Aliso: A Journal of Systematic and Floristic Botany Aliso: A Journal of Systematic and Floristic Botany A Phylogenetic Classification of Polemoniaceae A Phylogenetic Classification of Polemoniaceae

Three hundred seventy nine species of Polemoniaceae are arranged in a phylogenetic classification composed of three subfamilies, eight tribes. and 26 genera. Nomenclature of one tribe is clarified and the circumscription of several tribes differs greatly from previous classifications. Five new genera, Bryantiella, Dayia, Lathrocasis, Microgilia, and Saltugilia, are proposed. In addition, four new species are described from the genera Allophyllum, Dayia, Giliastrum, and lpomopsis. This treatment repre sents a major reclassification with 59 new combinations, and the application of several additional combinations not used in recent years . carpels (bicarpellate in some Navarettia; septacidal in Cobaea, sometimes both loculicidal and septicidal in Acanthogilia). Petals convolute in aestivation (imbri cate in some Cantua). Annular nectary disk present; ovules unitegmic and tenuinucellate; endosperm devel opment nuclear. (including Cobeaceae D. Don, Edin burgh Philos. J. 10: 109, 1824.) Type: Polemonium L.


INTRODUCTION
Supergeneric classification in Polemoniaceae during the 20th century has undergone considerable flux, despite the appearance of relative stasis. In his monograph of Polemoniaceae for Engler's Das Pflanzenreich,  recognized 13 genera in two subfamilies and four tribes (Table I). Investigations by a number of students of the family during the next 30 years Jed to Wherry's (1940) revision, which recognized 19 genera in three tribes and three subtribes. Almost 20 years later, Grant (1959) reevaluated the work of earlier students and synthesized these works with his own biosystematic and experimental studies to erect a classification of 18 genera in five tribes. This classification has served as the framework for sampling in comparative and evolutionary studies for forty years (Table 1). A number of broad comparative surveys, conducted in Polemoniaceae during this interval (e. g., Stuchlik 1967a, b; Taylor and Levin 1975;Harborne and Smith 1978;Smith et al. 1982;Wilken et al. 1982;Plitmann andLevin 1983, 1990;Carlquist et al. 1984;Steele and Vilgalys 1994;Johnson et al. 1996 ;Porter 1996), were interpreted in the framework of Grant's classification. As a result, some workers suggested a revised taxonomy may better reflect the accumulating evidence of character distributions supporting alternative views of phylogenetic relationship (e. g., Smith et al. 1977;Day and Moran 1986;Johnson et al. 1996;Porter 1996). Of these studies, comparative DNA sequence analyses most forcefully stimulated the reexamination of Polemoniaceae classification. These data were clearly the impetus leading to Grant's (1998a, b) revised taxonomy of 20 genera in two subfamilies and eight tribes (Table I). Similarly, these data have stimulated us to examine carefully many classes of characters and their distribution in Polemoniaceae. Grant's (1998a, b) newest classification, while in some instances remedying character conflict imposed by his earlier system (Grant 1959), inadequately reflects strongly supported and corroborated evidence for the underlying phylogeny of Polemoniaceae (Johnson et al. 1996;Porter 1996;Porter and Johnson 1998). Examination of both morphological and molecular data reveals well-corroborated patterns, providing evidence of the phylogeny of this family. Here we propose a new classification of Polemoniaceae, developed to reflect the phylogeny of this complex family.

Philosophy of Classification
Classifications frequently function, for better or for worse, as proxies for phylogeny in comparative studies (e.g., Taylor and Levin 1975;Rickleffs and Renner 1994). The inferential accuracy achieved in such studies depends upon how well the classification accurately reflects phylogeny. Though it is not possible to literally observe the diversification events of the past, philosophical and methodological advances over the last 40 years have provided a valuable framework for reconstructing phylogeny. These tools, including cladistic analyses, have grounded systematics as a science based in hypothesis testing (classifications representing testable hypotheses of relationship), using repeatable methods, and providing a means for assessing the taxonomic value of a character or intrinsic levels of data support for a particular group. We do not suppose that systematics has reached a methodological pinnacle, that current algorithmic data analyses can resolve all of the complex diversification events produced by nature, nor that interpretation has no role in modem systematics. We do suggest, however, that appropriate tools for implementing strong inference (Platt 1964) in systematic studies exist. The application of these tools, while recent in origin relative to less explicit approaches, can provide important insights into phylogeny of use to all taxonomists involved in classifying organisms (see Mayr, 1995 , for a similar conclusion albeit from a different perspective).
The classification presented here is an attempt to model what we have discovered regarding phylogeny in the Phlox family through our field and laboratory studies, observations, and primary literature (Fig. I); the classification reflects the phylogeny, rather than the phylogeny being determined a posteriori from the classification (Grant 1998a). Recognition of monophyletic groups has been central to our efforts. We use "monophyly" throughout tills paper sensu Hennig (1966). That is, an ancestor and all of the descendants of that ancestor represent a monophyletic group. This corresponds to the term "holophyletic" of Ashlock (1971). As a classification criterion, monophyly has received its share of criticism (e. g., Cronquist 1987 ;Sosef 1997). Monophyly, however, provides a strong theoretical foundation for comparative biology (e.g., Donoghue and Cantino 1988;Snow 1996) and reflects our view that classifications should mirror evolutionary patterns of diversification.
The processes of diversification are the processes of evolution; that is, descent with modification. These processes have produced a genealogical connection, or phylogeny, between all organisms. Characters, whether morphological traits, palynology, flavonoid chemistry, reproductive biology, or DNA sequences, are the products of tills genealogical descent as modified by mutation, drift, and selection. Each character may provide some degree of corroboration of the phylogeny, as well as some misleading information attributable to homoplasy or in accuracies in homology assessment. It is through the synthesis of all lines of evidence that the most reliable estimation of the phylogeny of a group will be found and genealogical connections discovered.
The classification presented here reflects such a synthesis. Published DNA sequence analyses (Steele and Vilgalys 1994;Johnson and Soltis, 1995;Johnson et al. 1996Johnson et al. , 1999Porter 1996;Spencer and Porter, 1997 ;Porter and Johnson 1998; Prather and Jansen 1998; Subfam Bell et al. 1999;Fergussen et al. 1999;Bell and Patterson 2000), as well as an equal volume of unpublished DNA data, have contributed heavily toward phylogeny estimation in Polemoniaceae and hence this classification. DNA sequences remain the only broadly surveyed and published class of data at the family level in Polemoniaceae that have measurable levels of support. Even so, sequences and cladistic analyses are by no means the only source of evidence considered by us. We have attempted to evaluate and incorporate all published comparative and biosystematic data. In addition, when feasible, we have reexamined many morphological, palynological, and anatomical features ourselves to verify published interpretations and to characterize states for many species not previously reported. Preliminary cladistic analyses of comparative morphological data (Porter. unpubl.) provide very few inferred groups with even moderate support due to extensive homoplasy in these data; nonetheless, examining patterns of character distribution has been enlightening. In some instances, morphology was considered "more compelling" evidence of common an-cestry than gene sequence data; in other instances, the reverse was true.

Scope and Format
Supergeneric classification primarily is of interest to systematists and comparative biologists, whereas interest in taxonomic changes involving generic circumscriptions and species relationships extends to government agencies, field biologists, and amateurs. Both types of change are understandably met with reserve and sometimes resistance. We have taken neither change lightly. We deliberated extensively between ourselves, and with others, on the bases and merit for all taxonomic change. We considered keeping some historical groups of doubtful monophyly for the sake of convenience and tradition while continuing to gather an " insur mountable body of evidence" before approaching the systematic community. At the same time, we surveyed and reviewed many classes of data with sufficient depth and breadth to express confidence in the composition of several lineages of Polemoni- Porter and lohnson ALISO Fig. I . Phylogenetic tree depi cting hypothesized relationships between genera of Polemoniaceae. The structu re of the [fee is based upon comparative morphological and molecular (DNA sequences of rnITS. cpmo l K. ctnrnl.-F, and mtnadl intron ) data, The circled groups represent major lineages. emphasized in the phylogenetic classification.
aceae not recognized by any currently accepted taxonomy. Ultimately, we abandoned tradition when corroborating data provided reasonable, substantive alternative hypotheses of phylogeny coupled with sufficient indication that the traditional groups in question were dubiously supported.
This classification is hierarchically arranged. Ranks of taxa are generally in correspondence with the traditional taxon inclusion of ranks in Polemoniaceae. However, taxa of equal rank are not intended to represent sister groups per se, or groups of equal age ; taxa are simply monophyletic groups. Consistent with this form of presentation, taxa at lower ranks are presented alphabetically within the taxa of higher ranks to which they belong. Additionally, this classification is exclusive of subgeneric ranks other than species; infrageneric ranks are considered valuable, but their circumscription extends beyond the scope of this paper. Their omission here reflects neither endorsement nor rejection of the subgeneric and sectional classifications of others. Recent and forthcoming contributions (e.g., Porter 1998a;Prather 1999) address subgeneric and/or sectional classification of several genera in Polemoniaceae.
For each rank, the correct name of the taxon with full citation is provided. The nomenclature has been checked with the source literature, with the exception of Amott's (1832) treatment in Encyclopedia Britannica, which was unavailable to us. For the sake of brevity, synonymy generally is not provided; however, some synonymy is included to clearly indicate circumscription of taxa. Following the family, subfamily, tribe, and genus names, brief descriptions are provided . A phylogenetic delineation, or "clade address," for supergeneric ranks follows. The phylogenetic delineation provides an explicit identification of the monophyletic group to which we associate a name, and specifies the meaning of a taxon name in terms of ancestry and descent (de Queiroz 1997) . This delineation (or definition) may take one of three forms: node-based, stem-based, or apomorphy-based (see de Queiroz and Gauthier 1990. Node-based definitions have the following form: the monophyletic group arising from the most recent common ancestor of taxon A and taxon B. In some cases we specify that the group is the "least incorporative," or the smallest monophyletic group that simultaneously includes two or more taxa. Less frequently, we employ a stem-based definition in the form: the most inclusive (or largest) monophyletic group that includes taxon C and taxon 0 , but does not include taxon A, or taxon B. Apomorphy-based definitions are of the form: the monophyletic group arising from the first common ancestor of taxon A to evolve a particular character. Be cause we are classifying under International Code of Botanical Nomenclature (lBCN) rules (Greuter et al. 1994), we have used type species to serve as taxon specifiers (taxon A or taxon B, above). The logic being that, for example, if the type of Cilia is within a clade, then the generic name must be associated with at least some portion of that clade. Following the phylogenetic delineation, the membership of each taxon is indicated. At the generic rank, a listing of species indicates membership. We have attempted to provide a complete and accurate listing of species; however, we do not consider ourselves experts in all genera, and some errors due to omission or inaccuracies may exist. Frequently, there are notes following the descriptions of the taxa that provide additional information regarding recognition. Following the classification, we present a generic key to Polemoniaceae.

