strong> ancestral<br />
conditions within the family, citing in particular<br />
the following characters as primitive: ‘‘diploid, . . . dorsiventral<br />
leaves, initial single traces to each sepal, ten stamens,<br />
. . . 6(-12) ovules per ovary.’’ Other species within<br />
the family possess dorsiventral or isolateral leaves, a<br />
higher but variable number <strong>of</strong> sepal traces, ten or more<br />
stamens, and a higher but variable number <strong>of</strong> ovules.<br />
Henrickson’s proposition that Fouquieria leonilae and F.<br />
ochoterenae possess the greatest number <strong>of</strong> plesiomorphic<br />
characteristics within the family (1972) is most consistent<br />
with the parsimony trees (Fig. 1), assuming that<br />
out <strong>of</strong> the three most reasonable root placements, the root<br />
is placed along the branch between subg. Fouquieria and<br />
the remaining taxa. From the lineage represented by F.<br />
leonilae and F. ochoterenae, Henrickson suggested the<br />
derivation <strong>of</strong> three additional lineages: (1) F. formosa,<br />
(2) the putatively polyploid series <strong>of</strong> F. macdougalii, F.<br />
diguetii, and F. burragei, and (3) the ocotillos, F. splendens<br />
and F. shrevei. Fouquieria leonilae and F. ochoterenae,<br />
together with F. formosa and the polyploid series,<br />
constitute the dendroid subg. Fouquieria sect. Fouquier-
586 AMERICAN JOURNAL OF BOTANY<br />
[Vol. 86<br />
TABLE 4. Percentage <strong>of</strong> trees <strong>from</strong> parsimony analysis <strong>of</strong> 2000 simulated<br />
data sets (100 for each <strong>of</strong> 20 examined root positions; see<br />
Materials and Methods) in which the root is placed between the<br />
taxa listed and the remaining taxa.<br />
Taxa %<br />
FB a<br />
FB(2)<br />
FC<br />
FC,FFA,FP,FO,FL,FF<br />
FD<br />
FDFB<br />
FFA<br />
FFA,FP<br />
FF<br />
FL<br />
FL,FO<br />
FO<br />
FS,FSH<br />
FSH<br />
FC,FFA,FP<br />
FP<br />
FS<br />
FM<br />
FF,FC,FP,FFA<br />
FSH,FS,FB(2)<br />
a Refer to Table 3 for abbreviations.<br />
0.0<br />
7.0<br />
5.0<br />
1.0<br />
0.0<br />
8.0<br />
1.0<br />
6.0<br />
5.0<br />
6.0<br />
2.0<br />
7.0<br />
0.7<br />
6.0<br />
15.0<br />
2.0<br />
11.0<br />
6.0<br />
1.0<br />
1.0<br />
ia. Fouquieria splendens and F. shrevei form subg. Fouquieria<br />
sect. Ocotilla. The remaining two groups Henrickson<br />
recognized are the succulent subg. Idria (F. columnaris)<br />
and subg. Bronnia (F. fasciculata and F. purpusii).<br />
The four groups Henrickson identified (1972) are not,<br />
he noted, <strong>of</strong> equal phenetic distinctiveness relative to<br />
each other. The two succulent subgenera are more distinct<br />
phenetically <strong>from</strong> each other and <strong>from</strong> the woody taxa<br />
than are the two woody sections <strong>from</strong> each other. This<br />
impression is supported by the clock-like <strong>molecular</strong> trees<br />
in which the three subgenera are on long branches relative<br />
to branches within the subgenera (Fig. 4). Henrickson<br />
pointed out that the succulent subg. Bronnia is phenetically<br />
intermediate between subg. Idria and woody<br />
subg. Fouquieria. His dendrograms indicate a closer phenetic<br />
similarity <strong>of</strong> subg. Bronnia with subg. Fouquieria<br />
using ecological and vegetative characters, and a closer<br />
similarity with subg. Idria using reproductive characters.<br />
Within subg. Fouquieria, Henrickson noted that F. burragei<br />
is intermediate between sect. Ocotilla and sect.<br />
Fouquieria. This is perhaps a reflection <strong>of</strong> the hypothesized<br />
