POLEMONIACEAE PHYLOGENY AND CLASSIFICATION:

Transcription

1 American Journal of Botany 87(9): POLEMONIACEAE PHYLOGENY AND CLASSIFICATION: IMPLICATIONS OF SEQUENCE DATA FROM THE CHLOROPLAST GENE NDHF 1 L. ALAN PRATHER, 2,5 CAROLYN J. FERGUSON, 3 AND ROBERT K. JANSEN 4 2 Herbarium and Department of Botany & Plant Pathology, Michigan State University, East Lansing, Michigan USA; 3 Herbarium and Division of Biology, Kansas State University, Manhattan, Kansas USA; and 4 Section of Integrative Biology, Plant Resources Center, and Institute of Cellular and Molecular Biology, University of Texas, Austin, Texas USA The chloroplast gene ndhf was used to study phylogenetic relationships of the Polemoniaceae at two levels: among members of the Ericales and among genera of the family. Sequence data for interfamilial analyses consisted of 2266 bp for 14 members of the Ericales, including four species of the Polemoniaceae, plus three outgroup taxa. The Polemoniaceae were found to be related to Diospyros, Fouquieria, the Primulales, Rhododendron, and Impatiens, but relationships among taxa were generally not well supported. The precise position of the Polemoniaceae within the Ericales remains obscure. Data for intrafamilial analyses consisted of 1031 bp for 27 species of the Polemoniaceae, including at least one species from most genera of the family, plus five outgroup taxa. A single most parsimonious tree was identified. The analyses suggested that subfamily Cobaeoideae, excluding Loeselia, is monophyletic and that Huthia is sister to Cantua. Acanthogilia was sister to the remainder of subfamily Cobaeoideae. Subfamily Polemonioideae plus Loeselia formed four subclades that were strongly supported as monophyletic and represent the major lineages of the subfamily. Key words: Acanthogilia; classification; Ericales; Huthia; Loeselia; molecular phylogeny; ndhf; Polemoniaceae. The Polemoniaceae is a relatively small but highly diverse family with 350 species. The species are distributed primarily in North America, and many are endemic to the western United States (Grant, 1998a). The family and its constituent genera and species have served as model systems for many systematic and evolutionary studies (e.g., Epling and Dobzhansky, 1942; Wright, 1943; Grant, 1959; Grant and Grant, 1965; Harborne and Smith, 1978; Carlquist, Eckhart, and Michener, 1984; Paige and Whitham, 1985; Schlichting and Levin, 1986; Campbell, 1989; Waser and Price, 1989; Barrett, Harder, and Worley, 1996). The family is taxonomically complex, and generic delimitation has been controversial and unstable (e.g., Greene, 1887; Grant, 1959, 1998a; Johnson and Soltis, 1995). The first molecular phylogenetic study of the Polemoniaceae was that of Steele and Vilgalys (1994) based on partial sequences of the chloroplast encoded gene matk. Their investigation was followed by a second study based on a more variable region of the same gene (Johnson and Soltis, 1995; Johnson et al., 1996) and by another based on sequences of the internal transcribed spacer (ITS) regions of nuclear ribosomal DNA (Porter, 1996). In addition, Grant (1998a) proposed a phylogeny based on both morphology and published sequence data; however, it was assembled using evolutionary systematics rather than cladistic methodology. Several phylo- 1 Manuscript received 10 August 1999; revision accepted 10 December The authors thank Frank Axelrod, Mark Chase, Amy David, Wendy Hodgson, Leigh Johnson, Ki-Joong Kim, Kathy Kron, Robert Patterson, Cynthia Morton, Mark Porter, and Stanley Spencer for providing plant material, DNA samples, and DNA sequences, and Jan Barber, Les Goertzen, Mark Mayfield, Amanda Posto, Anna Wiese, Dieter Wilken, Rachel Williams, and an anonymous reviewer for helpful comments on an earlier version of the manuscript. Support was provided by the National Science Foundation (LAP, DEB ; CJF, DEB ; RKJ, DEB ) and by a Garden Club of America/World Wildlife Fund Scholarship in Tropical Botany to LAP. 5 Author for correspondence ( alan@msu.edu) genetic studies have provided insight into evolution at lower taxonomic levels (e.g., in Ipomopsis [Wolf, Soltis, and Soltis, 1993]; Navarretia [Spencer and Porter, 1997]; Cobaea [Prather and Jansen, 1998]; and Phlox [Ferguson, Krämer, and Jansen, 1999]). Two recent studies (Porter and Johnson, 1998; Johnson, Soltis, and Soltis, 1999) have focused on the position of the Polemoniaceae in the Ericales (sensu Angiosperm Phylogeny Group, 1998). Concurrent with the interest in Polemoniaceae phylogenetics has been a resurgence of interest in classification of the family. Grant s early studies, especially Natural history of the phlox family (Grant, 1959), revolutionized Polemoniaceae classification and have served as the foundation for Polemoniaceae systematics for the last 40 yr. Grant (1998a) recently contributed a modified classification of the entire family that incorporated morphological and molecular data that had accumulated since At many levels the new classification and molecular phylogenies correspond with his 1959 classification, supporting in large part the basic tenets of the earlier work. In addition to Grant s contribution, several other workers have recently contributed to tribal (Porter, 1998a), generic (Grant, 1998b; Grant and Day, 1998; Porter, 1998a, b), or subgeneric (Day, 1993; Prather, 1994, 1999; Spencer and Porter, 1997; Ferguson, Krämer, and Jansen, 1999) classification. Incorporating phylogenetic information into classification has proven to be sometimes challenging and even controversial, but this provides yet another arena in which the Polemoniaceae might serve as a model system. Some examples of controversy are the generic disposition of the many disparate elements now included in Gilia s.l. (e.g., Grant, 1998a, b; Grant and Day, 1998; Porter, 1998a, b), tribal classification of subfamily Polemonioideae (Grant, 1998a; Grant and Day, 1998; Porter, 1998a), and the status of the genera Microsteris (e.g., Patterson and Wilken, 1993; Grant, 1998a; Ferguson, Krämer, and Jansen, 1999) and Loeseliastrum (e.g., Porter, 1996; Grant, 1998a).