RES ULTS AND DISCUSSION
An evaluation of molecular, morphological, palynological, karyological, phytochemical, and ecological data in Polemoniaceae provides an immediate realization of the immense complexity that has arisen during the diversification of this family. Homoplastic evolution is common. No fewer than eight genera possess some species occurring in very xeric habitats and other species that inhabit relatively mesic sites. Similarly, nine genera have species that display floral syndromes associated with hummingbird pollination (Y. Grant and K. Grant 1965;Porter 1993), and ten genera display floral syndromes associated with hawkmoth pollination. At least six genera have both hawkrnoth and hummingbird, as well as bee, fly, or beetle pollination. Given that multiple lineages appear to have undergone independent morphological change associated with shifts in habitat preference, and, at the same time, diversified greatly in pollination mechanisms, a high degree of homoplasy should not be surprising. Extensive homoplasy is reflected in the difficulty all workers in Polemoniaceae have faced in elucidating higher-order relationships in this family: all but a few genera are defined by suites of characters and exclusive diagnostic features for supergeneric groups are elusive.
Hybridization may also have contributed to complex patterns of variation in the Phlox family. Hybridization has been suggested to play a role in species origins, both at the diploid (Grant 1959) and the polyploid Tommerup and Porter 1996) levels, and has been hypothesized to have also played a role in the origin of more diverse lineages (Grant 1959). The patterns of distribution of morphological traits resulting from hybridization are varied and not easily predicted (McDade 1992). However, in the absence of substantive evidence for higher-order reticulation in the Phlox family, corroborative data supporting a model of simple divergence deserve consideration. Analytical methods under development that accommodate reticulate processes are promising (e. g., Hein 1993;Alroy 1995;Dickerman 1998), and such methods may provide new insights in Polemoniaceae in conjunction with additional data.
In Polemoniaceae, explicit analyses of congruence between nuclear and chloroplast genes reveal similar historic patterns of gene descent among genera for the regions surveyed (Johnson et al. 1995;Johnson 1996). Minor discrepancies exist between some species within genera, and in some weakly supported supergeneric relationships. Similarly, explicit comparisons of morphological data with DNA-based hypotheses show either no, or only weak conflict (Porter, unpubl.). Thus, while processes exist that can lead to discordance between gene-trees and species trees (e.g., Wendel and Doyle 1998), hard evidence for such discordance in the Phlox family does not currently exist. Furthermore, what is true for the gene, is true for the genotype and phenotype. Both morphology and gene sequences have the potential for providing misleading information, if approached with naivete. However, they both have great promise if considered in concert and through corroboration. We contend that, in concert, morphological and molecular data provide an indication of the underlying phylogeny of Polemoniaceae.
This phylogenetic classification of Polemoniaceae (Table 2, and below) includes three subfamilies, eight tribes, and 26 genera for the 379 species of Polemoniaceae. Nomenclature for one tribe is clarified and the circumscription of several tribes greatly differs from previous classifications of this family (Table 1). We propose five new genera and describe four new species. This treatment represents a major reclassification, resulting in 59 new combinations. We restate our belief that an accurate and unambiguous reflection of phylogenetic relationships in classification supersedes the inconvenience of nomenclatural change. As with Grant (l998a: 744), we consider classifications to represent hypotheses that "can be tested, supported, falsified, or modified by subsequent evidence." This classification is our attempt to incorporate concrete hypotheses of monophyly, based on our current understanding of phylogeny. We view this contribution as the commencement, rather than the consummation, of phylogenetic classification in the Phlox family, and welcome continued exploration of relationships. We anticipate additional, carefully considered data, both morphological and molecular, that may either corroborate or refute the hypotheses of relationships herein presented.
POLEMONIACEAE Juss. Gen. pL., p. 136, 1789. Vines (Cobaea), small trees (Cantua), shrubs, perennial, biennial, or annual herbs. Flowers typically of five sepals, five fused petals, and five epipetalous or basally adnate stamens (but perianth and androecium consistently 4-or 6-merous in some Linanthus). Superior ovary of three fused and loculicidally dehiscing Polernoniaceae, including Cobaea, are supported strongly by cladistic analyses as a monophyletic group that originated as part of an ericalean-ebenalean diversification. The sister family of Polemoniaceae is as yet uncertain (Johnson et al. 1996;Porter 1996;Nandi et aI. 1998;Porter and Johnson 1998;Johnson et aI. 1999). The origin of Polemoniaceae among ericalean and ebenalean families impacts the interpretation of traits heretofore considered characteristic of (i .e ., putative synapomorphies for) the family. For example, tricarpelly is not uncommon in the ericalean-ebenalean families (e.g., Clethraceae, Diapensiaceae, and Fouquieriaceae), and therefore is doubtfully a synapomorphy for Polemoniaceae. Phylogenetic delineation.-The monophyletic group des cended from the first ancestor of Acanthogilia gloriosa to evolve persistent, spinescent primary leaves, formed by the lignification of the vascular bundles of the leaf, and simultaneously zonocolporate pollen bearing verruca.

Membership.-Acanthogilia .
As a taxon, the name Acanthogilioideae is redundant with Acanthogilieae, and Acanthogilia; recognition of this subfarn.ilial rank is maintained for uniformity in treatment in this classification. However, as a monophyletic group, Acanthogilioideae are identified in a broader context that does not preclude the possibility that unknown species may also belong within this subfamily (i.e., more incorporative than only Acanthogilia gloriosa). Our exclusion of Acanthogilia from Cobaeoideae of Grant (1998a), and from our Polemonioideae, emphasizes the morphological intermediacy between this taxon, Cobaeoideae, and Polemonioideae. This monotypic taxon occupies a pivotal position with respect to the root of Polemoniaceae, and similarities in the morphological features of Acanthogilia exist with both subfamilies (Day and Moran 1986). For example, the verrucate pollen of Acanthogilia is reminiscent of the insulae (verrucae) on the pollen of Cantua (Day and Moran 1986). Even so, the pollen of Acanthogilia is zonocolporate, as is found in many members of Polemonioideae, rather than pantoporate, as is observed in Cobaeoideae (Stuchlik 1967a; Day and Moran 1986), and verrucate pollen also occurs in the Giliastrum foetidum (Gillies ex Benth.) J.M. Porter of the Polernonioideae. Like most members of Polemonioideae, Acanthogilia has a low , presumably diploid chromosome number of N = 9, in contrast to the high, presumably polyploid chromosome numbers of Cobaeoideae (N = 15,26,27). Seedlings of Cantua and Cobaea are described as possessing " large cordate or ovate cotyledons" (Grant 1959: 9). The cotyledons of seedlings of Acanthogilia are linear and acute (Porter unpubI.), whereas the cotyledons of members of Polemonioideae vary from ovate to linear. Collectively, there seems to be no overwhelming morphological evidence supporting any specific sister relationship for Acanthogilia. In addition, molecular cladistic studies are unclear concerning the sister group relationship of Acanthogilia. This taxon may be sister to the so-called "tropical" genera (Cobaeoideae), the "temperate" genera (Polemonioideae), or to all other members of Polemoniaceae.
Phylogenetic delineation.-See Acanthogilioideae, with which this taxon is redundant; recognition of this tribal rank is maintained for uniformity in treatment in this classification. Acad. Sci. 44: 111-115, 1986. Shrubs with persistent, alternate, spiny, pinnately lobed primary leaves and deciduous, linear secondary leaves that are axillary and fascicled on short shoots; glandular throughout. Calyx tube with a hyaline interval alternating with the herbaceous costae, the lobes with narrow hyaline margins, apex mucronate. Corolla funnelform to salverform and mostly radially symmetric. Pollen zonocolporate, with semitectate exine (perforate or perreticulate), insulae (verrucae) present. Fruits dehiscing only about half the length, either loculicidal or both loculicidal and septicidal. Seeds winged. N = 9. One species.
This circumscription of subfamily Cobaeoideae differs from Grant (1998a) in excluding Acanthogilia and Loeselia. The former taxon is here accorded in its own subfamily; the latter is placed in Polemonioideae. Encompassing considerable morphological diversity, Cobaeoideae separated early in the diversification of extant Polemoniaceae. The group as a whole is characterized by high, but not equivalent, chromosome numbers that may either stem from a single polyploid origin followed by several aneuploid changes, or multiple episodes of polyploidy and aneuploidy (Grant 1959;Johnson et al. 1996). Further investigations may allow a determination of which hypothesis explaining change in chromosome numbers in this group is most consistent with observed genome structure. BONPLANDIEAE Baill. Hist. pi. v.lO, p.342, 1890. Herbaceous to woody perennials, flowering the first year. Leaves alternate, broad, serrate or pinnately lobed, to linear and entire. Flowers axillary and paired. Calyx and corolla bilaterally symmetric. Seeds narrowly winged to wingless. Type: Bonplandia Cav.
Phylogenetic delineation . -The most incorporative monophyletic group of species that includes Bonplandia geminiflora and B. linearis, and shares a more recent common ancestor with B. geminiflora than with Acanthogilia gloriosa, Cantua buxifolia, Cobaea scandens, Gilia laciniata, Loeselia ciliata, Phlox glaberrima, or Polemonium caeruleum. Membership.-Bonplandia. As a taxon, Bonplandieae is redundant with Bonplandia; recognition of this tribal rank is maintained for uniformity in treatment in this classification.