parental contribution to F. burragei <strong>from</strong> both sections<br />
within subg. Fouquieria, a hypothesis supported by<br />
our <strong>molecular</strong> data.<br />
Outgroups and rooting—Rooting Fouquieriaceae using<br />
the outgroup method was problematic due to the levels<br />
<strong>of</strong> sequence divergence between Fouquieriaceae and<br />
its potential outgroups. Outgroup sampling in this study<br />
was not aimed at determining the placement <strong>of</strong> Fouquieriaceae<br />
relative to other angiosperm families, but was<br />
guided by other higher level phylogenetic analyses which<br />
included Fouquieriaceae (Anderberg, 1992; Hufford,<br />
1992; Olmstead et al., 1993). Results <strong>of</strong> analyses including<br />
only those portions <strong>of</strong> the outgroup sequences that<br />
were alignable suggest that an ericalean grouping is sister<br />
to Fouquieriaceae. This clade, however, has weak support.<br />
This result is in agreement with interpretations <strong>of</strong> a<br />
position <strong>of</strong> Fouquieriaceae close to Ericales (e.g., Anderberg,<br />
1992; Olmstead et al., 1993). The tree shown (Fig.<br />
2) was rooted using Tamarix based on recent <strong>evidence</strong><br />
that Tamaricaceae falls close to Caryophyllales, well outside<br />
<strong>of</strong> Asteridae sensu lato (Lledó et al., 1998).<br />
Using only the alignable portions <strong>of</strong> the outgroup sequences<br />
placed the root in Fouquieriaceae between Fouquieria<br />
subg. Idria (F. columnaris) and the remaining<br />
species (Fig. 2). A midpoint rooting results in a basal<br />
dichotomy between subg. Fouquieria and both subg. Idria<br />
and subg. Bronnia (equivalent to the root seen in Fig.<br />
4). Both <strong>of</strong> these root placements as well as a root placement<br />
along the subg. Bronnia branch are consistent with<br />
clock-like evolution <strong>of</strong> ITS sequences (Table 3). Results<br />
<strong>from</strong> data sets simulated with various root positions suggest<br />
that neither the subg. Idria branch nor the subg.<br />
Bronnia branch act as long-branch attractors, whereas the<br />
branch between the succulent and woody taxa may attract<br />
other long branches (Table 4). The root placement in trees<br />
resulting <strong>from</strong> the analyses including outgroup taxa (Fig.<br />
2, analysis 2) does not, therefore, appear to be a result<br />
<strong>of</strong> long-branch attraction. Because the root placement<br />
within Fouquieriaceae is uncertain, the implications <strong>of</strong> all<br />
reasonable root positions will be addressed in discussion<br />
<strong>of</strong> character evolution.<br />
Succulent and woody habits—Whether the ancestral<br />
condition in Fouquieriaceae is succulent or woody is<br />
equivocal, in part due to uncertainty about rooting <strong>of</strong> the<br />
ingroup tree topology and identity <strong>of</strong> the closest relatives<br />
<strong>of</strong> the family. If subg. Idria and subg. Bronnia fall on<br />
different sides <strong>of</strong> the basal dichotomy in Fouquieriaceae<br />
(e.g., a root as in Fig. 2) then succulence may well be<br />
the ancestral condition in the family, assuming monophyly<br />
<strong>of</strong> subg. Fouquieria. This would indicate a reversal<br />
to the woody condition in subg. Fouquieria. A transition<br />
<strong>from</strong> a succulent habit to a woody habit would be unusual,<br />
if not unique. In Cactaceae, the woody Pereskioideae<br />
appear to be basal (Hershkovitz and Zimmer, 1997,<br />