2 September 2000] PRATHER ET AL. PHYLOGENY OF THE POLEMONIACEAE 1301 TABLE 1. Voucher information for species from which ndhf sequences were generated. An asterisk (*) by a species indicates that only the 3 end was sequenced. Species ERICALES (except Polemoniaceae) Ardisia crenata Sims Fouquieria columnaris (Kellogg) Kellogg ex Curran Jacquinia umbellata A. DC. POLEMONIACEAE Acanthogilia gloriosa (Brandegee) A. G. Day & Moran* Aliciella latifolia (S. Watson) J. M. Porter* Allophyllum divaricatum (Nutt.) A. D. Grant & V. E. Grant* Bonplandia geminiflora Cav.* Cantua buxifolia Juss. ex Lam* Cobaea scandens Cav. Collomia linearis Nutt.* Eriastrum sapphirinum (Eastw.) H. Mason* Gilia leptalea (A. Gray) Greene* Gilia scabra Brandegee* Gilia sp. nov.* Giliastrum rigidulum (Benth.) Rydb. Gymnosteris parvula (Rydb.) A. Heller* Huthia coerulea Brand* Ipomopsis aggregata (Pursh) V. E. Grant* Ipomopsis tenuifolia (A. Gray) V. E. Grant* Langloisia matthewsii (A. Gray) Greene* Langloisia setosissima (Torr. & A. Gray) Greene* Leptodactylon californicum Hook. & Arn.* Linanthus ciliatus (Benth.) Greene* Loeselia glandulosa (Cav.) G. Don* Microsteris gracilis (Hook.) Greene* Navarretia intertexta (Benth.) Hook.* Phlox pilosa L. Polemonium foliosissimum A. Gray* Polemonium pauciflorum S. Watson* Voucher Kron 3001 (NCU) Prather 1927 (MSC) Axelrod 4552 (UPRRP) Porter & Heil 7987 (SJNM) Porter & Machen (RSA) Prather 1350 (TEX) Cultivated, San Francisco State University Cultivated, San Francisco State University Patterson, s. n. (RSA) Prather 1605 (MSC) Prather 1441 (TEX) Johnson (WS) Porter & Machen (RSA) Porter & Heil 7991 (SJNM) Porter 8723 (RSA) Patterson s. n. (WS) Hodgson 7924 (F) Prather 1618 (MSC) Prather 1440 (TEX) Prather 1411 (TEX) Liston (TEX) Prather 1348 (TEX) Prather 1397 (TEX) Porter & Campbell 9231 (SJNM) David 274 (TEX) Spencer (RSA) Ferguson 455 (MO) Porter 7576 (SJNM) Hinton (TEX) GenBank accession number a GBAN-AF GBAN-AF GBAN-AF GBAN-AF GBAN-AF GBAN-AF GBAN-AF GBAN-AF GBAN-AF GBAN-AF GBAN-AF GBAN-AF GBAN-AF GBAN-AF GBAN-AF GBAN-AF GBAN-AF GBAN-AF GBAN-AF GBAN-AF GBAN-AF GBAN-AF GBAN-AF GBAN-AF GBAN-AF GBAN-AF GBAN-AF GBAN-AF GBAN-AF a The prefix GBAN- has been added to link the online version of American Journal of Botany to GenBank, but is not part of the actual accession number. Our understanding of Polemoniaceae phylogenetics has advanced considerably and the classification has been improved, yet many uncertainties remain. Here we discuss implications of new data from the cpdna gene ndhf on phylogeny and classification. We focus on several issues related to the following taxonomic groups: (1) the order Ericales, (2) the subfamily Cobaeoideae, including Huthia, (3) the genus Acanthogilia, (4) the genus Loeselia, and (5) the subfamily Polemonioideae. We review the current state of Polemoniaceae phylogenetics and focus attention on the remaining questions. Furthermore, we discuss the ongoing attempts to reconcile molecular phylogenies of the family with morphological variation and to incorporate these data into the classification of the family. Finally we illustrate why it is important to take a conservative and holistic approach to nomenclature in the Polemoniaceae. MATERIALS AND METHODS Sampling Two sets of analyses were performed. The first included representatives of the Polemoniaceae and other ericalean families (interfamilial relationships) and the second included species representing genera of the Polemoniaceae (intrafamilial relationships). For the analysis of interfamilial relationships, we used 11 sequences from other sources (Olmstead, Sweere, and Wolfe, 1993; Olmstead et al., 2000): Anagallis arvensis L. (GenBank accession GBAN-AF130212), Camellia japonica L. (GBAN-AF130216), Cornus florida L. (GBAN-AF130220), Diospyros texana Scheele (GBAN- AF130213), Garrya elliptica Dougl. ex Lindl. (GBAN-AF147714), Halesia tetraptera L. (GBAN-AF130222), Impatiens biflora Walt. (GBAN- AF130210), Nicotiana tabacum L. (GBAN-L14953), Phlox drummondii Hook. (GBAN-AF130211), Rhododendron mucronulatum Turcz. (GBAN- AF130209), and Styrax americana Lam. (GBAN-AF130215). In addition, we sequenced the ndhf coding region for six species (Table 1). We attempted to sample Diapensiaceae but were unable to amplify ndhf from any DNA samples that we obtained. The data matrix included four species from the Polemoniaceae, ten from potentially related families, and three outgroup taxa (Cornus, Garrya, and Nicotiana). For the analysis of intrafamilial relationships, we sequenced over 1 kilobase (kb) from the 3 region of ndhf from 26 species, including at least one species from each genus of the Polemoniaceae (Table 1), except the recently established genera Maculigilia and Tintinabulum (Grant, 1998b). There has been some confusion concerning the identity of G. scabra in the literature. Gilia scabra of Johnson et al. (1996) and Porter (1996), as well as G. cf. scabra of Porter and Johnson (1998) is identical to the taxon we have called Gilia sp. nov. The earlier studies included DNA from an undescribed species that was confused with, and referred to as, G. scabra. The new species is currently being described by J. M. Porter (personal communication) who kindly furnished samples of plant tissue from both species, which are included here (Table 1). We used the following combinations of outgroup taxa because the sister group relationships of the Polemoniaceae remain obscure: (1) Diospyros, (2) Fouquieria, (3) the Primulales (Anagallis, Ardisia, and Jacquinia), (4) all five aforementioned taxa simultaneously, and (5) Fouquieria and Diospyros. Preliminary analyses resulted in a single tree that was topologically identical regardless of outgroup combination; therefore combination 5 was used in all subsequent analyses.