1800.
BONPLANDIA LlNEARIS B. L. Rob., Proc. Amer . A cad. Arts. 43: 24 . 1907. The two species of Bonplandia are distinguished on the basis of leaf and floral morphology: Bonplandia geminiflora has narrow to broad, serrate, Iyrate or pinnately lobed (but highly variable) leaves, and corolla (20-)22-29 mm long with spreading lobes; whereas, B. linearis has narrow, linear, entire leaves, and corolla 15-18(-20) mm long with reflexed lobes. Rzedowski et al . (1995) suggest reducing B. linearis to synonymy under B. geminifiora, citing the highly variable leaf morphology. However, floral morphology provides ev-idence contrary to such a treatment. Robinson (1907) described the corolla size of B. linearis as 20 mrn , which falls between the ranges of the two species here cited . This raises the question as to whether the holotype of B. linearis (not seen) represents the largeflowered or the small-flowered species. If the former is true, then Rzedowski et al. (1995) are indeed correct, and there remains an undescribed , small flowered species of Bonplandia. Directed studies are needed in this genus. Pflanzenfam. IV 3a, pA5 , 1891 . Branching shrubs to rarely small trees. Leaves alternate or fasciculate, coriaceous to fleshy, entire to deeply pinnately divided. Flowers tubular to salverform , imbricate or convolute in aestivation, calyx wholly herbaceous, or hyaline between vasculature. Seeds broadly winged. Type: Cantua Juss. ex Lam.
Phylogenetic delineation.-The most incorporative monophyletic group of species that includes Cobaea minor and Cobaea scandens, and that shares a more recent common ancestor with Cobaea penduliflora than with Acanthogilia gloriosa, Bonplandia geminiflora, Cantua buxifolia, Cilia laciniata, Lo eselia ciliata, Phlox glaberrima, or Polemonium caeruleum. Membership.-Cobaea. As a taxon, the name Cobaeeae is redundant with Cobaea; recognition of thi s tribal rank is maintained for uniformity in treatment in this classification.
COBAEA Cav., Icon. v.l , p.l l , 1791. Tropical vines, somewhat woody at the base. Leaves alternate, pinnate-compound with terminal leaflet modified into a tendril. Inflorescences reduced to axillary, long pedunculate, solitary flowers. Calyx radially syrn-metric, lobes nearly free to the base, herbaceous, no hyaline portion. Corolla campanulate, to nearly rotate, large, radially to bilateraJly synunetric, the lobes broad to narrow and caudate. Stamens included to exserted, adnate and affixed at the base of the corolla tube, base of filaments glabrous to densely hairy; pollen pantoporate, with reticulate exine. Capsules with septicidal dehiscence. Seeds large, broadly winged. N = 26. Eighteen species. Type: Cobaea scandens Cav.
Polemonioideae, though not paralleling Cobaeoideae in extreme morphological features (e.g., arborescent or vining habit, very large flowers, high chromosome numbers, etc.), exceed Cobaeoideae in degree of morphological diversity. Species of Polemonioideae range from near leafless annuals to woody subshrubs. Loeselia, retained by Grant (1998a) in Cobaeoideae, is placed in Polemonioideae and certainly shares recent conunon ancestry with members of this subfamily as here circumscribed, to the exclusion of members of Cobaeoideae and Acanthogilioideae (Johnson et Porter 1996;Porter and Johnson 1998). This subfamily frequently is referred to as the "temperate group." However, it should be noted that many members of Polemonioideae occur in the tropics and subtropics of North and South America, in addition to temperate regions. Tribal composition within our Polemonioideae differs substantially from that presented by Grant (1959Grant ( , 1998a. Herbae annuae (vel perennes rariores ut in Collomia) erectae glandulosae vel glabrae. Folia altema integra vel pinnatisecta (vel rara palrnata), Folium caulinum foliis rosularibus simile vel deminutus, vel folium congestus infra inflorescenciam. Calyx fructifer accrescens vel rumpens. Flores regulares et symmetrici. Corollae infundibuliformis vel hypocrateriformis. Pollinis albi vel caerulei. Seminis pallidi vel atrobrunnei. Typus: Gilia Ruiz et Pavon AnnuaJ herbs, rarely perennial (in Collomia); glandular throughout or primarily in the inflorescence.
We formally describe Tribe Gilieae, following many years of misapplication of this tribal name. The use of Gilieae at the tribal rank (e.g., Grant 1959Grant , 1997Grant , 1998a rests on Reichenbach's (1837) classification, subdividing Convolvulaceae into three "Groups" (Gruppen): Polemoniariae, Hydroleeae, and Convolvuleae. Polemoniariae are also divided into three groups (Phloginae, Gilieae, and Polemonieae) but no rank is specified. The Code does not provide a status for these rankless categories. It might be inferred that Reichenbach intended the designations to be equivalent to the tribal rank, given the ending of the names (e.g., Gilieae). But, because name endings were not standardized in 1837, and Reichenbach did not indicate a rank, such a supposition is groundless. Regardless, these names are subdivisions of Convolvulaceae, not Polemoniaceae. Grant (1997) has suggested that in taking up Reichenbach's Gruppe Gilieae and treating it at the tribal level, in his 1959 Natural History of the Phlox Family, the name is legitimized, according to the code. While he did provide a description (p . 51) and indirectly cited the type (p. 120) , Grant failed to identify that he was changing the rank of the taxon from unranked to tribe (i. e., stat. nov. or trib. nov.) or providing a new combination (transferring Gilieae from Convolvulaceae to Polemoniaceae, i.e., comb. nov.), as required by the 1959 Code. Unlike an orthographic error, the rank of a taxon cannot be changed as a correction; the taxon clearly must be transferred in rank, or redescribed at the tribal rank. Rather than continue the confusion involving the application of Reichenbach's Gruppe Gilieae, we describe thi s tribe anew.
Tribe Gilieae, as here circumscribed, differs considerably in generic inclusion from that of Grant (1959Grant ( , 1998a. Grant, EI Aliso 3: 98, 99, 1955. Annual herbs, erect to spreading, simple or divaricately branched, glandular. Leaves often forming a basal rosette, alternate, entire to once or twice pinnatifid, upper cauline leaves nearly palmate, lobes linear to narrowly oblanceolate, the tip not acute or mucronate. Flowers sessile to long pedicellate, somewhat congested or open inflorescences. Calyx tube with narrow hyaline intervals alternating with herbaceous costae, the lobes equal in length, not acute or mucronate. Corolla funnel form, radially to bilaterally symmetric. Stamens equally or unequally inserted in the distal corolla tube, the filaments unequal or nearly equal in length; pollen pantoporate, with pertectate-perforate exine and spinule-like processes. Seeds one to three per locule, producing copious mucilage (i.e ., spiricles) when wetted. N = 8, 9. Six species. Type Grant. El Aliso 3: 104, 1955.
El Aliso 3: 102, 1955. Cronquist (1984: 132) was "not convinced . .. that Allophyllum is actually more closely related to Collomia than to Gilia," and did "not see how it can reasonably be extracted from the remainder of the genus Gilia." Comparative DNA sequence data (Johnson et al . 1996, Porter 1996, however, strongly support the morphological evidence, cited by Grant and Grant (1955), providing additional support for the segregation of Allophyllum. As circumscribed here, Allophyllum does not incl ude Lathrocasis tenerrima (for specific reasons, see below) as suggested by Grant (I 998b); however, we include a previously undescribed species from southern Baja California, Mexico: Johnson, sp. nov .
This species is endemic to moist pockets of rocky, granitic slopes, mostly in the shade of larger rocks, 1000-2000 m., in the Sierra San Borja and Sierra la Asamblea, flowering March through April.
Repr esentative specimens.-MEXtCO. B  Restricted to desert mountains of the Vizciano phase of the Sonoran Desert, Allophyllum nemophilophyllum is ecologically unique in the genus: it is the only true subtropical, desert species. Morphologically it differs from all other species of Allophyllum in the exceptionally long pedicels (11-33 mm long). The small, white, radially symmetric flowers also serve to distinguish this species from A. glutinosum (with bilateraly symmetric, blue-violet flowers) and A. gilioides (with radially symmetric, dark blue-violet flowers), the only other representatives of this genus in Baja California, Mexico. As the specific epithet suggests, the lower cauline leaves are similar in morphology to some desert species of Nemophila (Hydrophyllaceae).
COLLOMIA Nutt., Gen. Amer. v. I 126, 1818. Annual or perennial herbs, erect to spreading, simple or divaricately branched , hairy or glandular. Leaves generally not forming a basal rosette, alternate, simple, entire or serrate to once pinnatifid, upper cauline leaves sessile. Flowers in terminal head -like inflorescences, rarely 1-3 in axils of leaves. Calyx tube with narrow, obscure hyaline intervals alternating with herbaceous portion, the lobes equal or unequal in length, sinuses pleated, calyx enlarging in fruit. Corolla funnelform to salverform, radially symmetric. Stamens equally or unequally inserted in the distal corolla tube, the filaments unequal or nearly equal in length; pollen zonocolporate or pantoporate, with striate , striate-reticulate, or reticulate exine. Seeds mostly one per locule (to three in two species), producing copious mucilage (i.e., spiricles) when wetted. Collomia is often characterized in the literature as possessing an entirely herbaceous calyx. However, the accrescent calyces in Collomia, which are easily distinguished from all other genera in this family, are composed of wide, rhombic, herbaceous lobes connected ± mid-length by a narrow yet identifiable membrane that folds outward at the sinus to form a pitcherlike projection. This arrangement gives the calyx a distinctive replicate or distended appearance. Most, if not all, species in this genus also possess explosively dehiscent capsules: in species with one seed per locule, each valve folds backwards lengthwise along the septae whereas in species with three seeds, each valve folds more or less horizontally between the seeds. The inclusion of Collomia heterophylla is supported equivocally by molecular cladistic analyses (Johnson and Porter unpubl.), but it is inferred to be part of a monophyletic Collomia, based on morphological and palynological evidence (Chuang et al. 1978).