citing others), with succulence perhaps achieved via a<br />
neotenic reduction <strong>of</strong> wood developmental rates (Altesor,<br />
Silva, and Ezcurra, 1994). Sufficient phylogenetic information<br />
is lacking to assess the direction <strong>of</strong> life-form evolution<br />
in other groups containing both succulent and<br />
woody members (e.g., Euphorbiaceae, Asclepiadaceae,<br />
and Asteraceae). If the root within Fouquieriaceae falls<br />
as suggested above, a woody or an intermediate ancestral<br />
condition rather than a succulent condition would dictate<br />
independent elaboration <strong>of</strong> succulence in the three subgenera<br />
<strong>of</strong> Fouquieria: cortical succulence in subg. Fouquieria<br />
and stem succulence in both subg. Bronnia and<br />
subg. Idria. The woody species <strong>of</strong> subgenus Fouquieria<br />
possess a cortical network <strong>of</strong> water storage tissue (Scott,<br />
1932) that is slightly more developed than that <strong>of</strong> the<br />
stem succulents (Humphrey, 1935; Henrickson, 1969c).<br />
Succulence in both subg. Idria and subg. Bronnia is due<br />
to parenchymatous water storage cells within the xylem<br />
(Humphrey, 1935; Henrickson, 1969c), originating <strong>from</strong><br />
both cambial division and division <strong>of</strong> pre-existing parenchymatous<br />
cells (Henrickson, 1969c). Succulent tissue
April 1999] SCHULTHEIS AND BALDWIN—MOLECULAR PHYLOGENETICS OF FOUQUIERIACEAE<br />
587<br />
is most extensive in F. columnaris (subg. Idria), which<br />
possesses a succulent pith, less extensive in F. purpusii,<br />
and least extensive in F. fasciculata (Henrickson, 1969b,<br />
c, 1972). If subg. Bronnia is sister to subg. Idria (i.e., a<br />
root as in Fig. 4), this would suggest a single origin <strong>of</strong><br />
stem succulence in Fouquieriaceae and an equivocal ancestral<br />
habit in the family.<br />
Determining the direction <strong>of</strong> evolution between ocotillo<br />
and dendroid growth forms relies on resolution <strong>of</strong><br />
relationships within subg. Fouquieria. Parsimony analysis<br />
indicates paraphyly <strong>of</strong> sect. Fouquieria and derivation<br />
<strong>of</strong> the ocotillo growth form <strong>from</strong> dendroid forms (Fig.<br />
1). However, ML analyses indicate paraphyly <strong>of</strong> sect.<br />
Ocotilla and derivation <strong>of</strong> dendroid growth forms <strong>from</strong><br />
ocotillo forms (Fig. 3). More data on relationships among<br />
the members <strong>of</strong> subg. Fouquieria are needed to resolve<br />
life-form evolution in the group.<br />
Polyploidy—Polyploids within Fouquieria include the<br />
tetraploid F. diguetii (n 24) and the hexaploids F. burragei<br />
(n 36) and F. columnaris (n 36). Henrickson<br />
(1972) suggested that F. diguetii may have descended<br />
directly <strong>from</strong> ‘‘F. macdougalii stock’’ and that F. burragei<br />
is perhaps an amphiploid derivative <strong>of</strong> F. diguetii<br />
and another, possibly extinct taxon. Fouquieria macdougalii<br />
and F. diguetii share similar floral and inflorescence<br />
structure (Henrickson, 1969b, 1972). Fouquieria macdougalii<br />
(n 12) is distributed in portions <strong>of</strong> the Sonoran<br />
desert and adjacent tropical deciduous and thorn forests<br />
in mainland Mexico (Henrickson 1969b, 1972). Fouquieria<br />
diguetii is found in Baja California, but overlaps<br />
with the range <strong>of</strong> F. macdougalii around Guaymas, in<br />
mainland Mexico (Henrickson, 1969b, 1972). The <strong>molecular</strong><br />