3 1302 AMERICAN JOURNAL OF BOTANY [Vol. 87 DNA extraction and amplification Total DNA was extracted from fresh or dried leaf material, the latter sometimes from herbarium specimens. The DNA extraction methods of Doyle and Doyle (1987) were used for fresh material and those of Loockerman and Jansen (1996) for dried material. A double-stranded DNA fragment was amplified using the ndhf primers of Jansen (1992). For those taxa for which the entire coding region was sequenced, the gene was amplified in two segments. Amplification components and parameters followed the protocol of Kim and Jansen (1995) except that hot-start or touchdown polymerase chain reaction (PCR) methods were sometimes employed. Product purification and sequencing Products were sequenced manually or on an automated sequencer. Samples that were manually sequenced were purified with glass beads as described by Kim and Jansen (1994). Samples sequenced with the automated sequencer were purified by spin columns either directly (QIAquick PCR Purification Kit, Qiagen, New Castle, Delaware, USA) or following separation in an agarose gel (QIAquick Gel Extraction Kit, Qiagen). Manual sequencing was performed using the snap-chill technique described by Kim and Jansen (1994), except that termination reactions were carried out at 42 C. Automated sequencing was performed on an ABI 377 DNA sequencer (Applied Biosystems, Inc., Foster City, California, USA). Sequencing was accomplished with the same primers used by Kim and Jansen (1995). Phylogenetic analyses Sequences were manually aligned. For the analyses of interfamilial relationships, the first 26 bp of the coding region were excluded because these data were missing for most taxa. For the analyses of intrafamilial relationships, we used only the 3 end of the sequence beginning with bp 1262 relative to tobacco. Insertion/deletion (indel) events were treated in four ways: (1) as missing data, (2) as missing data and each gap scored as an additional binary character equal in weight to a base substitution, (3), as additional binary characters with gap regions deleted from the matrix, and (4) as a new state (i.e., a fifth base). The alignment is available on request from the first author. Parsimony methods were implemented using PAUP* (version 4.0b2; Swofford, 1999). Heuristic searches were performed using TREE BISECTION RE- CONNECTION, COLLAPSE, and MULTREES options. The STEEPEST DE- SCENT option was not in effect. One hundred replicate searches with random taxon-entry were used to search for multiple islands of most parsimonious trees (Maddison, 1991; Page, 1993). The amount of support for monophyletic groups was assessed using bootstrap replicates (Felsenstein, 1985) with 100 addition-sequence replicates per bootstrap replicate for the interfamilial analysis and ten addition-sequence replicates for the intrafamilial analysis. Bootstrap analyses were performed using gap treatment 1 only. RESULTS Interfamilial relationships Of the 2266 bp of aligned sequence data, 861 sites (38.0%) were variable and 501 (22.1%) were potentially phylogenetically informative. Four of 13 indels (30.8%) were potentially phylogenetically informative. Missing sequence data constituted 1.05% of the data matrix, and there were no missing data for indels in the interfamilial analyses. For gap treatments 1, 2, and 4, a topologically identical set of three most parsimonious trees was identified (Fig. 1; treatment 1: 1809 steps, consistency index excluding uninformative characters [CI e ] 0.565, retention index [RI] 0.571; treatment 2: 1822 steps, CI e 0.566, RI 0.572; treatment 4: 1902 steps, CI e 0.572, RI 0.573). There were two unresolved nodes in the strict consensus of these three trees (Fig. 2). Six most parsimonious trees (1780 steps, CI e 0.564, RI 0.568) were identified by the search using gap treatment 3. Three of those six trees corresponded to the set of trees from the searches using other gap treatments. The strict consensus of six trees is topologically consistent, but less Fig. 1. One of three most parsimonious trees identified from analysis of interfamilial relationships in the Ericales (sensu Angiosperm Phylogeny Group, 1998) based on ndhf sequence data with gaps scored as missing data (gap treatment 1; 1809 steps, CI e 0.565, RI 0.571). Values along branches indicate number of steps. Polemoniaceae species are shown in boldface. resolved than the strict consensus of the former sets of three (Fig. 2). We favor the strict consensus tree for gap treatments 1, 2, and 4 and use it as the basis of discussion, because many phylogenetically informative sites are deleted from the matrix using gap treatment 3. The monophyly of the four Polemoniaceae taxa was strongly supported (Fig. 2; 100% bootstrap value), as was the monophyly of the Primulales (100% bootstrap value). Diospyros was sister to the Polemoniaceae, albeit with poor bootstrap support (37%). Fouquieria was placed in a trichotomy with the Polemoniaceae Diospyros clade and a clade of the Primulales plus Impatiens and Rhododendron, but this clade had poor bootstrap support (29%). Halesia was strongly supported as sister to Styrax (100% bootstrap value). The relationship among the Halesia Styrax clade, Camellia, and the clade of the remaining ingroup taxa was unresolved (Fig. 2). Intrafamilial relationships Of the 1031 bp of aligned sequence data used in the intrafamilial analyses, 404 sites (39.2%) were variable and 235 (22.8%) were potentially phylogenetically informative. Seven of 13 indels (53.8%) were potentially phylogenetically informative. Missing sequence data constituted 1.05% of the data matrix and three indel cells (0.72%) were scored as missing. The topology within the Po-

4 September 2000] PRATHER ET AL. PHYLOGENY OF THE POLEMONIACEAE 1303 of each of these four subclades was strongly supported ( 92% bootstrap values), but there was poor support (54 65% bootstrap values) for nodes resolving the relationships among these four groups. Fig. 2. Strict consensus of three most parsimonious trees from analyses of interfamilial relationships in the Ericales based on ndhf sequence data using gap treatments 1, 2, and 4. See text for details. Bootstrap values ( replicates; gap treatment 1) are shown along branches. Asterisks (*) indicate additional nodes that collapse in the strict consensus of six most parsimonious trees identified when scoring gaps as additional binary characters with gap regions deleted from the matrix (gap treatment 3). Polemoniaceae species are shown in boldface. lemoniaceae was identical regardless of which outgroup or outgroup combination was used (trees not shown) and discussion below is limited to analyses with Diospyros and Fouquieria as outgroups. Regardless of gap treatment, a single, topologically identical, most parsimonious tree (Fig. 3) was identified (treatment 1: 797 steps, CI e 0.601, RI 0.691; treatment 2: 809 steps, CI e 0.603, RI 0.694; treatment 3: 771 steps, CI e 0.601, RI 0.693; treatment 4: 865 steps, CI e 0.612, RI 0.704). There was