GlLI A Ruiz & Pav., Fl. peruv. prodr., p.2S, t. 4, 1794. Annual herbs, glandular, variously hairy, or almost glabrous, generally with a basal rosette of leaves. Leaves highly dissected, 1-3 pinnatifid, to dentate or rarely entire, the ultimate segments sometimes toothed; cauline leaves alternate, becoming reduced above, often greatly so . Inflorescence cymose, the flowers in dense heads, glomerate, loose, or solitary. Calyx tube with hyaline intervals alternating with the herbaceous costae, the lobes with hyaline margins, acute to mucronate. Corolla funnel form to salverform, large and showy to very small and inconspicuous, the tube narrow often with a characteristic flared region ("throat"), lobes long or short, spreading. Stamens mostly equally inserted near the sinuses of the corolla lobes, the filaments equal or unequal in length; pollen zonocolporate, apertures rarely anomotreme, with striate or rugulate exine. Seeds many per locule, producing copious mucilage (i.e. , spiricles) when wetted; each outer cell wall of the seed coat smooth when shed. N = 9, 18, 36. of section Saltugilia (Shevock and Day 1999), is also retained in Cilia. In our estimation, this species belongs with the leafy-stemmed gilias with close affinities to Cilia capitata, but simply lacking a capitate inflorescence. The seeds of all gilias produce copious mucilage (i.e., spiricles) when wetted, and the outer wall of the testa is shed as an areolate, but not verrucate, sheet (with areoles defined by cell boundaries). Fine, tangled, white arachnoid-tomentum is also shared by most species, at least in the leafaxils, but often on leaves and in the inflorescence.  A. Johnson Annual herbs, densely stipitate glandular on stems and pedicels, sparsely glandular on leaves and calyces.

Johnson
Lathrocasis tenerrima (A. Gray) L. A. Johnson. comb. nov. Basionym: Cilia ten errima A. Gray Inconpicuous and largely ignored, this species has recently been subjected to several generic realignments. Affinities with Allophyllum were suggested based largely on inaccurate interpretation of pollen morphology (Grant 1998b), Affinities with Tin tinabulum (Grant and Day 1998)  possesses a distinct suite of morphological features that, considered as a whole, distinguish this taxon at the generic level. This suite of characters includes the mostly entire, nonmucronate leaves, effuse branching with diverging to retrorse and threadlike pedicels, uniformity of short stipitate glands throughout, uniovulate locules, spiriliferous seeds with verrucate seedcoats, minute funnelforrn flowers with simple vasculature, and stamens equally inserted approximately mid-tube.
DNA sequence data (chloroplast matK, tmL-F, and nuclear ITS genes) of multiple populations of L. tenerrima corroborate this distincitive morphological characterization and show Lathrocasis well separated at the nucleotide sequence level from its nearest relatives in Gilieae (Johnson and Weese 2000).
Lathrocasis derives its name from the combination of the Greek "lathro-" (hidden, secret) and "kasis" (sister), True to its name, it is not clear presently whether Lathrocasis is sister to Cilia. a group composed of Allophyllum, Collomia, and Navarretia, or both of these groups combined (Johnson and Weese 2000). NAVARRETIA Ruiz & Pav., Fl. peruv. prodr., p,20, 1794. Annual herbs, erect or prostrate, simple or divaricately branched, variously hairy, glandular or glabrous, Leaves generally not forming a basal rosette, alternate or lower most opposite, entire to once or twice pinnatifid; upper cauline leaves usually bracteate, sometimes palmately lobed to divided, acerose or spine-tipped. Flowers usually sessile and congested in a head-like inflorescences, but pedicellate and in pairs in Navarretia capillaris, N. leptalea, and N. sinistra.
Phylogenetic analyses of chloroplast rnatK (Johnson and Soltis 1995;Johnson et al. 1996) and nuclear ITS (Porter 1996;Spencer and Porter 1997) sequences provide strong evidence that Navarretia is closely allied with CoLlomia, A LlophyLlum, and three species formerly recognized in Cilia section KeLloggia (D ay 1993a). Navarretia is here circumscri bed to include these latter species (i.e. , N. cap iLlaris, N. leptalea, and N. sin istra). The recent placement of these three species in ALlo-phyLlum (Grant and Day 1998) is understandable given the close affinities of ALlophyLlum to Navarretia and CoLlomia . These three genera possess distinct gross morphologies, as do the "kelloggioid Na varretias" : Even so, examination of the morphological traits, considered to be important indicators of relationship (Grant and Day 1998), provide support for placement of "kelloggioid Navarretias" in Navarretia rather than ALlophyLlum (Porte r and Johnson unpubl. ).
Navarretia cap iLLaris, N. LeptaLea, and N. sinistra are distinguished from other Navarretia on the basis of architecture and traditional " key characters" (i.e., they lack a dense, spinescent inflorescence and po ssess calyx lobes of equal length). Their inclusion adds heterogeneity at the gross-morphological level to an otherwise cohesive group. Nevertheless , morphological and molecular data indicate the se species are phylogenetically removed from Gilia and near Na varretia. In determining how best to treat N. cap iLLaris, N. Lep-taLea, and N. sinistra taxonomicaJly, we at first considered erecting a new genus. Isozyme and distribution-wide population sampling of both chloroplast and nuclear genes, however, indicate these three species do not form an exclusive monophyletic group (Johnson, unpubl.). An cient hybridization may be involved in the origin of at least some taxa. If so , their exact parentage has been ob scured by time; nonetheless, ancestors of Navarretia, but not CoLlomia or ALLophyLlum, appear implicated mo st stro ngly (Johnson, unpubl.). By including N. capiLLaris, N. leptalea, and N. sinistra in Nava rret ia, we provide a treatment that emphasizes monophyly in the classification of these species. Any attempt to classify N. capillaris, N. LeptaLea, and N. sinistra in any single genus exclusive of Navarretia would make the alternative genus polyphyletic, and Navarretia paraphyletic.
Examination of fine-scale morphological features yields support for this taxonomy despite gross morphological appearances. Navarretia capillaris, N. leptalea, and N. sinistra possess linear cotyledons as do most other Navarretia (but not Allophyllum or Collomia). They are also most similar to other Navarretia in seed morphology and seed germination requirements (involving long periods of moist chilling; Johnson pers. observ.). Additionally, N. capilla ris, N. leptalea, and N. sinistra possess pantoporate pollen grains, a type ubiquitous in Navarretia, Allophyllum, and present in some Collomia, but unknown in Gilia. Exine sculpting, however, differs greatly among these species (Spencer and Porter 1997;Johnson unpubl.). Navarretia capillaris and N. leptalea have a striate-reticulate exine also found in N. hamata and N. prolifera (striato-reticulate type II of Spencer and Porter 1997); however, N. leptalea subsp. bicolor has variable pollen, some with the striate-reticulate exine similar to that of N. breweri (striato-reticulate type I of Spencer and Porter 1997). This contrasts with the rugulose, superficially pertectate pollen of N. sinistra, which in some ways is similar to the pollen of N. leucocephala, N. plieantha, N. prostrata, N. myersii, and N. fossalis, as well as Allophyllum. This exine diversity further supports the hypothesis that N. capillaris and N. leptalea form a lineage apart from N. sinistra (Johnson et al., 1994;Johnson 1996;unpubl.). Additional comparative surveys of micromorphological characters and two additional nuclear genes are underway to provide further insights into the history of these species (Johnson unpubl). Erect annual herbs; stems typically I, much branched above, glabrous, glaucous, or glandular, mostly 15-100 em in height. Leaves principally in a basal rosette, (1-)2(-3) pinnatifid, reduced upwards and ultimately subulate bracteate. Trichomes on lower leaves and stem translucent, tapering, terminating in a single, minute, globose, glandular cell (gland obscure; appearing chain-like villous when dry). Trichomes on upper (and lower in some) stems, pedicels, and calyces short stalked and terminating in a flat-topped multicellular gland. Inflorescence cymose, loose, flowers borne singly or paired on unequal to subequal pedicels. Flowers mostly showy. Calyx lobes equal, the herbaceous costae narrower than and united above midlength by a hyaline membrane. Corolla radially symmetric, funnelform, yellow spotted on throat. Stamens unequal to subequal, the shortest one often borne at right angles to the corolla wall. Pollen blue, zonocolporate, with striate exine. Seeds light tan, angular, many per locule, producing copious mucilage (i.e., spiricles) when wetted; seed coat verrucate. N = 9. Three species. Type: Saltugilia grinnellii (Brand) L.

A. Johnson
SaItugilia grinnellii (Brand )  Saltugilia (=woodland gilias) was first accorded nomenclatural status as a section of Gilia when Grant (1954c) placed G. splendens, G. caruifolia, and G. australis with G. leptalea and G. capillaris. A more recent circumscription of this section (Shevock and Day 1998) reflects the inclusion of G. scopulorum and G. stellata by Grant (1959), the removal of G. leptalea and G. capillaris by Day (1993a), and the incusion of G . yorkii by Shevok and . Saltugilia grinnellii, S. caruifolia, and S. australis are distinguished morphologically from these other taxa, and other Gilia, by the combination of leaf, trichome, and seed coat characteristics, despite sharing a similar habit with Gilia . This habit, consisting of a typically well-developed basal rosette of leaves subtending an erect, much branched but usually solitary stem bearing reduced leaves, is found elsewhere in Polemoniaceae, notably in Aliciella and Ipomopsis sonorae (tribe Loeselieae) and , in a reduced form, in Linanthus (e.g., L. campanulata of tribe Phlocideae). Hybridization studies indicate Saltugilia is isolated reproductively from Gilia (Grant and Grant 1954;Johnson pers. observ.). Chloroplast and nuclear DNA sequences sampled widely from populations across the geographic range of these species also strongly support the removal of Saltugilia from Gilia (Johnson, 1996;Johnson et al., 1996;Weese and Johnson unpubl.). Saltugilia is sister to a monophyletic group encompassing the remainder of Gilieae. The relationship between morphological and molecular variation at the population level among species of Saltugilia is being explored (Weese and Johnson unpubl.). A. Johnson Aliso 17: 84, 1998.