data shown here indicate a close relationship between<br />
F. diguetii and F. macdougallii that is consistent<br />
with (but not directly supportive <strong>of</strong>) Henrickson’s interpretation<br />
<strong>of</strong> a progenitor–derivative relationship between<br />
F. macdougalii and F. diguetii.<br />
Fouquieria burragei is vegetatively very similar to F.<br />
macdougalii and, especially, F. diguetii, but is very different<br />
in floral characteristics <strong>from</strong> the two species, with<br />
more open flowers, more numerous stamens, pink to<br />
white corolla pigmentation (vs. red), and more elongate<br />
inflorescences (Henrickson, 1969b, 1972). Samples <strong>of</strong> F.<br />
burragei fall in two places in the <strong>molecular</strong> trees: one<br />
with F. diguetii in the parsimony and ML trees and the<br />
other with the ocotillo clade (F. splendens and F. shrevei)<br />
in the parsimony trees or within the ocotillo grade in the<br />
ML trees. Different phylogenetic placements <strong>of</strong> F. burragei<br />
ITS sequences might be the result <strong>of</strong> bidirectional<br />
concerted evolution (Wendel, Schnabel, and Seelanen,<br />
1995), i.e., different populations <strong>of</strong> F. burragei may have<br />
become fixed for different ITS repeat types inherited<br />
<strong>from</strong> the two parental taxa involved in its putative allopolyploid<br />
origin. This interpretation is consistent with<br />
Henrickson’s (1972) hypothesis that F. diguetii served as<br />
one parent to F. burragei, while a species similar to a<br />
white-flowered F. splendens served as the other. Fouquieria<br />
burragei is distributed along the eastern coast <strong>of</strong><br />
lower Baja California (Henrickson, 1969b, 1972), overlapping<br />
in range with F. diguetii. The current range <strong>of</strong><br />
F. splendens overlaps with that <strong>of</strong> F. diguetii on the Baja<br />
peninsula, but does not reach quite as far south as the<br />
northern known range limit <strong>of</strong> F. burragei (Henrickson,<br />
1969b, 1972). Different subspecies <strong>of</strong> F. splendens differ<br />
in part in floral pigmentation, ranging <strong>from</strong> red to creamwhite,<br />
but none <strong>of</strong> the white forms <strong>of</strong> F. splendens are<br />
known <strong>from</strong> Baja.<br />
Henrickson’s (1972) suggestion that a representative<br />
<strong>from</strong> the Fouquieria purpusii and F. fasciculata lineage<br />
may have served as a parent in the amphiploid derivation<br />
<strong>of</strong> F. columnaris is neither supported nor refuted here,<br />
but can be reconciled with our results most easily if the<br />
ingroup root falls between the woody and succulent species.<br />
Fouquieria purpusii and F. fasciculata (subg. Bronnia)<br />
share with F. columnaris (subg. Idria) both vegetative<br />
and floral features, including succulent trunks and<br />
white to cream-white decandrous flowers (Henrickson,<br />
1972). Fouquieria columnaris is found in central Baja<br />
California and as a small population at Punta Cirio in<br />
Sonora, Mexico (Henrickson, 1969b, 1972; Humphrey<br />
and Marx, 1980); Fouquieria fasciculata and F. purpusii<br />
are highly restricted, found in pockets <strong>of</strong> arid tropical<br />
scrub vegetation (Henrickson, 1972) in Hidalgo (F. fasciculata)<br />
and in Puebla and Oaxaca (F. purpusii)(Henrickson,<br />
1969b, 1972). Extensive <strong>molecular</strong> divergence<br />
between these two species and F. columnaris<br />
corresponds with their highly disjunct distributions in<br />