a basal split between two major Polemoniaceae lineages (Fig. 3). The first was composed of Acanthogilia, Bonplandia, Cantua, Cobaea, and Huthia, but was weakly supported (Fig. 3; 52% bootstrap support). This lineage corresponds to Grant s (1998a) subfamily Cobaeoideae excluding Loeselia, and we will refer to this as the Cobaeoideae clade. The clade consisting of Bonplandia, Cantua, Cobaea, and Huthia will be referred to as the core Cobaeoideae. Within this clade, Huthia was strongly supported (100% bootstrap support) as sister to Cantua (Fig. 3). The second lineage was composed of the remainder of the genera and was strongly supported (98% bootstrap support). It corresponds for the most part to Grant s (1998a) subfamily Polemonioideae, and we will refer to it as the Polemonioideae clade. This lineage included four subclades (following the nomenclature of Porter, 1996): (1) Polemonium, (2) the Gilieae subclade (Allophyllum, Collomia, Gilia leptalea, and Navarretia), (3) the Linanthieae subclade (Gymnosteris, Leptodactylon, Linanthus, Microsteris, and Phlox), and (4) the Loeselieae subclade (Aliciella, Giliastrum, Eriastrum, Gilia scabra, Gilia sp. nov., Ipomopsis, Langloisia [incl. Loeseliastrum], and Loeselia). The Loeselieae subclade was sister to the remaining three, and Polemonium was sister to a clade composed of the Gilieae and Linanthieae subclades. Monophyly DISCUSSION Interfamilial relationships The ndhf phylogeny (Fig. 1) places the Polemoniaceae within the Ericales (sensu Angiosperm Phylogeny Group, 1998) in agreement with other molecular analyses (reviewed in Porter and Johnson, 1998). In contrast, the traditional placement of the Polemoniaceae has been near the Hydrophyllaceae or Convolvulaceae, for example in the Solanales of Dahlgren (1980) and Cronquist (1981). The sister-group relationship of Diospyros and the Polemoniaceae (Fig. 2) is poorly supported and has not been uncovered by other molecular analyses. However, a large number of potential sister-group relationships have been proposed by various cladistic analyses, and there is no consensus among studies (see Fig. 1 in Porter and Johnson, 1998). Low bootstrap values and a lack of resolution at the node uniting the Polemoniaceae Diospyros lineage with two others, Fouquieria and a clade consisting of the Primulales, Impatiens, and Rhododendron (Fig. 2), prevent us from making any strong conclusions; precise relationships of the Polemoniaceae to these families remain obscure. This study is the fourth molecular study to focus on resolving the phylogenetic position of the Polemoniaceae (Porter and Johnson, 1998; Johnson et al., 1996; Johnson, Soltis, and Soltis, 1999). These four investigations complement broader comparisons (e.g., Olmstead et al., 1992; Chase et al., 1993; Morton et al., 1996). Overall, results of molecular studies are inconsistent from analysis to analysis with regard to precise placement of the family, and many nodes in critical areas are poorly supported. Resolution of relationships among these groups will require a concerted effort and involve sampling of many taxa and genes. Intrafamilial relationships The ndhf phylogeny is largely congruent with other molecular phylogenies and is in general agreement with Grant s (1998a) classification. There are, however, some important differences among our results, other molecular phylogenies, and Grant s classification. These differences result from disparate phylogenetic hypotheses, conflicting perspectives on how to incorporate phylogenetic information into classification, or a combination of these sources. Here we place our data within the context of ongoing issues in Polemoniaceae phylogenetics and classification. We summarize several key differences among studies, explicitly identify whether the issues are phylogenetic or classification related, and identify the problems remaining and propose how best to approach them. Subfamily Cobaeoideae The Cobaeoideae clade is one of two major groups of the Polemoniaceae in the ndhf analyses (Fig. 3), and the core Cobaeoideae is monophyletic. Other studies have sampled only three genera of the core Cobaeoideae, Bonplandia, Cantua, and Cobaea, and these three taxa are monophyletic in most molecular phylogenies (Fig. 4). There are two exceptions: in the matk study of Steele and Vilgalys (1994), members of the core Cobaeoideae plus Acanthogilia formed an unresolved polytomy at the base of the Polemoniaceae (Fig. 4B), and in the ITS tree (Porter, 1996;

5 1304 AMERICAN JOURNAL OF BOTANY [Vol. 87 Fig. 3. Single most parsimonious tree identified from analysis of intrafamilial relationships of the Polemoniaceae based on ndhf sequence data with gaps scored as missing data (gap treatment 1; 797 steps, CI e 0.601, RI 0.691). Major clades are indicated on the right. Bootstrap values ( replicates; gap treatment 1) are given above branches. The number of steps is indicated below each branch in italics. Subclade nomenclature follows Porter (1996). Fig. 4. Simplified phylogenetic diagrams illustrating relationships of Acanthogilia to the core Cobaeoideae and Polemonioideae clade as inferred from six separate molecular phylogenetic analyses. (A) ndhf (from Fig. 3, this study). (B) matk 1 (from Fig. 3 in Steele and Vilgalys, 1994). (C) matk 2 (from Fig. 3 in Johnson et al., 1996). (D) nad1b (from Fig. 2 in Porter and Johnson, 1998). (E) ITS (from Fig. 1 in Porter, 1996). (F) 18S (from Fig. 3 in Johnson, Soltis, and Soltis, 1999). Sampling of core Cobaeoideae and Polemonioideae clade varied in each study. Fig. 4E) these same taxa formed a basal group that was paraphyletic to the remainder of the family. However, branches in the critical portion of the ITS tree were weakly supported and Porter stated that the branching pattern was suspect (Porter, 1996, p. 69). Our ndhf phylogeny is the only molecular study to include Huthia. Given our sampling, H. coerulea is sister to Cantua buxifolia and the relationship is strongly supported. A close relationship between Huthia and Cantua has long been hypothesized based on the woody habit, simple leaves, actinomorphic and tubular corollas, and their primarily Andean distributions. Our results are in agreement with morphology and current classification (Grant, 1998a). Subfamily Cobaeoideae sensu Grant (1998a) consists of Acanthogilia, Bonplandia, Cantua, Cobaea, Loeselia, and Huthia. Because the bulk of the molecular evidence suggests that Bonplandia, Cantua, and Cobaea form a monophyletic group and because the sister-group relationship between Cantua and Huthia in the ndhf phylogeny is consistent with morphological evidence and classification, monophyly of the core Cobaeoideae is well established. Cobaea is sister to Bonplandia in the ndhf tree (Fig. 3) and is nested within the Polemoniaceae in every molecular analysis. This is an important finding because Cobaea has often been placed in other families or segregated to its own (reviewed in Prather, 1999a). Our data,