Tribe Loeselieae, although diverse morphologically, has consistent support for monophyly from comparative DNA studies (Johnson et Porter 1996;Porter and Johnson unpubl.). Several notable evolutionary trends appear to be repeated in Loeselieae, including perennial habit, woody species, bilateral corollas, and aneuploid reduction (in Aliciella, Giliastrum, and the Eriastrum, Ipomopsis, Langlois ia, Loeseliastrum, Microgilia group). AUCIELLA Brand, Helios 22: 78, 1905. Taprooted perennials, biennials, or annuals, mostly glandular pubescent. Leaves alternate, entire, or onceor twice-pinnatifid, leaf tips cuspidate, mucronate or aristate, often forming a basal rosette, cauline leaves reduced either zt gradually or abruptly in size, but ultimately diminished. Inflorescence cymose, open. Calyx composed of herbaceous costae and hyaline intercostal regions, glandular. Corolla salverform to funnelform, concolored, bicolored or tricolored, glandular or glabrous externally, glabrous internally, corolla veins often anastomosing at the base of the lobes and rarely also in the lobe, :::!: radially symmetric. Stamens equally or unequally inserted in the distal corolla tube or in the sinuses of the corolla lobes, the filaments unequal or equal in length. Pollen zonocolporate and either striate-reticulate or reticulate, or zonoporate and pertectate, blue, yellow, or cream. Ovary glabrous, seeds not (or only slightly) becoming mucilaginous when wetted. N = 8, 9,16,17,18,25. Twenty-one species. Type: Aliciella triodon Brand. AUCIELLA CESPITOSA (A. Gray) J. M. Porter. Aliso 17: 34, 1998 Formerly included in Cilia (Grant 1959(Grant , 1998b, Aliciella is more closely related to Loeselia and Ipomopsis than to Cilia (Johnson et al. 1996;Porter 1996), based on DNA sequence data. Morphologically, too, Aliciella differs from Cilia in chromosome number (most, but not all, species of Aliciella are N = 8) and seed morphology (seeds do not produce copious mucilage; large seeded species tend to have an irregular wing). However, in spite of its small size, Aliciella is diverse morphologically and complicated by hybridization and polyploidy. Some species (e.g., A. leptomeria and A. lottiae) represent complexes with multiple polyploid origins (Tommerup and Porter 1996). The genus is also varied in pollination mechanisms and breeding systems (including the only verified instance of heterostyly in Polemoniaceae; Cochrane and Day 1994 Plants stout annuals or perennials, 5-60 e rn tall , glabrescent, glandular pubescent, or viscid throughout. Leaves narrowly linear, entire or pinnatifid, the lobes narrowly linear, axils sometimes with white hairs; cauline leaves reduced in size, ultimately bract-like. Rowers cymose, solitary or in pairs, pedicels 0.5-3.0 cm long. Calyx campanulate, 5-lobed or 5-c1eft; the rube with a scarious region alternating with a green co sta (lobe), glandular puberulent. Corolla rotate to campanulate , white, blue, violet, or deep pink, tube shorter than the calyx, lobes obovate to ovate, api culate or denticulate. Stamens equally inserted low in the tube and included, glabrous; pollen zonocolporate with a striate exine. Ovary glabrous or glandular apically, fruit ovate capsule, longer than calyx; seeds tan to brown, becoming mucilaginous when wetted. Two species. Type: Bryantiella palmeri (S. Wats.) J. M Although traditionally included within Cilia, Bryantiella differs from the former in the lack of a persistent basal rosette of leav es and the perennial life history (some times persist ing only a single year, appearing annual; Gibson 1967 ;Wiggins 1980). Ecologically, Bryantiella also differs from Cilia, being adapted to the driest de serts of both North America (San Felipe Desert) and South Am erica (Atacama Desert). Phylogenetic inferences from the gene phylogenies of nr ITS and cp tmL-F noncoding regions (Porter et al. unpubl.) unambiguously place the relationships of Bryantiella (B. palmeri) within Tribe Loe selieae, near Ipomopsis, and aw ay from Cilia s.s.
The two species were included in Cilia section Giliastrum by Grant ( 1959, but not I998b). Unlike Gilia strum with yellow poll en , B ryantiella has white to cream-colored pollen. Pollen exine morphology of B ryantiella also differs from Giliastrum (see above) in that both spec ies have striate exines with narrow lirae .
The generic name honors Su sanna Bixby Bryant, founder of Rancho Santa An a Botanic Garden (RSABG) and stro ng advocate for systematic research of California plants. Through RSABG, Su sanna Bixby Bryant has provided both infrastructure and financi al support that has contributed immensely to our knowled ge of Polemoniaceae.
Dayia scabra apparently is very local and restricted to several washes just north of Santa Rosalia, on the eastern coast of Baja California Sur, Mexico. Most of the se areas are now under min e tail ing s and the continued existence of this species in the wild is in question. This species has been overlooked as a result of taxonomic confusion. Although the historical collections of D. scabra are representative and quite distinct, they have been misidentified inadvertently as Linanthu s nuttallii (Grant 1959) or Gil iastrum rigidulum (as Cilia rigidula, Wiggins 1980), or omitted (D ay 1964) from treatments. Because this species is known so poorly, we provide a full description of D. scabra.
Ere ct herbaceous perennials with woody base or more freque ntly sub-shrubs, 15-65 cm high and 13-44 em wide, much branched , the young part s glandular-pubescent with 3-5(-7) celled trichomes mostly less than I mm long, each tipped with a unicellular or multicellular yellowish gland. Woody base to 11-22 mm thick, the bark light to dark tan or gray, spliting into a network of narrow verti cal stri ps . Branches ascending, branching pattern axillary along the primary axis, ultimately sympodial in the inflorescence, the branches flowering terminally, 9-40 e m long and 1.0-2.0 mm thick the first year, tan to somewhat anthocyanic, green terminally, subterete ; internodes 1.0-22.0 nun long, generally shorter than the leaves. Primary leaves alternate or rarely sub-opposite, pinnatifid to nearly palmate, 10.0-39.0 nun long, gradually reduced in size in the inflorescence, rachis 0 .6-1 .0 nun wide, 3-6 pairs of opposite to sub-opposite lobes, 9.0-35 .0 mm long, 0.4-0.9 nun wide, often two pairs of lobes are located at the base of the mucronate tipped leaf, sparsely glandular. Inflorescence densely glandular puberulent, composed of reduced (1-) 2-f1owered cymes, forming a thyrsoid inflorescence. Pedicels erect to spreading, pedicel of terminal flower 2-7 nun long, 0.2-0.4 nun thick, pedicel of lateral flower 5-14 nun long, 0.2-0.4 nun thick. Calyx 5.5-7.5 nun long, 3.0-4.0 nun in diameter at the sinuses, cylindric to campanulate, tapering to a rounded base, glandular externally, less so internally, calyx tube 3.0-4.3 nun long, segments (lobes) equal, erect, 1.2-3.2 nun long, attenuate-acute, with a weak mucro, green costae with three primary veins within, scarious margined except near the apex; sinuses v-shaped, the scarious intervals wider to about equal the chlorophyllous costae, generally not rupturing in fruit. Corolla funnelform to salverform, 13.0-20.0 nun long, glabrous externally, pale blue, with a pale or white center, the orifice often with purple streaks, tube subequal or slightly longer than lobes, straight, 6.4-10.0 nun long, ca . 2.5-3.5 nun in diameter at 1,.2 length, slightly flaring at the orifice, 3.0-5.0 nun wide; lobes 6.0-9.5 nun long, 4.0-6.0 nun wide at 1/2 length, oval to oblong or nearly orbicular, entire to emarginate, muriculopapillose within, the lobes convolute in bud , in anthesis spreading, with (15-)17-22(-24) close-spaced parallel veins per lobe, the veins not connected in the lobes. Filaments glabrous above, glandular pappillose at the base, unequal in length, the shortest 3.0-4.0 nun long, the longest 6.0-7.0 nun long, sub-equally attached in the lower tube, the filaments superficially to distinctly adnate to the corolla tube, flowers protandrous; anthers 1.5-2.0 nun long, 0.2-0.6 nun wide, linear to linear-ovate, erect to versatile, mostly simultaneously dehiscing as the corolla lobes open, along the theca from the terminal point and downward, two to three anthers slightly exserted from the tube, the remaining anthers included. Pollen grains suboblate to spheroidal; apertures zonate, 5-7 colporate; exine striate, the lirae radiating from the apertures like lines of force in a magnetic field. Nectary disk green, ca, 2 nun wide shallowly cupped, the margin regularly undulate to form erect lobules opposite the calyx segments and spreading ones opposite the corolla lobes. Ovary three celled, 2.0-3.0 nun long, ca. 1.0 nun wide at the base, glandular at the apex; style 6.0-8.0 nun long, subequal to longer than the longest anthers; stigma lobes linear, acute, 0.8-2.5 nun long, spreading when receptive; ovules anatropous, unitegmic, axial placentation, 24-48 per cell. Fruit a capsule, obovoid, tan to golden brown, oft en suffused with purple, 3.0-5.5 mm long, 1.5-3.5 nun in diameter, apex acute, loculicidally dehiscing, valves slightly recurving to erect, fruit shorter than the calyx. Seeds 12-46 per cell , minute, 0.8-1.5 nun long, 0.4-0.8 mm wide, ovoid, nearly round in cross-section, golden to pale tan, the outer testa with hygroscopic mucilage cells, producing copious fibrils when wetted. Embryo achlorophyllous, surrounded by a more or less thin layer of endosperm, the cotyledons ovate, equal to or slightly longer than the radical. Chromosomes: N = 9. Erect herbaceous perennials with woody base or more frequently sub-shrubs, (18)30-60 em high and 20-50 cm wide, much branched, the young parts glandular-pubescent with 2-8( I 0) celled trichomes mostly less than 0.10-0.33 nun long, each tipped with a unicellular or multicellular yellowish gland. Woody base to 8.5 nun thick, the bark light to dark tan or gray, spliting into a network of narrow vertical strips. Branches ascending, branching pattern axillary along the primary axis, ultimately sympodial in the inflorescence, the branches flowering terminally, 8-30 cm long 1-4 nun thick the first year, tan to strongly anthocyanic becoming gray, chlorophylous terminally, subterete; internodes 0 .5-55.3 nun long, mostly 5.0-12.0 nun, generally shorter than the leaves. Primary leaves alternate or rarely sub-opposite, pinnatifid to nearly palmate, 8.0-32.6 mm long, gradually reduced in size in the inflorescence, rachis 0.6--1.3 nun wide, (I )2-4 pairs of opposite to sub-opposite lobes, 5.0-12.6 nun long, 0.6--1.1 nun wide, often two sets of lobes are located at the base of the mucronate tipped leaf, sparsely glandular. Inflorescence densely glandular puberulent, composed of reduced (1-) 2-flowered cymes, forming a thyrsoid inflorescence. Pedicels erect to spreading, pedicel of terminal flower 1.0-3.2 nun long, 0 .35-0.60 nun thick, pedicel of lateral flower 5.0-10.5 nun long, 0.30-0.60 nun thick  nun in diameter at the sinuses, cylindric, tapering to a rounded base, densely glandular externally, less so internally, tubular for 3.3-4.8 nun (the lower 213 to V-i), segments (lobes) equal, erect or somewhat out-curved, 1.2-3.0 nun long, triangular lanceolate, pungent-acuminate, aristate, with three primary veins within, scarious margined except near the apex; sinuses v-shaped, the scarious intervals about equal to the chlorophyllous costae, generally not rupturing in fruit. Corolla funnelform to salverform, 14.0-22 .0 nun long, glabrous both externally and internally, pale to deep blue, with a distinct yellow center, tube shorter than lobes, straight, 6.0-9.0 nun long, ca. 3.0 nun in diameter at 1;2 length, slightly flaring at the orifice, 4.0-5.0 nun wide; lobes 8.5-12.0 nun long, 6.0-9.5 nun wide at Ih length, oval to oblong or nearly orbicular, entire to emarginate, muriculopapillose within, the lobes convolute in bud, in anthesis spreading, with (15-)17-22(-24) close-spaced parallel veins per lobe, the veins not connected in the lobes. Filaments gl abrous, 13.5-17.6 nun long, sub-equally attached in the upper tube, the filaments superficially to distinctly ad nate to the corolla tube, filaments declinate, flowers protandrous; anthers 3.5-4.5 nun long, 0.6-0.8 nun wide, linear to linear-ovate, erect to versatile, mostly simultaneously dehiscing as the corolla lobes open, along the theca from the terminal point and downward, well exserted from the corolla tube but slightly shorter than the lobes. Pollen grains suboblate to spheroidal; apertures zonate, 5-7 colporate ; exine striate, the lirae radi ating from the apertures like lines of force in a magnetic field . Nectary disk green, ca , 2 mm wide shallowly cupped, the margin regularly undulate to form erect lobules opposite the calyx segments and spreading ones opposite the corolla lobes. Ovary three celled, 3.0-4.0 nun long, ca.I.O nun wide at the base, glabrous; style 10.0-15.0 nun long, subequal to longer than the anthers; stigma lobes linear, acute, 1.5-2.0 mm long, spreading when receptive; ovules anatropous, unitegmic, axial placentation, 24-48 per cell. Fruit a capsule, obovate, tan to golden brown, often suffused with purple, 6.0-9.0 nun long , 2.5-3.5 nun in diameter, apex acute, loculicidally dehiscing, valves slightly recurving to erect, fruit shorter than the calyx.
Seeds 12-46 per cell, minute, ca. 1.2 nun long, 0.6 mm wide, ovoidal, nearly round in cross-section, golden to pale tan, possessing no vestiges of a wing, the outer test a with hygroscopic mucilage cells, producing copious fibrils when wetted. Embryo achlorophyllous, surrounded by a more or less thin layer of endosperm, the cotyledons ovate, equal to or slightly longer than the radical. Chromosomes: N = 9.
Dayia grantii apparently is endemic to a very small region of the Vizcaino desert, along the western coast of northern Baja California Sur. It is known only from the type locality, at the western foot of Cerro Mesa (a.k.a. Cerro Prieto) between San Hipolito and Punta Prieta, 24 km south of Ciudad Bahia Ascuncion. This species occurs along a wash and on adj acent alluvial slopes, associated with Eneelia palmeri, Fouqueria dugettii, Jatropha einera, Euphorbia miser, Bursera hindsiana, and Krameria sp., at approximately 50 ft. Dayia is similar in general appearance to both Ipomopsis and Giliastrum; however, morphological data, as well chromosome number, provide evidence that it is isolated from these genera. The haploid chromosome counts (from dividing pollen mother cells) of D. seabra and D. grantii are x = 9 . This contrasts with all known members of Ipomopsis, which possess a base chromosome number of N = 7 (Grant 1959). Pollen of both D. seabra and D. grantii is blue, with a thick striate-re ticulate exine; seeds are minute; and a well developed corolla tube (6--9 nun long; longer than the calyx) is present. This differs from all members of Giliastrum, that have yellow, pertectate pollen grains, larger seeds, and a very short corolla tube (less than 5 nun long; shorter than the calyx).
Given the relationship between Dayia and Loeselia inferred by ITS (Porter 1996) and matK (Johnson et al. 1996) sequence data, it is important to note the gre at morphological differences between Da yia and Loeselia that parallels the large number of nucleotide changes that distinguish these genera. A well-characterized genus of Polemoniaceae, members of Loeselia ALISO possess simple, dentate leaves and large, wingless or winged seeds. In addition, most members of Loeselia have bilateral1y symmetric flowers (except L. mexicana) and a shrubby to subshrubby habit. By contrast, Dayia possess pinnatifid leaves and minute, wingless seeds and radial1y symmetric corollas. Further, in molecular phylogenetic analyses of nrITS and cp trnL-F, including all species of tribe Loeselieae, Dayia does not share immediate and exclusive common ancestry with Loeselia, nor with Giliastrum, nor with lpomopsis (Porter et a!. unpub!.).
The DNA studies of Johnson et al. (1996) and Porter (1996) included "Gilia scabra" in their analyses. Following the nomenclatural changes provided here, the DNA sample used in these studies actually represents Dayia grantii.
The generic name honors Alva G. Day, who has contributed much to the understanding of morphological diversity in Polemoniaceae. The specific epithet of Dayia grantii honors Verne E . Grant, who has dedicated much of his research to developing an understanding of the diversification process in the genus Gilia and Polemoniaceae as a whole.
Most authors (Craig 1934;Mason 1945;Harrison 1972) have disagreed with  inclusion of Eriastrum within Navarretia. For example Grant (1959) suggests a close relationship between Eriastrum, Ipomopsis, and Langloisia. Both chloroplast and nuclear DNA sequence data (Johnson et al. 1996;Porter 1996) support Grant's hypothesis. Further, the morphological and molecular data provide support that Eriastrum, as circumscribed by Grant (1959) and Mason (1945), is a monophyletic group.
Even though there is little debate regarding generic delimitation of Eriastrum, species and subspecies boundaries are problematic (compare Mason 1945 andPatterson 1993). We fol1ow the species recognized in the two most recent monographic/revisional studies (Mason 1945;Harrison 1972). Thorough comparative studies of this genus are needed.

1917.
Retrospectively, it is remarkable that a genus, as cohesive and well marked as is Giliastrum, has remained entangled with an unrelated group of species (i.e, Gilia, see Grant 1959Turner 1994a). Giliastrum floral morphology is quite different from Gilia. Species of Giliastrum have short tubes, large, rotate lobes, and bright yellow pollen. Giliastrum displays a pattern of corolla vasculature that lacks anastomoses in the corolla lobes, a pattern not documented in Gilia (Day and Moran 1986), as here circumscribed.
Giliastrum exhibit an unusual floral "behavior." Members of this genus have flowers that open and close at specific times of the day . Giliastrum acerosum, G. ludens, and G. rigidulum possess flowers that open in the early afternoon and close at dusk. By contrast, G. insigne and G. purpusii have flowers that open with the early morning sun, but close around noon. This phenomenon is particularly striking at locations where representatives of the two types occur sympatrically. The flowers of G. foetidum open in the early morning and remain open until late afternoon, providing a third, contrasting pattern.
Apparently Giliastrum castillanosii is very local and restricted. The only known collections are from the same region, toward the northern end of Valle de Chaschuil, near the Cordillera de San Buenaventura, in the Andes Mountains of western Catamarca, Argentina. Closely related to Giliastrum foetidum , G. ca stellanosii differs from the former in both vegetative and floral traits. The leaves of G. castellanosii have linear lobes, dissected to the midrib, contrasting with the broadly lobed leaves that are often shallowly lobed in G. foetidum. The flowers of G. castellanosii are pale blue, 15 nun long or less, with obovate, acute lobes. Giliastrum foetidum has pink corollas, usually more than 15 nun long, with ovate, rounded lobes. In addition, pollen of G. castellanosii is zonocolporate, with a rugulose exine, covered with small, but sometimes densely arranged, verrucae. Giliastrumfoetidum (Prov. Mendoza, Cruz del Pararnillo, C. A. O'Doneli 1172. LIL) has zonocolporate pollen with a striate exine, bearing both highly scattered verrucae and spinule-like processes. These are strikingly different exine sculpting patterns. Although the se two Giliastrum species occur at similar elevations, G. castellanosii occurs farther north in Provo Catamarca, whereas G. foetidum is found to the south, in Provo Mendoza and Provo San Juan.
The specific epithet honors A. Castellano, eminent botanist of Argentina, who has contributed much to the knowledge of Cactaceae of Argentina (and Polemoniaceae, through his coJlections).
The segregation of Ipomopsis from GiLia ), ignored or rejected by some authors (Cronquist 1984;Welsh et al. 1993), has been supported overwhelmingly by molecular data (Steele and Vilgalys 1994;Johnson et al. 1996;Porter 1996). Although chromosome numbers of Ipomopsis (N = 7) are consistently different from that of GiLia (N = 9), the morphological traits that distinguish these genera are surprisingly cryptic. Rather than any single feature, trends and combinations of traits distinguish Ipomopsis. Ipomopsis has many perennial species (Gilia species are annual; Aliciella, Giliastrum, Eriastrum and Dayia also have perennial species) and bilateral corollas are common (CiLia species have mostly radial corollas). Some species of Ipomopsis have thyrsoid inflorescences, an architecture never found in Gilia nor most other genera (save a few species of ALicieLLa). These traits, while singly not separating every species of Ipomopsis from Cilia, are helpful if considered together.

Ipomopsis monticola
Known only from the type collection in the Sierra Surutato, Sinaloa, Mexico, this species occurs on steep, moist slopes at 6500 ft elevation, associated with ALnus, Styrax, Mahonia, CLethrea, Pinus, Quercus, and JugLans. Ipomopsis monticoLa is apparently quite uncommon, and has been mistaken for I. aggregata. It differs in having ovate corolla lobes that lack the more attenuate apex observed in I. aggregata. It also differs in having a glandular ovary and filiform leaf lobes, whereas I aggregata has a glabrous ovary and linear (but not filiform) leaf lobes. In some other respects it is similar to I. wendtii, of Coahuila, Mexico (Henrickson 1987). For example, both possess pink corollas, with ovate lobes, and are associated with pine-Oak woodlands in the mountains of Mexico. However, I. monticoLa lacks the enlarged calyx characteristic of I. wendtii, possesses a longer corolla tube , and differs in pollen morphology. Ipomopsis wendtii has oblate spheroidal pollen, 38.3-48.0 I.l. (equatorial), 35.3-40.0 I.l. (polar), 6-7(-8) zonocolporate apertures, with a heterobrochate, reticulate exine, unlike pollen of I. monticoLa (see above). Phylogenetic analyses of chloroplast trnL-F DNA sequences strongly support relationships between I. monticola, I. pinnata, and I. pringLei; however, nuclear sequences (ITS) provides ambiguous resolution concerning I. monticoLa (Porter et al. unpubl.).
Annual herbs, branching from the base, hairy with branched trichomes but not glandular. Leaves alternate, simple, linear to oblanceolate, dentate to oncepinnatifid, lateral lobes with 2-3 bristles, not forming a basal rosette, cauline leaves little reduced in size. Inflorescence mostly terminal and head-like, subtending bracts with bristle-tipped lobes. Calyx composed of herbaceous costae and narrow hyaline intercostal regions, lobes equal in length, bristle tipped. Corolla funnelform, radially symmetric. Stamens equally inserted below the sinuses of the corolla lobes, the filaments equal in length. Pollen zonocolporate and striate to striate-reticulate, blue or white; Ovary glabrous apically, seeds becoming mucilaginous when wetted. N = 7. One species. Generic delimitation of LangLoisia is discussed by Timbrook (1986), wherein he cites compelling morphological and phytochemical evidence for the segregation of LangLoisia and LoeseLiastrum. LangLoisia has branched trichomes (but lacks glands), has a radially symmetric corolla, equally inserted straight filaments, white to blue pollen, reticulate seed coat sculpturing, and possesses three different 6-methoxyflavonols, while lacking the common flavonols, kaempferol and quercetin. In contrast, Loeseliastrum has unbranched trichomes (and also glands), has a bilaterally symmetric corolla, unequally inserted sigmoid filaments, yellow pollen, linear seed coat sculpturing, and possesses the common flavonols, kaempferol and quercetin while lacking the 6-methoxyflavonols. Indeed, morphological evidence for the segregation of Langloisia and Loeseliastrum is overwhelming.
Leaves alternate or opposite, simple, entire, dentate or aristate. Flowers perfect, in bracteate dichasia or congested in head-like inflorescences. Calyx campanulate to tubular, radially or slightly bilaterally symmetric; with herbaceous costae and scarious sinuses, or entirely scarious. Corolla funnelform to salverform, radially to strongly bilaterally symmetric, external surface glandular to glabrous, internal tube glandular pubescent or glabrous. Stamens equally inserted in the mid to lower portion of the corolla tube, filaments pubescent, papillate or glabrous. Apex of ovary glandular puberulent to glabrous. Pollen pantoporate, pertectate exine with fine striations and spinule-like processes. Fruit a loculicidal capsule, ellipsoidal or globose in shape. Seeds several per carpel, lacking a wing or rarely narrowly winged, seed coat producing mucilage when wetted. N = 9. Thirteen species. Type: Loeselia ciliata L.
Nearly all classifications of Polemoniaceae prior to that of  include Loeselia in the same tribe (or "Gruppe") as species here included in tribe Loeselieae (along with others; see Table I) . It was with  treatment that Loeselia was isolated into another tribe, along with Bonplandia; however, both were still included in subfam. Polemonioideae. Grant (1959, 1998a associates Loeselia with the "tropical genera" of subfam. Cobaeoideae. In spite of this taxonomic assignment, Grant (1959; suggests that Loeselia is involved in the origin of Ipomopsis and Giliastrum (hi s "Gilia sect. Giliastrum'"). Although Grant's (1998a, b) general thesis of relationships (subfamilial, tribal, and within "Gilia") is without support from molecular data (morphological data being equivocal), there is evidence for common ancestry between Loeselia, Giliastrum, and Ipomopsis, along with the other genera here included in Loeselieae. Nomenclature and species delimitation in Loeselia remains in an unsatisfactory state and revisional work is needed. At least four named taxa have had their types destroyed and neither lectotypes nor neotypes have been identified (Turner 1994b Acad. Arts 16: 106, 1880. LOESELIASTRUM MATfHEWSIl (A. Gray) Timbrook, Madroiio 33: 171, 1986. LOESELIASTRUM SCHOTfIl (Torr.) Timbrook, Madroiio 33: 172, 1986 The segregation of Loeseliastrum from Langloisia has been discussed by Timbrook (1986) and is supported by morphological evidence; Loeseliastrum is related to Langloisia and Eriastrum more closely than to Gilia, as suggested by Timbrook. We include Loeseliastrum depressum, formerly treated within lpomopsis , in this small genus. Although morphological data is somewhat equivocal, L. depressum shares bilaterally symmetric flowers, glandular trichomes (in addition to long nonglandular trichomes), subequally inserted filaments that are unequal in length, and corolla vasculature with veins connected only well above the base of the corolla lobe, with Loeseliastrum. The corolla vasculature trait has not been observed in lpomopsis (Day and Moran 1986), and in Loeseliastrum occurs only in L. matthewsii. This is noteworthy because chloroplast trnL-F DNA sequences provide evidence supporting the hypothesis that L. depressum and L. matthewsii share immediate common ancestry (Porter et al. in prep.). It must be recognized, however, that nuclear ITS sequences support (albeit weakly) L. depressum as sister to Loeseliastrum, Langloisia, and Eriastrum (but not to lpomopsis).
Tribe Phlocideae incorporates Grant's (1998a) Tribe Leptodactyloneae, and three members of Grant's Tribe Polemonieae. All of these genera display a tendency toward possessing opposite leaves (Grant 1959); however, not all species in this tribe have opposite leaves (Mason 1941). Similarly, most species possess salverform corollas, but other forms are also found . Nuclear and chloroplast gene sequences also support the monophyly of Phlocideae (Johnson et Porter 1996).
Annual herbs, stems naked or nearly so, but bearing persistent cotyledons. Leaves reduced to a whorl of entire, basally connate involucral bracts, subtending a more or less capitate inflorescence. Inflorescences cymose, more or less congested or reduced to a single flower. Calyx almost entirely hyaline, save the herbaceous lobes. Corolla salverform, the tube narrow, lobes somewhat large (to 6.0 rnrn long) to very short, spreading, acute to rounded apically. Stamens equally inserted on the distal corolla tube; pollen pantoporate, with reticulate exine. Seeds 3 or more per locule, producing copious mucilage when wetted. N = 6. Two Affinities of Gymnosteris lie with PhLox and Microsteris (Johnson et al. 1996) rather than CoLLomia (W herr y 1944; Grant 1959Grant , 1998a. Gymnosteris pollen is nearly identical to that of PhLox (Stuchlik 1967a ; Taylor and Levin 1975), and seeds produce copious mucilage when wetted that appears to be truly mucilaginous in origin (as does Microsteris), rather than the expansion of obvious spiricles as observed in CoLLomia (Schnopf and Deichgraber 1983). Also, the congested inflorescence of Gymnoster is, although similar to those found in Collomia, is equally similar to those found in PhLox and elsewhere in PhJocideae. Chloroplast and nuclear DNA sequences strongly support this view (Johnson 1996; Johnson and Soltis 1998).
Low to erect and slender annuals, or perennials, sometimes woody at the base. Leaves opposite, palmately lobed, rarely entire (L. floribundu s subsp. haL-Lii), with 3-9 lobes, nearly glabrous to scabrous, tomentose or glandular. Inflorescence either open or head-like, flowers pedicellate, rarely sessile (L. Lemmonii). Corolla salverform to funnelform, often with a ring of hairs within the tube. Stamens equal in length, attached at the same level in the corolla, pollen yellow, pantoporate, rarely pantocolporate, with pilate, reticulate exine. Seeds becoming mucilaginous when wetted, or remaining unchanged. N = 9, 18. Thirty-one species of western North America and Chile. Type: Leptosiphon androsaceus Benth. Synonymy includes Siphonella A. Heller, MuhLenbergia 8: 57, 1912;and Linanthastrum Ewan, Jour. Wash. A cad. Sci. 32: 139, 1942. LEPTOSI PHON ACICULARIS (Greene ) Jeps .• School jI. Pacif. Coa st, p.77. 1902. LEPTOSIPHON Table 1), Some morphological support for the distinction between Leptosiphon and Linanthus was indicated by Wherry (1961: 10), but he did not make nomenclatural changes: The aim of this note has been to point out that, if calyx characters are given adequate consideration, the so-called genus Linanthus is polyphyletic. The small group of its members which includes the generitype, and so here is termed colloquially the eulinanthus section, seems to have arisen from the Leptodactylon complex. Most of them have on the other hand evolved from perennials which may be assigned to the genus Linanthastrum, and so could be classed as generically distinct from Linanthus Bentham (1833), sens. strict.
Although all of the details of Wherry's note have not found support, his hypotheses of relationship between Linanthus sect. Linanthus and Leptodactylon, and between Linanthastrum (= Leptosiphon nuttallii and relatives) and much of what previously has been referred to as "Linanthus" is endorsed strongly by molecular phylogenetic analyses (Johnson et al. 1996;Porter 1996;Bell et al. 1999;Bell and Patterson 2000). Leptosiphon, as here circumscribed, includes taxa formerly included in Linanthus sections Siphonella, Leptosiphon, Pacificus, and Dactylophyllum. LINANTHUS Benth., Edward's Bot. Reg. 19: sub. t. 1622, 1833. Linanthus, while morphologically diverse, is evidently monophyletic as here circumscribed (Johnson et Porter 1996;Bell et al. 1999; Bell and Patterson in press). Species formerly treated as the genus Leptodactylon are , without question, within this monophyletic group. This is supported not only by molecular phylogenetic studies, but also morphology ; see also Fig. I a & c, in Patterson 1977). For example, with the exception of Linanthus californicus, all of the species formerly of Leptodactylon possess vespertine flowers, as is found in Linan thus section Linanthus (Linanthus arenicola, L. bigelovii, L. dichotomous, L. jonesii, and L. viscainensis).
We also include section Dianthoides, which, along with L. maculatus, L. campanulatus, L. inyoensis, and L. filiformis, are a grade leading to the remainder of Linanthus. While it may seem desirable to remove some of these to another genus (e. g., Tintinabulum and Maculigilia, Grant 1998b), such an attempt, without additional generic circumscription ("splitting"), gives taxonomic status to a paraphyletic group. While L. maculatus, L. campanulatus, L. inyoensis, and L. filiformis all possess alternate leaves (at least above the lowermost nodes), seemingly discordant with Linanthus, it is important to also observe that L. arenicola, L. californicum, L. pungens, and L. viscainensis also possess alternate leaves. Moreover, the floral morphology of L. maculatus, L. campanulatus, L. inyoensis, and L. filiformis is strikingly similar to members of sect Dianthoides in having a calyx with a short tube, and rotate to campanulate corolla with red marks near the orifice of the corolla tube.
It is noteworthy that only a decade has passed since Linanthus maculatus was returned to Cilia (Patterson 1989). This decision was based on a comparative morphological study that demonstrated great morphological similarity between L. maculatus and three species of "Cilia" (L. campanulatus, L. inyoensis, and L. filiformis). Recent comparative DNA studies of nrlTS (Porter 1996;Bell et al. 1999), and cp tmL-F (Johnson and Porter, unpubl.) support Patterson's (1989) assertions regarding affinities among these species; however, all four species share most recent common ancestry with Linanthus, rather than Cilia.
Close examination of Porter ( 1996) reveals that Linanthus inyoensis was inferred (albeit weakly) to share common ancestry with Phlox, Mierosteris and Leptosiphon, in some analyses of nr ITS sequences (and also cp trnl .-F, Johnson and Porter unpubl.). While no additional molecular data has been surveyed that suggests these independent chloroplast and nuclear genes are revealing erroneous inferences of relationship for this species (due to introgression, lineage sorting, or systematic error), there is an overwhelming similarity between L. inyoensis and L. campanulatus (the two were considered conspecific by some authors: e.g. , Jepson 1957). They differ primarily in flower size. Rather than recognize a new, monotypic genus based upon flower size and weak molecular evidence, we include L. inyoensis within Linanthus and recommend further study of this species.
Annual herbs, stems erect and branched. Leaves opposite below, alternate above, simple and entire. Inflorescences cymose, reduced to a pair or single flowers. Calyx tube with hyaline intervals alternating with the herbaceous costae, the lobes with hyaline margins. Corolla salverforrn, the tube narrow, lobes short, spreading, somewhat notched apically. Stamens unequally inserted on the corolla tube. Pollen pantoporate, with reticulate exine. Seeds one per locule, producing copious mucilage when wetted. N = 7. One species. Microsterls is generally considered to include a single, polymorphic species with perhaps two widespread races. These races differ largely in architecture and flower size: subsp. gracilis is unbranched with larger flowers, subsp. humilior is branched with small flowers (see Cronquist 1984). In actuality, morphological variation across western North America is much more complicated and appears to involve phenotypically plastic traits. Moreover, Microsteris occurs in an incredibly diverse array of habitats, ranging from alpine slopes to Sonoran Desert washes. A nearly identical degree of variation is observed in South America, both in terms of morphology and habitat. Microsteris may be more diverse than previously indicated. Directed research is required to determine the limits of geographically partitioned genetic diversity (subspecies) and the role of phenotypic plasticity in the diversification of Microsteris.
PHLOX L., Gen. pl. ed. 1, p.52, 1737. Annual or perennial herbs, stems erect to decumbent, sometimes cushion-like or mat-forming. Leaves opposite, the uppermost sometimes alternate, simple, entire, often connate at the base. Inflorescences cymose, sometimes reduced to a single flower. Calyx tube with hyaline intervals alternating with the herbaceous costae, the lobes with hyaline margins. Corolla salverform, the tube narrow, lobes large, spreading, acute to notched apically. Stamens unequally inserted on the corolla tube; pollen pantoporate, with reticulate exine. Seeds 1(-4) per locule, not producing copious mucilage when wetted. N = 7. About 69 species. Type: Phlox glaberrima L.  Revis . W. N. Amer. Phlox., p .24 & 25 . 1899. Phlox is characterized by simple, entire leaves, salverform corolla, stamens unequally inserted on the corolla tube, and pantoporate pollen with reticulate exine, However, all of these traits are also found in one or another of the closely related genera, Leptosiphon, Linanthus, and Microsteris. Thus, in spite of the morphological cohesiveness, there are apparently few unambiguous synapomorphies for Phlox. Although some species are distinctive, intergradation is common in many species groups ) and hybridization has been proposed to playa major role in the origin of species in some groups (e.g., Levin and Smith 1965;Levin 1966). The hybrid origin of some Phlox species is consistent with recent ITS sequence analyses (Ferguson et al. 2000).
There has been little controversy regarding generic limits, save the debate concerning the inclusion of Microsteris within Phlox (Mason 1941). Recent molecular cladistic analyses with improved sampling in Phlox (Ferguson et al. 2000), taken together with previous studies (Johnson et al. 1996;Porter 1996), provide evidence that Microsteris is the sister group to Phlox, rather than sharing close relationship to other annual Phlox species (Mason 1941). Thus, segregation of Microsteris from Phlox rests on the relative importance placed on distinguishing characters, such as seed coat mucilage and flower size, in generic delimitation (see also Grant 1998a). POLEMONIEAE Dumort. Anal. Jam. pl., p.25 , 1829 . Perennial or annual (Polemonium micranthum) herbs. Leaves pinnately compound. Calyx accrescent, regular; corolla regular, rotate, campanulate or funnelform. Stamens usually inserted equally. Seeds dark, angular, and shiny. Type: Polemonium L.
Phylogenetic delineation.-the most incorporative monophyletic group of species that includes Polemonium ca eruleum and P. micranthum. and shares a more recent common ancestor with Polemonium viscosum than with Aliciella triodon, Gilia laciniata, Gymnosteris nudicaulis, Loeselia ciliata, Phlox glaberrima, or Saltugilia grinnellii. Membership.-Polemonium. As a taxon, the name Polemonieae is redundant with Polemonium; recognition of this tribal rank is maintained for uniformity in treatment in this classification. POLEMONIUM L., Sp. pl. ed. J, v.l, p.162, 1753 . Perennial (one annual) herbs, stems generally erect, often glandular. Leaves alternate, pinnately compound, or very deeply pinnatifid, the terminal leaflet confluent with the rachis. Inflorescences variously cymose. Calyx tube mostly herbaceous, rarely with a very narrow or shallow hyaline interval alternating with the broad herbaceous costae, the lobes lacking hyaline margins. Corolla campanulate to funnelform, rarely nearly rotate. Stamens equally inserted on the mid to lower corolla tube, pollen pantoporate, with striate or reticulate exine. Seeds 1-10 per locule, producing copious mucilage when wetted. N = 9. About 28 species. Type: Polemonium caeruleum L.
gen er ally herbaceous, lackin g a per sistent, woody, above-ground stem, woody ca ude x may be present 26 30 25 St am en filaments unequ ally affix ed in the co ro lla tube . St amen filaments equall y affixe d in the corolla tub e ers so litary or in pairs, radiall y sym metric; filaments atta ch ed at the mid-tube "Bryantiella Plants mor e or less strict and th yrsoid, branching from the base. sometimes so me what o pe n, bearing eg land ular, w hite trichomes and so me times glandular trichomes (not viscid ); co ro lla bilat erall y sy mmetr ic (rare ly radi al ): filam ent s attac hed above the mid-tube . . . . . . . . . . . . . . . . . . . . . . .. ' Ipo mopsis Cal yx wh olly green (herbaceou s) in flower. or ca rinat e with a narrow intercostal membrane that termin ates in a spo ut-like project ion at the apex of the si nus 22 Ca lyx wi th a well-defined hyal ine inte rcos ta l membra ne (some times narrow); ca lyx sinuses not pleated or ex pa nde d (tho ug h the entire cal yx may be accrescent ), and not terminat ing in a spo ut-like proje cti on 23 Caly x who lly green, accresent , si nus not pleated in fruit ; coroll a rotate campanul at e ; . · ' Polemo nium (P. micranthum ) Ca lyx w ith a narrow intercostal membrane that termin ates in a spo ut-like project ion at the ape x of the si nus ; s inus of cal yx pleated to expa nde d in fru it: coroll a funne lform to sa lve rfo rm, Co llomi a Leaves primarily opposit e o n low er ste m, generall y cauline , leaves often alternate in inflor escence or o n di stal portions of branches 24 Leaves alterna te , cauline andIor w ith a rosette (lower-most nod es sometimes subopposite to opposite) description differs slightly, and the authors take full responsibility for any errors. Kenneth D. Heil aided in collections and field studies, including collections of Dayia grantii, and Terri Weese aided in laboratory observations. This research was carried out under NSF research grant, DEB-9509121 to JMP and support from North Carolina State University and Brigham Young University to LAJ.