suggesting that a direct ancestor–descendent relationship<br />
between subg. Bronnia and subg. Idria is untenable. A<br />
possibility that cannot be discounted is that F. columnaris<br />
is an allopolyploid involving a member <strong>of</strong> subg. Bronnia<br />
and an extinct taxon, with the ITS repeat type sequenced<br />
<strong>from</strong> F. columnaris derived <strong>from</strong> the extinct taxon.<br />
Floral features—While Henrickson did not mention<br />
floral pigmentation in his assessment <strong>of</strong> ancestral features<br />
within the family, a root placement along the branch between<br />
subg. Fouquieria and the remaining taxa (the root<br />
most consistent with Henrickson’s hypothesis that F.<br />
leonilae and F. ochoterenae are best representative <strong>of</strong><br />
ancestral conditions within the family) suggests a shift<br />
<strong>from</strong> white/cream floral pigmentation to red pigmentation<br />
in ML trees (Figs. 3–4), but is equivocal in parsimony<br />
trees (Fig. 1). If subg. Idria and subg. Bronnia fall on<br />
opposite sides <strong>of</strong> the basal dichotomy, white- to creamcolored<br />
flowers appear ancestral in the family, with a single<br />
derivation <strong>of</strong> red pigmentation in subg. Fouquieria<br />
and, in the parsimony trees (Fig. 1), a reversal(s) in the<br />
clade containing F. burragei, F. shrevei, and F. splendens.<br />
While shifts in floral pigmentation and shape may be<br />
indicative <strong>of</strong> pollinator shifts, there does not appear to be<br />
strict conformity to stereotypical pollination syndromes<br />
within Fouquieriaceae. A temporal match between hummingbird<br />
migrations and flowering time in red-flowered<br />
F. splendens has been shown to favorably affect seed set<br />
(Waser, 1979). However, carpenter bees are also important<br />
pollinators and can be the most frequent visitors<br />
(Scott, Buchmann, and O’Rourke, 1993). Similarly, while<br />
15 species <strong>of</strong> bees have been observed to visit the white/<br />
cream flowered F. columnaris, observed visitors to the<br />
white/cream flowered F. fasciculata include hummingbirds,<br />
bees, and various other insects (Henrickson, 1972).<br />
In general, observations <strong>of</strong> pollination in Fouquieriaceae<br />
are limited.
588 AMERICAN JOURNAL OF BOTANY<br />
[Vol. 86<br />
Biogeography—Most species <strong>of</strong> Fouquieriaceae are<br />
endemic to mainland Mexico. Those found outside <strong>of</strong><br />
mainland Mexico (i.e., in Baja California and the southwestern<br />
United States) are all polyploid, with the exception<br />
<strong>of</strong> the widespread Fouquieria splendens. While F.<br />
diguetii (tetraploid) and F. burragei (hexaploid) are<br />
closely related, the two hexaploids F. columnaris and F.<br />
burragei, although geographically proximal to one another,<br />
must represent independent origins <strong>of</strong> the hexaploid<br />
condition. The overall distribution <strong>of</strong> Fouquieria species,<br />
with all but one diploid species endemic to mainland<br />
Mexico, suggests a mainland Mexican origin for the family.<br />
The highly localized distribution <strong>of</strong> many species (e.g.,<br />
F. shrevei, F. leonilae, F. ochoterenae, F. fasciculata, F.<br />
purpusii) may be indicative <strong>of</strong> relictualism or neoendemism.<br />
Considering the magnitude <strong>of</strong> the disjunction between<br />
the hexaploid F. columnaris and the morphologically<br />
similar species <strong>of</strong> subg. Bronnia it seems likely that<br />