6 September 2000] PRATHER ET AL. PHYLOGENY OF THE POLEMONIACEAE 1305 and in fact all molecular analyses, suggest that Loeselia should be excluded from the subfamily (see below). The relationship between the core Cobaeoideae and Acanthogilia is unclear and is discussed in detail below. Phylogenetic position and classification of Acanthogilia The position of Acanthogilia as sister to the core Cobaeoideae in the ndhf phylogeny, albeit with weak support (Fig. 3), is novel among molecular studies (Fig. 4). In other molecular studies Acanthogilia always appeared as a basal lineage, although its exact placement varied among phylogenies (Fig. 4). In the two matk analyses, Acanthogilia was in an unresolved polytomy at the base of the Polemoniaceae (Fig. 4B, C; Steele and Vilgalys, 1994; Johnson et al., 1996), a position consistent with, but less resolved than, the ndhf phylogeny. The nad1b data placed the core Cobaeoideae as sister to a clade consisting of Acanthogilia and the Polemonioideae clade (Fig. 4D; Porter and Johnson, 1998). Based on the ITS data, Acanthogilia was the sister taxon to the entire family except Bonplandia (Fig. 4E; Porter, 1996). The 18S data placed the genus as sister to all seven remaining Polemoniaceae taxa sampled, including Bonplandia, Cantua, and Cobaea (Fig. 4F; Johnson, Soltis, and Soltis, 1999). The placement of Acanthogilia in the ndhf phylogeny is in general agreement with morphological features and classification. When erecting the monotypic genus Acanthogilia, Day and Moran (1986) concluded that it was most closely related to Cantua, based on morphological and palynological features. Grant (1998a) placed Acanthogilia in subfamily Cobaeoideae, based on morphological evidence, but found the genus distinct enough to place it in its own tribe, tribe Acanthogilieae. The questions concerning Acanthogilia involve both phylogeny and classification: What are its relationships? Is it best placed in subfamily Cobaeoideae or Polemonioideae, or perhaps in a third subfamily? Because of the agreement among morphology, classification, and the ndhf phylogeny, as well as consistency with the phylogenetic position in other cpdna studies (Fig. 4), we conclude that Acanthogilia is best included in subfamily Cobaeoideae. However, among molecular phylogenies, lack of resolution and/or weak support for the relationships to other genera suggest that additional comparisons are needed to firmly establish its phylogenetic and taxonomic position. Classification of Loeselia The ndhf tree places Loeselia sister to a clade of two Gilia species, G. scabra and G. sp. nov., and this clade falls within the Loeselieae subclade. The ITS (Porter, 1996) and matk (Johnson et al., 1996) phylogenies identified these same relationships, albeit with different sampling. In fact, except for nad1b, all molecular studies that have sampled both taxa have placed Loeselia and G. scabra in a monophyletic Loeselieae subclade. Grant hypothesized a close relationship between Loeselia and some members of the Loeselieae subclade, particularly the Gilia rigidula group (more or less equivalent to Giliastrum; Porter, 1998a). In fact, he stated it is hypothesized that the Gilia rigidula group evolved from Loeselia in the Madro-Tertiary flora... (Grant, 1998a, p. 747). It is noteworthy that Loeselia and Giliastrum, plus a few other taxa, share a relatively recent common ancestor (Fig. 3) and that this pattern agrees with Grant s evolutionary hypothesis. Why then, did Grant place Loeselia in subfamily Cobaeoideae and not in his tribe Gilieae of subfamily Polemonioideae (Grant, 1998a)? This decision stems from the fact that Grant did not use a cladistic definition of monophyly (Grant, 1998a, p. 748) and chose, rather, to emphasize similarities of Loeselia to members of subfamily Cobaeoideae. We choose to use a cladistic definition of monophyly and therefore include Loeselia in subfamily Polemonioideae. The evolutionary relationships are not in conflict among previous and current studies; we merely differ in how we choose to represent those relationships in classification. We agree that there are many similarities between Loeselia and some genera of subfamily Cobaeoideae, especially Bonplandia. For instance, seeds of the species of subfamily Cobaeoideae are broadly winged, except for those of Bonplandia, which are narrowly winged and very similar to wings of Loeselia seeds. Wings are absent from seeds of species of subfamily Polemonioideae except for some species of Polemonium that have ridge-like vestigial wings (Grant, 1959). The small size of chromosomes of Loeselia is also similar to that of members of subfamily Cobaeoideae, but this information has been quantified for few species of Loeselia, and some species of subfamily Polemonioideae (e.g., Leptodactylon californicum) have chromosomes approaching those of Loeselia in size (fig. 62 in Grant, 1959). But we also note many similarities to some members of subfamily Polemonioideae. For example, the chromosome number of Loeselia species is n 9, a number common in subfamily Polemonioideae, but unknown in subfamily Cobaeoideae, except in Acanthogilia. The pollen of Loeselia is not similar to that of any species in subfamily Cobaeoideae, but is very similar to some species in subfamily Polemonioideae (Stuchlik, 1967a, b; Taylor and Levin, 1975). The veins of the corolla lobes of Loeselia species are either free, or connected well above the base, both conditions that occur only among species of subfamily Polemonioideae. All species of subfamily Cobaeoideae have veins that are connected at the base, as well as sometimes in the upper lobes (Day and Moran, 1986). Because morphological and cytological evidence is equivocal, yet molecular data strongly place Loeselia in the Polemonioideae clade, we choose to place Loeselia in subfamily Polemonioideae. Phylogeny and classification of subfamily Polemonioideae Grant s subfamily Polemonioideae plus Loeselia, our Polemonioideae clade, is strongly supported as monophyletic. These genera, which include most species and genera of the family, also formed a monophyletic group in every other phylogeny except for that based on the 18S data, in which Phlox was sister to the rest of the family except Acanthogilia (Fig. 4F). That placement of Phlox is incongruent with all other molecular data as well as morphological evidence. The focus of the 18S study was not on intrafamilial relationships and sampling within the family was quite limited (eight species), therefore we urge caution in interpreting the 18S data with regard to relationships within the Polemoniaceae. The preponderance of evidence strongly supports a monophyletic group of the genera included in Grant s subfamily Polemonioideae plus Loeselia (Fig. 4). The four major subclades of the Polemonioideae clade in the ndhf tree are strongly supported (Fig. 3) and provide a context for grouping genera and species of the subfamily. The four subclades, Polemonium, Gilieae, Linanthieae, and Loeselieae, correspond to groups detected by most other molecular phylogenetic studies. Except for sampling differences, the sub-