considerable extinction has occurred in the history <strong>of</strong><br />
Fouquieriaceae. This impression is reinforced by the isolation<br />
<strong>of</strong> each <strong>of</strong> the three subgenera on long branches <strong>of</strong><br />
the clock-like ITS trees (e.g., Fig. 4).<br />
Concluding remarks—Any discussion <strong>of</strong> character<br />
evolution depends on an accurate tree topology and an<br />
accurate placement <strong>of</strong> a root within that topology. Rooting<br />
using an outgroup comparison method is problematic<br />
when sister taxa are too divergent or unknown. Outgroup<br />
taxa that are too divergent <strong>from</strong> the ingroup effectively<br />
act as random sequences and tend to root an ingroup<br />
topology along the longest branch (Wheeler, 1990). Due<br />
to its uncertain phylogenetic placement and its apparently<br />
high divergence <strong>from</strong> other angiosperm taxa, rooting<br />
Fouquieriaceae with the outgroup method is problematic,<br />
but this problem is certainly not unique.<br />
Our strategy for identifying a root in Fouquieriaceae<br />
was to employ the outgroup method using only the conservative<br />
(alignable) regions within the outgroup sequences<br />
while retaining all ingroup sequence data (Bruns<br />
et al., 1992), a strategy shown to be preferable to retention<br />
<strong>of</strong> all data when portions <strong>of</strong> the data are highly variable<br />
(Smith, 1994). We rejected the possibility that the<br />
root placement resulting <strong>from</strong> the outgroup method was<br />
confounded by long-branch attraction (Huelsenbeck,<br />
1997). Additionally, we found only three root placements<br />
consistent with clock-like evolution <strong>of</strong> ITS sequences.<br />
Our discussion <strong>of</strong> character evolution within Fouquieriaceae<br />
is based on a set <strong>of</strong> tree topologies and root placements<br />
that appear to be the best hypotheses given the<br />
limitations <strong>of</strong> our data. Future examination <strong>of</strong> additional<br />
morphological and <strong>molecular</strong> <strong>evidence</strong> may allow further<br />
refinement <strong>of</strong> our understanding <strong>of</strong> diversification in this<br />
fascinating family.<br />
LITERATURE CITED<br />
ALBERT, V.A.,AND B. D. MISHLER. 1992. On the rationale and utility<br />
<strong>of</strong> weighting nucleotide sequence data. Cladistics 8: 73–83.<br />
ALTESOR A., C. SILVA, AND E. EZCURRA. 1994. Allometric neoteny and<br />
the evolution <strong>of</strong> succulence in cacti. Botanical Journal <strong>of</strong> the Linnean<br />
Society 114: 282–292.<br />
ANDERBERG, A. A. 1992. The circumscription <strong>of</strong> the Ericales, and their<br />
cladistic relationships to other families <strong>of</strong> ‘‘higher’’ dicotyledons.<br />
Systematic Botany 17: 660–675.<br />
BALDWIN, B. G. 1992. Phylogenetic utility <strong>of</strong> the internal transcribed<br />
spacers <strong>of</strong> <strong>nuclear</strong> ribosomal DNA in plants: an example <strong>from</strong> the<br />
Compositae. Molecular Phylogenetics and Evolution 1: 3–16.<br />
BESSEY, C. E. 1915. The phylogenetic taxonomy <strong>of</strong> flowering plants.<br />
Annals <strong>of</strong> the Missouri Botanical Garden 2: 109–164.<br />
BRUNS, T. D., R. VILGALYS, S.M.BARNS, D.GONZALEZ, D.S.HIBBETT,<br />
D. J. LANE, L.SIMON, S.STICKEL, T.M.SZARO, W.G.WEISBURG,<br />
AND M. S. SOGIN. 1992. Evolutionary relationships within the fungi:<br />
analyses <strong>of</strong> <strong>nuclear</strong> small subunit rRNA sequences. Molecular<br />
Phylogenetics and Evolution 1: 231–241.<br />
CANTINO, P. D., R. G. OLMSTEAD, AND S. J. WAGSTAFF. 1997. A comparison<br />
<strong>of</strong> phylogenetic nomenclature with the current system: a<br />
botanical case study. Systematic Biology 46: 313–331.<br />
CRONQUIST, A. 1981. An integrated system <strong>of</strong> classification <strong>of</strong> flowering<br />
plants. Columbia University Press, New York, NY.<br />
DAHLGREN, R., S. JENSEN, AND B. NIELSEN. 1976. Iridoid compounds<br />
in Fouquieriaceae and notes on its possible affinities. Botaniska<br />
notiser 129: 207–212.<br />
DECANDOLLE, A. P. 1828. Prodromus Systematis Naturalis Regni Vegetabilis,<br />
vol. 3. Treuttel and Wurtz, Paris.<br />
DOYLE, J. J., AND J. L. DOYLE. 1987. A rapid DNA isolation procedure<br />
for small quantities <strong>of</strong> fresh leaf tissue. Phytochemical Bulletin 19:<br />
11–15.<br />
FELSENSTEIN, J. 1988. Phylogenies <strong>from</strong> <strong>molecular</strong> sequences: inferences<br />
and reliability. Annual Review <strong>of</strong> Genetics 22: 521–565.<br />
HENRICKSON, J. 1967. Pollen morphology <strong>of</strong> the Fouquieriaceae. Aliso<br />
6: 137–160.<br />
———. 1969a. Anatomy <strong>of</strong> periderm and cortex <strong>of</strong> Fouquieriaceae.<br />
Aliso 7: 97–126.<br />
———. 1969b. An introduction to the Fouquieriaceae. Cactus and<br />
Succulent Journal 61: 97–106.<br />
———. 1969c. The succulent Fouquierias. Cactus and Succulent Journal<br />
61: 178–184.<br />
———. 1972. A taxonomic revision <strong>of</strong> Fouquieriaceae. Aliso 7: 439–<br />
537.<br />
HERSHKOVITZ, M.A., AND E. A. ZIMMER. 1997. On the evolutionary<br />
origins <strong>of</strong> the cacti. Taxon 46: 217–232.<br />
HIGGINS, D. G. 1994. CLUSTAL V: multiple alignment <strong>of</strong> DNA and<br />
protein sequences. In A. M. Griffin and H. G. Griffin [eds.], Computer<br />
analysis <strong>of</strong> sequence data, part II. Methods in <strong>molecular</strong> biology,<br />
vol. 25, 307–318. Humana Press, Totowa, NJ.<br />
HUELSENBECK, J. P. 1997. Is the Felsenstein zone a fly trap? Systematic<br />
Biology 46: 69–74.<br />
———, AND B. RANNALA. 1997. Phylogenetic methods come <strong>of</strong> age:<br />
testing hypotheses in an evolutionary context. Science 276: 227–<br />
232.<br />
HUFFORD, L. 1992. Rosidae and their relationships to other nonmagnoliid<br />
dicotyledons: a phylogenetic analysis using morphological<br />
and chemical data. Annals <strong>of</strong> the Missouri Botanical Garden.79:<br />
218–248.<br />
HUMBOLDT, A., A. BONPLAND, AND C. S. KUNTH. 1823. Nova Genera<br />
et Species Plantarum, vol. 6. Gide, Paris.<br />
HUMPHREY, R. R. 1935. A study <strong>of</strong> Idria columnaris and Fouquieria<br />
splendens. American Journal <strong>of</strong> Botany 22: 184–206.<br />
———, AND D. B. MARX. 1980. Distribution <strong>of</strong> the boojum tree (Idria<br />
columnaris) on the coast <strong>of</strong> Sonora, Mexico as influenced by climate.<br />
Desert Plants 2: 183–187.<br />
HUTCHINSON, J. 1926. The families <strong>of</strong> flowering plants I. dicotyledons.<br />
MacMillan and Co., London.<br />
JENSEN, S.R.,AND B. J. NIELSEN. 1982. Iridoid glucosides in Fouquieriaceae.<br />
Phytochemistry 21: 1623–1629.<br />
LLEDÓ, M. D., M. B. CRESPO, K.M.CAMERON, M.F.FAY, AND M. W.<br />
CHASE. 1998. Systematics <strong>of</strong> Plumbaginaceae based upon cladistic<br />
analysis <strong>of</strong> rbcL sequence data. Systematic Botany 23: 21–29.<br />
MABRY, T. J. 1977. The order Centrospermae. Annals <strong>of</strong> the Missouri<br />