7 1306 AMERICAN JOURNAL OF BOTANY [Vol. 87 clades are identical to the clades of Porter (1996). The three latter groups also correspond to the Allophyllum Gilia splendens clade, Phlox Gilia filiformis clade, and Ipomopsis Gilia subnuda clade, respectively, of Johnson et al. (1996). The agreement in subclade membership among nearly all molecular analyses and strong support for monophyly in our phylogeny allow us to be reasonably certain that these four groups of the Polemonioideae clade are monophyletic. The only phylogenetic analysis that suggested any of these groups is nonmonophyletic was the nad1b study, in which the Linanthieae subclade was polyphyletic (Fig. 3 in Porter and Johnson, 1998). The authors considered placement of the Linanthieae members an anomalous result, possibly because of missing data for those taxa (Porter and Johnson, 1998). Like the 18S study, focus of the nad1b study was on interfamilial relationships, therefore sampling within the Polemoniaceae was limited. The ndhf phylogeny places the Linanthieae and Gilieae subclades as sister groups, with Polemonium and the Loeselieae subclade as successively more basal lineages. Different placements have been suggested by other phylogenetic analyses and relationships between subclades typically have been poorly supported. It appears that the major lineages of the Polemonioideae clade are well defined, but relationships among the subclades remain unresolved. As an example we consider the phylogenetic position of Polemonium. The ITS data placed Polemonium in a monophyletic group with the Linanthieae and Gilieae subclades, albeit with different sister-group relationships than in the ndhf phylogeny (Porter, 1996). In the 18S phylogeny, Polemonium was sister to a lineage consisting of the Loeselieae and Gilieae subclades. The results from the matk studies were inconsistent. The Steele and Vilgalys (1994) study supported Polemonium as sister to a lineage consisting of the Linanthieae and Gilieae subclades, in agreement with the ndhf data. The Johnson and Soltis (1995) phylogeny placed Polemonium sister to the Linanthieae subclade only, while the phylogeny of Johnson et al. (1996) placed Polemonium as sister to the remainder of the Polemonioideae clade, as did the nad1b data (Porter and Johnson, 1998). Grant suggested that Polemonium, especially section Polemonium, may have been one of the earliest derived members of the temperate lineage (Grant, 1998a, p. 748) and molecular data are in general agreement with his hypothesis. The tribal classification of subfamily Polemonioideae is certain to be one of the major issues of Polemoniaceae classification in the near future. In Grant s (1998a) revision, tribal circumscriptions within the subfamily were largely the same as in his 1959 treatment except that Navarretia was moved from the Gilieae to the Polemonieae, and Leptodactylon and Linanthus were excluded from the Gilieae and included in the newly erected tribe Leptodactyloneae. Concurrently, Porter (1998a) recognized two tribes not treated by Grant, tribe Phlogieae (our Linanthieae subclade) and tribe Loeselieae (our Loeselieae subclade). Porter s tribe Phlogieae is an expanded Leptodactyloneae, and his tribe Loeselieae is essentially Grant s (1998a) tribe Gilieae plus Loeselia, but excluding many species included in Gilia, most notably the type of Gilia, G. laciniata. The bulk of molecular evidence from several genes (see above) supports Porter s new tribes. However, if those tribes are recognized, all that would remain of tribe Gilieae would be Gilia s.s., whereas the Polemonieae would include Allophyllum, Collomia, Navarretia, and Polemonium. Based on molecular data tribe Polemonieae would not be monophyletic. If the Gilieae were expanded to include Allophyllum, Collomia, and Navarretia (the Gilieae subclade) and if the Polemonieae were restricted to Polemonium alone, all the tribes would be monophyletic based on the molecular phylogenies. The question is whether molecular analyses should be the basis for classification. Molecules and morphology The degree to which it is appropriate to use morphological characters vs. molecular data in classification of the Polemoniaceae has recently become an issue. Grant (1998a, p. 750; 1998b, pp ) and Grant and Day (1998, pp ) have emphasized morphological data and criticized what they perceived as an overemphasis on molecular data, especially by Johnson et al. (1996) and Porter (1996). Grant went so far as to say (1998b, p. 84) In any incongruence between the evidence from one or two genes and that from multifactorial phenotypic characters, the latter must be given great weight. On the other hand, Porter (1998a, b) emphasized molecular data when making nomenclatural changes, although not to the exclusion of discussions of morphological features. Every recent student of the Polemoniaceae has agreed, at least implicitly, that data from both morphology and molecules can be valuable indicators of relationship and are therefore likely to be useful in classification. The recent discussions of the utility of different types of data, however, have been sometimes misleading for two reasons. First, the molecular data have been analyzed using cladistic methodology, while the morphological characters were analyzed using evolutionary systematics (Grant, 1998a). Differing methodology could lead to different outcomes regardless of whether there is conflict among data sets. Second, the conflict discussed thus far in the literature is primarily between molecular data and morphological characters that have been traditionally considered important, i.e. those used in classification. It has not been shown that there is conflict between morphological characters, in general, and molecular data. Detailed study of morphological features combined with phylogenetic analyses is the only appropriate method to address potential conflicts between morphological and molecular data. This is a very difficult task at the intergeneric level because of the remarkable morphological diversity within and among genera. But it is an important future goal for Polemoniaceae systematists and is absolutely critical to understanding evolution in the family. Phylogenetic analyses of morphological data are not unheard of at lower levels in the Polemoniaceae. Three such studies exist: the Ipomopsis spicata complex (Wilken and Hartman, 1991), Navarretia (Spencer and Porter, 1997), and Cobaea (Prather, 1999b). The latter two are the only examples for which molecular phylogenies have also been estimated (Spencer and Porter, 1997; Prather and Jansen, 1998). Interestingly, comparisons of molecular and morphological phylogenies revealed much congruence. However, these studies found some morphological characters traditionally used in sectional circumscription to be homoplasious when examined in a phylogenetic context and advised against continued use of those morphological characters. A comparison of molecular and morphological data using cladistic methodology would not allay the concerns of those people, including Grant, who object to cladistic methodology in the first place. But it would at least allow the question to be refined (i.e., is it the methodology that leads to different