Botanical Garden 64: 210–220.<br />
MADDISON, W.P.,AND D. R. MADDISON. 1992. MacClade: analysis <strong>of</strong><br />
phylogeny and character evolution, ver. 3.0. Sinauer, Sunderland,<br />
MA.<br />
MARKOS, S. E. 1995. Phylogeny <strong>of</strong> Arctostaphylos hookeri (Ericaceae)
April 1999] SCHULTHEIS AND BALDWIN—MOLECULAR PHYLOGENETICS OF FOUQUIERIACEAE<br />
589<br />
based on ribosomal DNA sequence <strong>from</strong> the ITS region. M. A.<br />
thesis, San Francisco State University. San Francisco, CA.<br />
MORTON, C. M., M. W. CHASE, K.A.KRON, AND S. M. SWENSEN. 1996.<br />
A <strong>molecular</strong> evaluation <strong>of</strong> the monophyly <strong>of</strong> the order Ebenales<br />
based upon rbcL sequence data. Systematic Botany 21: 567–586.<br />
NASH, G. V. 1903. A revision <strong>of</strong> the family Fouquieriaceae. Bulletin<br />
<strong>of</strong> the Torrey Botanic Club 30: 449–459.<br />
OLMSTEAD, R. G., B. BREMER, K.SCOTT, AND J. D. PALMER. 1993. A<br />
parsimony analysis <strong>of</strong> the Asteridae sensu lato based on rbcL sequences.<br />
Annals <strong>of</strong> the Missouri Botanical Garden 80: 700–722.<br />
PORTER, J. M. 1996. Phylogeny <strong>of</strong> Polemoniaceae based on <strong>nuclear</strong><br />
ribosomal internal transcribed spacer DNA sequences. Aliso 15:<br />
57–77.<br />
ROEMER, J.J.,AND J. A. SCHULTES. 1819. Systema Vegetabilium, vol.<br />
4. Stuttgardt.<br />
SCOGIN, R. 1977. Anthocyanins <strong>of</strong> Fouquieriaceae. Biochemical Systematics<br />
and Ecology 5: 265–267.<br />
SCOTT, F. M. 1932. Some features <strong>of</strong> the anatomy <strong>of</strong> Fouquieria splendens.<br />
American Journal <strong>of</strong> Botany 19: 673–678.<br />
SCOTT, P. E., S. L. BUCHMANN, AND M. K. O’ROURKE. 1993. Evidence<br />
for mutualism between a flower-piercing carpenter bee and ocotillo:<br />
use <strong>of</strong> pollen and nectar by nesting bees. Ecological Entomology<br />
18: 234–240.<br />
SMITH, A. B. 1994. Rooting <strong>molecular</strong> trees: problems and strategies.<br />
Biological Journal <strong>of</strong> the Linnean Society 51: 279–292.<br />
SWOFFORD, D. L. 1993. PAUP: phylogenetic analysis using parsimony,<br />
version 3.1. Computer program distributed by the Illinois Natural<br />
History Survey, Champaign, IL.<br />
TAKHTAJAN, A. 1969. Flowering plants: origin and dispersal. Smithsonian<br />
Institution Press, Washington DC.<br />
THORNE, R. F. 1969. Synopsis <strong>of</strong> a putatively phylogenetic classification<br />
<strong>of</strong> the flowering plants. Aliso 6: 57–66.<br />
WASER, N. M. 1979. Pollinator availability as a determinant <strong>of</strong> flowering<br />
time in ocotillo (Fouquieria splendens). Oecologia 39: 107–<br />
121,<br />
WENDEL, J. F., A. SCHNABEL, AND T. SEELANAN. 1995. Bidirectional<br />
interlocus concerted evolution following allopolyploid speciation<br />
in cotton (Gossypium). Proceedings <strong>of</strong> the National Academy <strong>of</strong><br />
Sciences, USA 92: 280–284.<br />
WHEELER, W. C. 1990. Nucleic acid sequence phylogeny and random<br />
outgroups. Cladistics 6: 363–367.<br />
WHITE, T. J., T. BRUNS, S.LEE, AND J. TAYLOR. 1990. Amplification<br />
and direct sequencing <strong>of</strong> fungal ribosomal RNA genes for <strong>phylogenetics</strong>.<br />
In M. Innis, D. Gelfand, J. Sninsky, and T. White [eds.],<br />
PCR protocols: a guide to methods and applications, 315–322. Academic<br />
Press, San Diego, CA.<br />
WIGGINS, I. L. 1980. Flora <strong>of</strong> Baja California. Stanford University<br />
Press, Stanford, CA.