8 September 2000] PRATHER ET AL. PHYLOGENY OF THE POLEMONIACEAE 1307 conclusions regarding phylogeny and classification, or is it conflict between types of data?). Some recent changes in classification, such as Grant s transfer of Navarretia from tribe Gilieae to tribe Polemonieae, resulted from consideration of both morphological and molecular data. We support this practice. The ultimate goal should be to find consensus among all data and to explain any conflict, not merely to find morphological characters that support molecular phylogenies. The studies on Navarretia (Spencer and Porter, 1997) and Cobaea (Prather and Jansen, 1998; Prather, 1999a, b) provide examples of this endeavor. A cautionary note on nomenclature Systematists generally recognize that there are two main goals of plant classification. Classification should reflect our understanding of phylogeny and provide a system that can be easily used to refer to plants (Cantino, Wagstaff, and Olmstead, 1998). Both of these goals are extremely important and every effort should be made to achieve them in tandem. Unfortunately, the dual goals are sometimes in conflict. Given the conflict discussed in this paper and the likelihood that still more nomenclatural changes will be made in the near future, the Polemoniaceae may exemplify the problem of developing a classification that meets these dual goals. At present we advocate a conservative approach to nomenclatural changes in the Polemoniaceae. For this reason we follow Grant (1998a) in recognizing the monotypic genus Microsteris, while some workers include the species in Phlox as P. gracilis E. Greene. The recognition of Microsteris is equivocal based on ITS sequences and cpdna restriction site data of Phlox given current sampling of the major lineages (Ferguson, Krämer, and Jansen, 1999; C. Ferguson and R. Jansen, unpublished data). If further study suggests that Phlox is paraphyletic to Microsteris or detailed morphological studies across the range of variation lead to a strong argument that characters used to segregate Microsteris are weak or problematic, it would be reasonable to reduce the genus to synonymy within Phlox. We also follow Grant (1998a) in including Loeseliastrum in Langloisia ( Langloisia s.l.). Little is to be gained by segregating three species between two genera, because they are morphologically very similar. There is more diversity between pairs of species in other genera [e.g., Cobaea scandens Cav. and C. penduliflora (H. Karst.) Hook. f. or Loeselia glandulosa (Cav.) G. Don and L. mexicana (Lam.) Brand] than among these three taxa. The phylogenetic relationships of these species are troublesome because Loeseliastrum was paraphyletic to Eriastrum and Langloisia s.s. in the ITS tree (Porter, 1996). Notably, Langloisia s.l. is monophyletic in the cpdna phylogenies (Fig. 3; Johnson et al., 1996). Subsuming Loeseliastrum does not remedy the potential problem of paraphyly but it does minimize a rather cumbersome nomenclature. This issue is best resolved in context of the phylogeny of the entire Loeselieae subclade; until such an undertaking is completed we advocate following Grant (1998a). Perhaps the example that best highlights our concerns regarding nomenclature is the ultimate disposition of species currently placed in Gilia s.l. The situation is extremely complex. In Grant s (1998a) revision he kept Gilia s.l. intact, with much the same composition as in the 1959 treatment (Grant, 1959). Contemporaneously, Porter (1998a, b) segregated Aliciella and Giliastrum from Gilia s.l. Later, Grant (1998b) reclassified Gilia and reduced Porter s Aliciella and Giliastrum to synonymy within Gilia and simultaneously segregated from Gilia s.l. two additional genera, Maculigilia and Tintinabulum. Additionally, Grant transferred one species of Gilia, G. tenerrima A. Gray, to Allophyllum (Grant, 1998b). Later, that same taxon was transferred to Tintinabulum, as T. tenerrimum (A. Gray) A. Day & V. Grant and four more species of Gilia were transferred to Allophyllum (Grant and Day, 1998). Based on molecular phylogenies there are several remaining Gilia species, aside from those already transferred or split into the four genera mentioned above, that render the genus polyphyletic. The recent trend has been to segregate most of the lineages into separate genera, whence came Aliciella, Giliastrum, Maculigilia, and Tintinabulum. If this continues, we estimate that there will be at least four, and possibly more, additional genera segregated from Gilia s.l. This situation is not unique to Gilia. Based on a perusal of available molecular evidence, many genera may not be monophyletic (Ipomopsis, Langloisia s.l., Linanthus, Leptodactylon, and Navarretia). We suggest that if these taxa are studied in context of their respective lineages with comprehensive sampling, preferably with both molecular and morphological data, we may discover that some of the genera are monophyletic, rendering nomenclatural changes unnecessary. Furthermore, if the phylogeny of a lineage is well understood it may reveal that, when nonmonophyly occurs, some of the species could be accommodated in existing genera. Thus, an increasingly cumbersome taxonomy would be avoided. The ndhf data provide insight into several important issues of Polemoniaceae phylogeny. Perhaps most interestingly, the data support monophyly of subfamily Cobaeoideae (excluding Loeselia) and suggest that Acanthogilia is basal to other members of the subfamily. Furthermore, for the first time, molecular data are available for Huthia and indicate that the genus is sister to Cantua. The subclades of the Polemonioideae clade identified by ndhf are identical in composition, allowing for sampling differences, to those identified by most other molecular analyses. This provides convincing evidence for monophyly of the four lineages. Because some relationships differ from analysis to analysis, and some relationships are weakly supported, we promote a cautious approach in incorporating molecular data into classification and nomenclature and suggest that phylogenetic analyses of morphological data are sorely needed at the generic level in the Polemoniaceae. LITERATURE CITED ANGIOSPERM PHYLOGENY GROUP An ordinal classification for the families of flowering plants. Annals of the Missouri Botanical Garden 85: BARRETT, S. C. H., L. D. HARDER, AND A. C. WORLEY The comparative biology of pollination and mating in flowering plants. Philosophical Transactions of the Royal Society, London B, 351: CAMPBELL, D. R Measurements of selection in a hermaphroditic plant: variation in male and female pollination success. Evolution 43: CANTINO, P. D., S. J. WAGSTAFF, AND R. G. OLMSTEAD Caryopteris (Lamiaceae) and the conflict between phylogenetic and pragmatic considerations in botanical nomenclature. Systematic Botany 23: CARLQUIST, S., V. M. ECKHART, AND D. C. MICHENER Wood anatomy of the Polemoniaceae. Aliso 10: CHASE, M. W., ET AL Phylogenetics of seed plants: An analysis of nucleotide sequences from the plastid gene rbcl. Annals of the Missouri Botanical Garden 80: CRONQUIST, A An integrated system of classification of flowering plants. Columbia University Press, New York, New York, USA. DAHLGREN, R A revised system of classification of the angiosperms. Botanical Journal of the Linnean Society 80:

9 1308 AMERICAN JOURNAL OF BOTANY [Vol. 87 DAY, A. G New taxa and nomenclatural changes in Allophyllum, Gilia, and Navarretia (Polemoniaceae). Novon 3: , AND R. MORAN Acanthogilia, a new genus of Polemoniaceae from Baja California, Mexico. Proceedings of the California Academy of Sciences 44: DOYLE, J. J., AND J. A. DOYLE A rapid DNA isolation procedure for small quantities of fresh leaf tissue. Phytochemical Bulletin 19: EPLING, C., AND T. DOBZHANSKY Genetics of natural populations. VI. Microgeographic races in Linanthus parryae. Genetics 27: FELSENSTEIN, J Confidence limits on phylogenies: an approach using the bootstrap. Evolution 39: FERGUSON, C. J., F. KRÄMER, AND R. K. JANSEN Relationships of eastern North American Phlox (Polemoniaceae) based on ITS sequence data. Systematic Botany 24: GRANT, V Natural history of the phlox family: systematic botany. Martinus Nijhoff, The Hague, The Netherlands a. Primary classification and phylogeny of the Polemoniaceae, with comments on molecular cladistics. American Journal of Botany 85: b. Classification of the genus Gilia (Polemoniaceae). Phytologia 84: , AND A. G. DAY Transfer of some species from Gilia to Allophyllum and Tintinabulum, and the effects of the transfer on the generic definition of Gilia (Polemoniaceae). Phytologia 84: , AND K. A. GRANT Flower pollination in the Phlox family. Columbia University Press, New York, New York, USA. GREENE, E. L Some American Polemoniaceae I. Pittonia 1: HARBORNE, J. B., AND D. M. SMITH Correlations between anthocyanin chemistry and pollination ecology in the Polemoniaceae. Biochemical Systematics and Ecology 6: JANSEN, R. K Current research. Plant Molecular Evolution Newsletter 2: JOHNSON, L. A., AND D. E. SOLTIS Phylogenetic inference in Saxifragaceae sensu stricto and Gilia (Polemoniaceae) using matk sequences. Annals of the Missouri Botanical Garden 82: , J. L. SCHULTZ, D. E. SOLTIS, AND P. S. SOLTIS Monophyly and generic relationships of Polemoniaceae based on matk sequences. American Journal of Botany 83: , D. E. SOLTIS, AND P. S. SOLTIS Phylogenetic relationships of Polemoniaceae inferred from 18S ribosomal DNA sequences. Plant Systematics and Evolution 214: KIM, K.-J., AND R. K. JANSEN Comparisons of phylogenetic hypotheses among different data sets in dwarf dandelions (Krigia, Asteraceae): additional information from internal transcribed spacer sequences of nuclear ribosomal DNA. Plant Systematics and Evolution 190: , AND ndhf sequence evolution and the major clades in the sunflower family. Proceedings of the National Academy of Sciences, USA 92: LOOCKERMAN, D. J., AND R. K. JANSEN The use of herbarium material for DNA studies. In T. F. Stuessy and S. H. Sohmer [eds.], Sampling the green world, Columbia University Press, New York, New York, USA. MADDISON, D. R The discovery and importance of multiple islands of most-parsimonious trees. Systematic Zoology 40: MORTON, C. M., M. W. CHASE, K.A.KRON, AND S. M. SWENSEN A molecular evaluation of the monophyly of the order Ebenales based upon rbcl sequence data. Systematic Botany 21: OLMSTEAD, R. G., H. J. MICHAELS, K.M.SCOTT, AND J. D. PALMER Monophyly of the Asteridae and identification of their major lineages inferred from DNA sequences of rbcl. Annals of the Missouri Botanical Garden 79: , J. A. SWEERE, AND K. H. WOLFE Ninety extra nucleotides in ndhf gene of tobacco chloroplast DNA: a summary of revisions to the 1986 genome sequence. Plant Molecular Biology 22: , K.-J. KIM, R. K. JANSEN, AND S. J. WAGSTAFF The phylogeny of the Asteridae sensu lato based on chloroplast ndhf gene sequences. Molecular Phylogenetics and Evolution 16: PAGE, R. D. M On islands of trees and the efficacy of different methods of branch swapping in finding most-parsimonious trees. Systematic Biology 42: PAIGE, K. N., AND T. G. WHITHAM Individual and population shifts in flower color by scarlet gilia: a mechanism for pollinator tracking. Science 227: PATTERSON, R. W., AND D. H. WILKEN Phlox. In J. C. Hickman [ed.], The Jepson Manual: higher plants of California, University of California Press, Berkeley, California, USA. PORTER, J. M Phylogeny of Polemoniaceae based on nuclear ribosomal internal transcribed spacer DNA sequences. Aliso 15: a. Nomenclatural changes in Polemoniaceae. Aliso 17: b. Aliciella, a recircumscribed genus of Polemoniaceae. Aliso 17: , AND L. A. JOHNSON Phylogenetic relationships of Polemoniaceae: inferences from mitochondrial nad1b intron sequences. Aliso 17: PRATHER, L. A A new species of Phlox (Polemoniaceae) from northern Mexico with an expanded circumscription of subsection Divaricatae. Plant Systematics and Evolution 192: a. Systematics of Cobaea (Polemoniaceae). Systematic Botany Monographs 57: b. The relative lability of floral vs non-floral characters and a morphological phylogenetic analysis of Cobaea (Polemoniaceae). Botanical Journal of the Linnean Society 131: , AND R. K. JANSEN The phylogeny of Cobaea (Polemoniaceae) based on sequence data from the ITS region of nuclear ribosomal DNA. Systematic Botany 23: SCHLICHTING, C. D., AND D. A. LEVIN Effects of inbreeding on phenotypic plasticity in cultivated Phlox. Theoretical and Applied Genetics 72: SPENCER, S. C., AND J. M. PORTER Evolutionary diversification and adaptation to novel environments in Navarretia (Polemoniaceae). Systematic Botany 22: STEELE, K. P., AND R. VILGALYS Phylogenetic analyses of Polemoniaceae using nucleotide sequences of the plastid gene matk. Systematic Botany 19: STUCHLIK, L. 1967a. Pollen morphology in the Polemoniaceae. Grana Palynologica 7: b. Pollen morphology and taxonomy of the family Polemoniaceae. Review of Palaeobotany and Palynology 4: SWOFFORD, D. L PAUP*: Phylogenetic analysis using parsimony (*and other methods). Version 4. Sinauer, Sunderland, Massachusetts, USA. TAYLOR, T. N., AND D. A. LEVIN Pollen morphology of Polemoniaceae in relation to systematics and pollination systems: scanning electron microscopy. Grana 15: WASER, N. M., AND M. V. PRICE Optimal outcrossing in Ipomopsis aggregata: seed set and offspring fitness. Evolution 43: WILKEN, D., AND R. L. HARTMAN A revision of the Ipomopsis spicata complex (Polemoniaceae). Systematic Botany 16: WOLF, P. G., P. S. SOLTIS, AND D. E. SOLTIS Phylogenetic significance of chloroplast DNA restriction site variation in the Ipomopsis aggregata complex and related species (Polemoniaceae). Systematic Botany 18: WRIGHT, S An analysis of local variability of flower color in Linanthus parryae. Genetics 28: