Ex uno, multis: taxonomic revision in Navarretia divaricata (Polemoniaceae) and the recognition of four additional cryptic or near-cryptic species

Abstract Navarretia divaricata, endemic to western North America and most recently considered a single species with two subspecies, was re-examined in light of field work, DNA sequences, comparative morphology, and a review of herbarium specimens including types. From these studies, we lectotypify the material on which N. divaricata is based, elevate N. divaricata subsp. vividior, which is an allotetraploid, to species rank (as N. vividior comb. et stat. nov.), and recognize three additional species: N. modocensis sp. nov., N. aeroides sp. nov., and N. torreyella sp. nov. Navarretia modocensis, the diploid paternal progenitor of N. vividior, is morphologically cryptic with respect to its allotetraploid offspring and difficult to distinguish on herbarium sheets. Navarretia aeroides, the diploid maternal progenitor of N. vividior, is nearly cryptic, but more easily distinguished from both N. modocensis and N. vividior by its smaller, more glandular inflorescences. Navarretia torreyella is readily distinguished from all of these species, but has been generally mistaken for N. divaricata subsp. vividior given its colored corolla tube and rare co-occurrence with the other vividior-like species. Conservation assessments, an identification key, and table of comparative morphological features are provided for each species, emended descriptions for N. divaricata and N. vividior, and a discussion of the syntypes for Gilia divaricata Torr. ex A.Gray.


Introduction
Multiple criteria applied to assess the presence of homogenizing gene flow under the framework of the unified species concept (de Queiroz 2007) has provided strong evidence for the existence of previously unrecognized lineages in Polemoniaceae, including in Navarretia (Johnson and Cairns-Heath 2010. Overlooked in the past because they resemble other species, re-examination of morphological variation in conjunction with molecular data has enabled these cryptic or nearly cryptic lineages to be delineated and lifted from obscurity as recognized taxa. In some cases, molecular data provided the first clues as to the uniqueness of these taxa (e.g., Johnson andCairns-Heath 2010, Johnson et al. 2012), whereas in other cases, morphological variation was recognized first, and the biological significance of this variation confirmed with molecular data (Johnson et al. 2013(Johnson et al. , 2016. Stimulated by a need for nomenclatural housekeeping along with observations of variation in both morphological and molecular data, we here delimit several near-cryptic species from what previously has been considered a single species, Navarretia divaricata Greene. Within Navarretia Ruiz & Pav., N. divaricata is one of the more widely distributed species with populations extending from central California (and adjacent Nevada) to Idaho and British Columbia. It is also one of the smallest-flowered Navarretia. The taxonomic concept associated with N. divaricata has been fairly stable since Greene's treatment of the species in this genus (Greene 1887), and Jepson's subsequent partitioning of "coarser" specimens with more glandular-puberulent bracts and calyces, and blue or blue-lobed corollas into variety vividior Jeps. & V.L. Bailey (in Jepson 1943; currently subsp. vividior (Jeps. & V.L.Bailey) H. Mason). To preserve this stability, clarification of the type for N. divaricata Greene and lectotypification of the basionym upon which Greene's replacement name is based is necessary. Furthermore, stemming from observations that N. divaricata subsp. divaricata does not form a monophyletic group with N. divaricata subsp. vividior in comparative DNA sequence analyses (Johnson et al. 2016), that the two taxa co-occur without signs of hybridization, and that material assigned to N. divaricata subsp. vividior is polymorphic, we initiated fieldwork, a comprehensive review of herbarium specimens, and investigation of morphological and molecular variation across the geographic ranges of these taxa. Results indicate that N. divaricata subsp. vividior merits recognition at the species level, and that four evolutionarily unique lineages exist in the material heretofore generally referred to subsp. vividior. The taxonomy of this group is revised accordingly.

Historical background
Among plants provided to Asa Gray by John Torrey, two Navarretia collections are relevant to the present study. The first, labeled simply "Navarretia" by Torrey, was collected during his 1865 trip to California. The second, collected by Mr. Shelton, was labeled by Torrey, "Navarretia divaricata, n. sp." Gray considered these specimens con-specific, but having a confused view of generic relationships in Polemoniaceae (Mason 1945), he published this new species, attributed to Torrey, in the genus Gilia (i.e., Gilia divaricata Torr. ex A. Gray) where Gilia divaricata Nutt. (≡ Allophyllum divaricatum (Nutt.) A.D.Grant & V.E.Grant) already existed. Consequently, Gray's action created an illegitimate later homonym that was replaced by Greene (1887). In the protologue, Gray (1870) states, "California, along the foot hills of the Sierra Nevada, coll. Shelton, Rattan, Bolander, Torrey, Mrs. Davis, C. Lee," providing four additional syntypes beyond the two that can be traced to Torrey. The last of these, a specimen attributed to C. Lee, cannot be located and its identity remains unknown to us. The remaining five syntypes represent four distinct taxa. Consequently, the choice of lectotype is critical for maintaining the long-used taxonomic concept for N. divaricata. Gray (1876Gray ( , 1878 continued to recognize Gilia divaricata Torr., using descriptions similar to his original. These descriptions accentuate characteristics of the syntype Rattan s.n. (NY!; probable duplicates at UC × 2 (scan!), US (scan!), KEW (scan!)), which has larger flowers and a more densely pubescent head compared to the other syntypes, and is also a mixed collection from two gatherings. These specimens are today recognized as N. prolifera Greene, which has the narrowest distribution of the taxa represented by Gray's syntypes.
Emphasizing the distinctiveness of the calyx in Navarretia, Greene (1887) began the work of narrowing Gray's concept of Gilia and provided a replacement name for Gray's illegitimate homonym by equating N. divaricata Greene with Gilia divaricata Torr. ex Gray. There is no indication that Greene viewed the syntypes Gray worked with, and good reason to believe he did not-for example, he described N. prolifera as a new species immediately preceding N. divaricata, with no reference to any of the Gilia divaricata Torr. ex Gray syntypes. Greene characterized N. divaricata flowers as "minute" and distributed in the Sierra Nevada from Kern County northward into Oregon; he distinguished this taxon from N. prolifera and N. peninsularis Greene in part based on smaller flower size. Beyond the description provided for N. divaricata, Greene's taxonomic concept can be reconstructed by examining the specimens he worked with. Four sheets with five collections exist at NDG (scans!) labeled "Navarretia divaricata Greene", in Greene's hand, with collection dates prior to Greene's publication, and a sixth specimen, also on one of the four sheets, with a preprinted label collected in the year of his publication, labeled "Gilia divaricata Torr." Of these, Greene referenced two in his publication.
M. K. Curran s.n. , from Tehachapi, Kern County, California, was cited as representing the southern extent of the species range, but differs taxonomically from the other specimens and is equivalent to N. peninsularis. Interestingly, this collection represents the northern extent of the latter species rather than the southern extent of N. divaricata. Greene described N. peninsularis based on larger plants from Lower (Baja) California, considerably distant from Tehachapi, so perhaps the Tehachapi plants being more depauperate and proximal to other N. divaricata specimens played a role in his misidentification and efforts to circumscribe taxa with little material on hand. The second specimen, C. F. Sonne s.n. (Donner Lake) , was cited as a putative hybrid between N. divaricata and N. minima Nutt. without explanation. This specimen, along with C. C. Parry s.n., (Yosemite) , D. Cleveland s.n. (Butte County)  , and T. Howell s.n. (near Waldo, Oregon)   all correspond to material congruent with Gray's syntypes Bolander 4908 and Mrs. Davis 49, while a second M. K. Curran s.n. specimen, from Colusa County, California , is congruent with Gray's syntype Torrey 302 discussed further below. Gray's syntypes Bolander 4908 and Mrs. Davis 49 have minute flowers and the species represented by them is the most common and widely distributed geographically of Gray's syntypes, as well as the entity to which the name N. divaricata subsp. divaricata has been applied.
Greene's taxonomic concept influenced nearly all later botanists. For example, Howell (1901) indicated N. divaricata occurs from "Washington to California, in the high mountains," with a description that borrowed more from Greene's description of the species than Gray's. Though referencing neither Gray nor Greene, the publication of Gilia atrata M.E.Jones with its type and two of three paratypes matching Bolander 4908 and Mrs. Davis 49 (and the third a similarly small-flowered entity that would fit within Greene's concept) may stem from Jones ' (1908) recognition that Gray's name was illegitimate while disagreeing with Greene in recognizing Navarretia as distinct from Gilia. Gilia atrata has not appeared in print, other than as a synonym, beyond its original publication.
The syntype Torrey 302 was gathered in Lake County, California, well removed from the Sierra Nevada. Circa 1936, Virginia Bailey began a study of N. divaricata and noticed four specimens from Butte, Lake, and Mendocino counties that were stouter of habit, slightly larger flowered, produced more seeds per locule, and possessed blue corollas or corolla lobes (note attached to Austin 827, UC!). Torrey 302, though not seen by her, fits naturally into the group she was segregating. Jepson (1943) with Bailey, recognized this material as a variety of N. divaricata and chose a specimen also from Lake County as the type. Jepson and Bailey's taxonomic concept for the two varieties of N. divaricata is clear, with all 17 representative specimens listed for N. divaricata var. divaricata morphologically congruent with Bolander 4908 and Mrs. Davis 49, and nine representative specimens listed for var. vividior (paratypes) that all share the larger flowering heads, more robust habit, and larger/more glandular heads as described by Jepson and Bailey. Thus, in recognizing var. vividior, Jepson and Bailey followed a taxonomic concept for the typical variety of N. divaricata that matches the syntypes Bolander 4908 and Mrs. Davis 49, despite not examining these syntypes or making reference to them. Mason (1951) treated these varieties as subspecies, but otherwise did nothing to alter the taxonomic concept for them. The description by Cronquist et al. (1984) of N. divaricata includes features, such as unequal valves and two of the three stigmas almost entirely fused, that exist only in material congruent with var. divaricata as put forth by Jepson and Bailey, and congruent with the syntypes Bolander 4908 and Mrs. Davis 49. The remaining syntype, Mr. Shelton s.n. (NY!), is distinct morphologically from the other syntypes. It has smaller heads and slender branches, like N. divaricata subsp.
divaricata, yet has a purple corolla throat and tube, three equal valves, and three equal stigma lobes that readily separates it from that taxon. While material matching this syntype was available to Jepson and Bailey, none of those collections were included in the list of representative specimens for either var. divaricata or var. vividior. No locality information is provided for Mr. Shelton's collection, though the material that it matches occurs in only a portion of the Sierra Nevada, with known occurrences in close proximity to Sacramento where he lived (Ewan 1981). Mr. Shelton's specimen is of particular interest because it is the only syntype labeled "Navarretia divaricata n. sp." in Torrey's hand. An unpublished annotation attached to the sheet at NY, upon which Mr. Shelton s.n. and three of the remaining five syntypes are mounted, suggests Shelton's collection is the lectotype-attributing the first step to Jepson (1943) and the second step to Cronquist (1984). We, however, disagree with the inferences made by this annotation. Jepson did not specifically state he was indicating the type, but instead restated the type locality published by Gray with his exact words reading, "type loc. 'foothills of the Sierra Nevada,' Shelton (the first named collector)." This falls short of specifying the Shelton collection as the lectotype, particularly since there is no indication on the sheet that the Shelton collection comes from the Sierra Nevada (even though we have determined, based on morphology, that it does). Jepson (1943) elsewhere explicitly indicates his observation of types, such as the isotype for N. peninsularis, but no such indication is provided for N. divaricata. It is unlikely Jepson would have specified as lectotype a specimen that varied from the concept he and Bailey defined through extensive specimen citations. Cronquist et al.'s actions also do not constitute a second step lectotypification. The treatment of N. divaricata in this work states "Shelton s.n., Sierra Nevada, Calif., is the first specimen cited by Gray, and the only one annotated by Torrey as 'n. sp.'; isotype at NY!" Importantly, the citation above is not an act of lectotypification; the first volume of the Intermountain Flora expressly states, "When one of several collections cited in the protologue is obviously the primary basis for a name, we have given it as the type, without further comment. Sometimes the typification is less clear and a more cautious statement is necessary. Our citation is not to be taken as the formal selection of a lectotype, except when specifically so indicated" (Cronquist et al. 1972: 8).
To summarize, for over 125 years, the name Navarretia divaricata has been equated with a divaricately branched plant with minute flowers and branches that often arise immediately below a flowering head. Since 1943, the typical variety or subspecies has been associated with specimens having whitish flowers that dry with pink lobes, and a whitish to yellowish lower throat (sometimes streaked with red) and upper tube. Such plants are represented among the syntypes of Gilia divaricata Torr. ex Gray by Bolander 4908 and Mrs. Davis 49. Choosing either Rattan s.n. (based on Gray's original description, though multiple gatherings is problematic) or Mr. Shelton s.n. (based on Torrey's designation) as lectotype would result in needless and confusing nomenclatural shuffling for one of the most recognizable and geographically widespread Navarretia species. Similarly, considering N. divaricata Greene to represent a new name with Curran's cited specimen as type, independent of G. divaricata Torr. ex A.Gray (which was clearly not Greene's intent), would make N. divaricata synonymous with N. peninsularis. This action would necessitate a new combination for the entity under consideration based on Gilia atrata M.E.Jones, a name buried in obscurity and unused since its original publication. As allowed by the code, we advocate the continued interpretation of Greene's name as a replacement for Gray's, with Gray's syntypes as the original material, and designate one of these syntypes as lectotype to serve as the basionym of N. divaricata Greene. Type. Based on Gilia divaricata Torr. ex A.Gray, non Gilia divaricata Nutt. 1848. Emended description. Taprooted annual herbs to 15 cm tall and 20 cm wide, sometimes larger, often wider than tall. Primary stem erect, terminating in an inflorescence head 1-5 cm above the cotyledons; generally greatly exceeded by secondary stems, with tertiary, and quaternary stems present on larger plants (these higher order stems may be reduced in length and the inflorescence heads ± congested); higher order branches arise from axils of proximal inflorescence bracts, axils of leaves subtending the primary head, or less commonly, leaves within 1 cm of an inflorescence head; branches ascending to spreading and ± leafless, except for leaves subtending higher order branches or within 1 cm of a head; stem and branches tan to reddish-brown, glabrous or sparsely minutely glandular pubescent to glabrescent, less often villous, the trichomes generally less than 0.5 mm long. Cotyledons two, linear, entire, united at base. Leaves somewhat finely stipitate-glandular proximally, less so distally; leaves at the lowermost nodes opposite, linear-filiform, and widened at the point of stem attachment, the proximal nodes often congested with overlapping leaf bases. More distal leaves alternate, entire, or more commonly with 1-3 paired or unpaired linear lateral lobes 1-5 mm long attached along the proximal 3(-15) mm of the leaf, with an elongated, linear terminal segment. Inflorescences head-like, generally ≤ 10 mm diameter (exclusive of bract tips; ~15 mm with bract tips), mostly less than 15 flowered, villous proximally, obscurely glandular. Inflorescence bracts < 10(-13) mm long, ± palmatifid to subpalmatifid; outermost 1-2 bracts with a short achlorophyllous base and 2-3 pairs of lateral lobes flanking an elongate terminal lobe, the distal pair of lateral lobes sometimes shorter and reflexed somewhat out of plane relative to the other lobes; bract bases become larger and clasping centripetally with lateral lobes reduced to a single pair departing from near the apex of the bract base flanking the central terminal lobe, all bract lobes chlorophyllous, entire, long tapering acute. Bracts sparsely villous abaxially, more densely villous adaxially and proximally along the lobes just above their point of attachment, with the distal 1/2 of each lobe more or less glabrous or with a few minute stipitate glands. Flowers actinomorphic, calyces mostly 4.5-7.5 mm long, tube ~ 1.5-2 mm; costae entire, long tapering acute, strongly unequal to subequal with typically two costae longer than the other three; costae narrowing proximally, the shorter ones narrower at base than the intercostal membrane and the longer ones subequal with the membrane; calyx tube achlorophyllous, minutely glandular-puberulent on the intercostal membrane with the costae at least somewhat villous, the trichomes longest along the costae at the junction with the intercostal membrane, the free portion of the costae glabrous to very sparsely and minutely glandular distally; intercostal membrane v-shaped at sinus. Corolla generally shorter than longest calyx costae at anthesis but exceeding the calyx as fruit matures, narrowly funnelform, glabrous, 3.5-5.0 mm long, lobes 0.6-1.0 mm long × 0.4-0.9 mm wide, proximal tube white, distal tube and throat yellow, sometimes red-streaked, transitioning to white lobes suffused with pink or lavender at tips (drying pink); tube base expanding and investing the fruit apex. Stamen filaments unequal, 0.1-0.6 mm long, inserted unequally to subequally 0.2-1.0 mm below corolla sinuses, included in throat to slightly exserted; pollen white, apertures pantoporate, acolpate; sexine seimitectate, reticulate, heterobrochate. Ovary two-chambered, unequally three-valved with two values bearing a septum and the third valve smaller and lacking a septum entirely or nearly so, stigma obscurely threelobed, minute, unequally divided with two stigmatic lobes nearly entirely fused, generally included in the corolla throat; capsules mostly 2-2.5 mm long, dehiscing circumcisally around the base with valves splitting upward loculicidally (completely) and septicidally (partially). Seeds generally 5-9(-12) per fruit, medium brown, ovoid-angular, mucilaginous when wet. Nuclear gene loci showing diploid PCR amplification patterns.

Navarretia divaricata
Habitat, distribution, and phenology. Navarretia divaricata occurs on a variety of soils from (800)1000-2600 meters in foothill and mountain habitats. It is widely distributed from Santa Barbara and Tulare Counties, California in the south to just within the borders of British Columbia, Canada, in the north, and east to Nevada and Idaho (Fig. 2). Flowering time is (May)June-July(September).
Conservation status. Navarretia divaricata has many occurrences throughout its broad range, and is typically abundant when encountered. Occurrences near the periphery of its range (e.g., in British Columbia) may be limited in numbers, but the Etymology. From the Latin divaricatus, to spread or diverge at a wide angle, which aptly describes the characteristic repeated pattern of nearly leafless stems diverging from under flowering heads to give rise to additional, well-separated flowering heads in this species and its morphological allies.
Representative specimens examined. CANADA. British Columbia: Anarchist Mt., 1.6 km north of Hwy 3 rest stop, 6.5 km due east of Osoyoos Lake, 49. Notes. By elevating Navarretia divaricata subsp. vividior to species status, as done below, N. divaricata subsp. divaricata henceforth is designated simply N. divaricata. Fresh N. divaricata have no noticeable odor. The reported presence of this species in Montana is based on a single, misidentified specimen collected in an agricultural field; that specimen differs from N. divaricata in branching structure, having pinnately divided leaves, and possessing different bract and calyx morphology (our determination is N. squarrosa (Eschsch.) Hook. & Arn. [MONT-68910 scan!]. Though the protologue provides no mention of N. divaricata, Peck's rationale for recognizing N. prolifera var. breviflora as distinct from N. divaricata can be inferred from the key and species descriptions in the first edition of his Manual of the Higher Plants of Oregon (Peck 1941). Examination of the type reveals it is simply a narrow, more erect N. divaricata that fits well within the range of variation observed in this species. Peck, perhaps influenced by Mason (1951), came to realize this and removed N. prolifera var. breviflora from the second edition of his manual (Peck 1961).
Navarretia divaricata differs from the remaining divaricately branched taxa of Navarretia (i.e., those treated below plus N. crystallina L.A. Johnson & D.Gowen, N. miwukensis D.Gowen & L.A.Johnson, and N. prolifera, in having two of its three stigmatic lobes nearly entirely fused, and in having unequal fruit valves with two of the three valves normal sized and bearing a septum, while the third valve is smaller and lacks a septum entirely or nearly so (Cronquist et al. 1984;personal observation). These other species have three lobed stigmas and three equal valves, each bearing a septum. Navarretia divaricata's flowers, ± 4 mm long, are among the smallest in Navarretia and typically dry with pink lobes with a white or yellowish throat and tube, though the throat and tube may at times be streaked with red. On herbarium specimens, the contrast between darker, pink lobes and light, whitish or yellowish tube is preserved on many flowers even on specimens 150 years old (Fig. 1). This coloration pattern enables this species to be determined without dissecting flowers to observe the stigma or fruit valves. Emended description. Taprooted annual herbs to 12(-15) cm tall and 20(-25) cm wide, sometimes larger, often wider than tall. Primary stem erect, terminating in an inflorescence head 1-2(-4) cm above the cotyledons; generally greatly exceeded by secondary stems, with tertiary, and quaternary stems present on larger plants; higher order branches arise from axils of proximal inflorescence bracts, axils of leaves subtending the primary head, or less commonly, leaves within 1 cm of an inflorescence head; branches ascending to spreading and ± leafless, except for leaves subtending higher order branches or within 1 cm of a head; stem and branches reddish-brown, commonly glandularvillous, the trichomes generally greater than 0.5 mm (often ± 1mm) long, sometimes glabrescent; distal-most branches generally 0.3-0.5 mm in diameter. Cotyledons two, linear, entire, united at base. Leaves somewhat glandular-pubescent proximally, less so distally; leaves at the lowermost nodes opposite, linear-filiform, and widened at the point of stem attachment, the proximal nodes often congested with overlapping leaf bases. More distal leaves alternate, entire, or more commonly with 1-3(-5) paired or unpaired linear lateral lobes 1-10 mm long attached along the proximal 3-5(-15) mm of the leaf, with an elongated, linear terminal segment. Inflorescences head-like, largest generally ≥ 12 mm diameter (exclusive of bract tips; ≥ 18 mm with bract tips), mostly 10-25-flowered, glandular-pubescent with some minutely-glandular villous trichomes proximally. Inflorescence bracts < 15(-20) mm long, palmatifid to subpalmatifid; outermost 1-2 bracts with a short achlorophyllous base and 2-3(-4) pairs of lateral lobes flanking an elongate terminal lobe, the distal pair of lateral lobes sometimes shorter and reflexed somewhat out of plane relative to the other lobes; bract bases become larger and clasping centripetally with lateral lobes reduced to a single pair departing from near the apex of the bract base flanking the central terminal lobe, all bract lobes chlorophyllous, entire, long tapering acute. Bracts somewhat minutely-glandular villous abaxially, often more densely villous adaxially and proximally along the lobes just above the bract base, glands becoming more prominent and their stipe diminishing in length toward the bract tips. Flowers actinomorphic, calyces mostly 4-7.5(-8.5) mm long, tube ~ 1.5-2.5 mm; costae entire, long tapering acute, strongly unequal to subequal with typically two costae longer than the other three; costae narrowing proximally, the shorter ones narrower at base than the intercostal membrane and the longer ones subequal with the membrane; calyx tube achlorophyllous, glandular-puberulent on the intercostal membrane with the costae at least somewhat glandular-villous, the gland stipes longest along the costae at the junction with the intercostal membrane, diminishing in length toward the chlorophyllous costae tips (may be glabrescent with age); intercostal membrane v-shaped at sinus. Corolla generally equal or shorter than longest calyx costae at anthesis but exceeding the calyx as fruit matures, narrowly funnelform, glabrous, 5.0-7.2 mm long, lobes 0.8-1.1(-1.4) mm long × 0.6-0.9(-1.2) mm wide, tube white proximally, distal tube white or yellow, throat bluish, sometimes streaked with magenta, lobes medium bluish-lavender; tube base expanding and investing the fruit apex. Stamen filaments unequal, 0.3-1.2 mm long, inserted unequal-ly 0.3-1.3 mm below corolla sinuses, anthers ± included in throat to exserted less than half the length of the corolla lobes; pollen blue (white rarely?), apertures pantoporate, acolpate; sexine seimitectate, reticulate, heterobrochate. Ovary three-chambered, stigmatic lobes three, included in to slightly exserted from corolla throat; capsule ~ 2.3-3.3 mm long, dehiscing circumcisally around the base with valves splitting upward. Seeds generally 5-7(8) per locule, medium brown, ovoid-angular, mucilaginous when wet. Nuclear gene loci showing allotetraploid PCR amplification patterns.
Habitat, distribution, and phenology. Navarretia vividior occurs in soils influenced by volcanic activity with favorable water status such as the edges of ephemeral pools and transient rivulets, to open flats or gentle slopes in forested areas from 300-1600 meters elevation. It occurs predominately in the North Coast Range from Lake and Sonoma Counties, California in the south to Humboldt, Trinity, and the western edge of Shasta County in the north (Fig. 4). A few populations cross the central valley and occur on the western flank of the northern Sierra Nevada/southern Cascade Range in Butte County, California, where they flower in late May-early June, rather than late June-July as is typical for species in the North Coast Range.
Conservation status. As defined here, Navarretia vividior is distributed more narrowly than previously considered. Some historical populations have not been revisited for ±100 years, though other known occurrences have been revisited multiple times over the past 20 years with no apparent change in local numbers (beyond what may be expected during drought years). Following IUCN (2012) Red List version 3.1 criteria, this species is most accurately characterized as data deficient, though it likely borders between being Vulnerable to a species of Least Concern.
Etymology. This specific epithet is derived from the Latin vividus, lively or vigorous, with the comparative ending -ior, more so, in reference to the larger, more robust habit and flowering heads this taxon possesses in comparison to N. divaricata. Alternative epithets of var. fertilior and var. vividia were considered by Jepson and Bailey for this entity, as indicated by annotations on the type sheet.    , is difficult to place with confidence; it was collected from 'garden soil' and thus its original source is uncertain, but its smaller corolla features suggest N. vividior. The Torrey 302 syntype of Gilia divaricata Torr. ex Gray also belongs here.
Navarretia vividior is readily distinguished from N. divaricata with its larger flowering head, larger flowers, equally divided 3-lobed stigma, typically blue pollen, and bluish-lavender corolla lobes and throat (Fig. 3). Stems are typically larger in diameter and invested with long, gland tipped hairs. The evenly 3-lobed stigmas, larger flowers, and some variation in flower color with colored lobes, throat, or both apply to three additional species that have heretofore generally fallen under the umbrella of N. divaricata subsp. vividior. These species are described and differentiated below. Diagnosis. A species similar to Navarretia vividior, but distinguished by being diploid, rather than allotetraploid, generally being more conspicuously villous in the proximal inflorescence, possessing slightly larger flowers and more exserted corollas that tend toward pinkish-lavender lobes with darker throat above a yellowish tube rather than bluish to bluish-lavender lobes with darker throat above a yellowish or whitish tube, and having white (rarely blue) rather than blue pollen.
Description. Taprooted annual herbs to 12(-15) cm tall and 20(-25) cm wide, sometimes larger, often wider than tall. Primary stem erect, terminating in an inflorescence head 1-2(-4) cm above the cotyledons; generally greatly exceeded by secondary stems, with tertiary, and quaternary stems present on larger plants; higher order branches arise from axils of proximal inflorescence bracts, axils of leaves subtending the primary head, or less commonly, leaves within 1 cm of an inflorescence head; branches ascending to spreading and ± leafless, except for leaves subtending higher order branches or within 1 cm of a head; stem and branches reddish-brown, commonly glandular-villous, the trichomes generally greater than 0.5 mm (often ± 1mm) long, sometimes glabrescent; distal-most branches generally 0.3-0.5 mm in diameter. Cotyledons two, linear, entire, united at base. Leaves somewhat glandular-pubescent proximally, less so distally; leaves at the lowermost nodes opposite, linear-filiform, and widened at the point of stem attachment, the proximal nodes often congested with overlapping leaf bases. More distal leaves alternate, entire, or more commonly with 1-3(-5) paired or unpaired linear lateral lobes 1-10 mm long attached along the proximal 3-5(-15) mm of the leaf, with an elongated, linear terminal segment. Inflorescences head-like, largest generally ≥ 12 mm diameter (exclusive of bract tips; ≥ 18 mm with bract tips), mostly 12-20-flowered, sometimes more, glandular-pubescent, generally with conspicuous, minutely-glandular villous trichomes proximally. Inflorescence bracts < 15(-20) mm long, ± palmatifid to subpalmatifid; outermost 1-2 bracts with a short achlorophyllous base and 2-3 pairs of lateral lobes flanking an elongate terminal lobe, the distal pair of lateral lobes sometimes shorter and reflexed somewhat out of plane relative to the other lobes; bract bases become larger and clasping centripetally with lateral lobes reduced to a single pair departing from near the apex of the bract base flanking the central terminal lobe, all bract lobes chlorophyllous, entire, long tapering acute. Bracts minutely-glandular villous abaxially, often more densely villous adaxially and proximally along the lobes just above the bract base, glands becoming more prominent and their stipe diminishing in length toward the bract tips. Flowers actinomorphic, calyces mostly 5-7.5(-12) mm long, tube ~ (1.5-)2-2.5(-3) mm; costae entire, long tapering acute, strongly unequal to subequal with typically two costae longer than the other three; costae narrowing proximally, the shorter ones narrower at base than the intercostal membrane and the longer ones subequal with the membrane; calyx tube achlorophyllous, glandular-puberulent on the intercostal membrane with the costae at least somewhat glandular-villous, the gland stipes longest along the costae at the junction with the intercostal membrane, diminishing in length toward the chlorophyllous costae tips (may be glabrescent with age); intercostal membrane vshaped at sinus. Corolla generally equal to exceeding the calyx costae at anthesis and exceeding the calyx further as fruit matures, narrowly funnelform, glabrous, 6.0-8.2 mm long, lobes (1.0-)1.2-1.5(-1.95) mm long × 0.8-1.4 mm wide, tube white proximally, yellow distally, throat lavender-purple or purplish streaked, lobes light to dark pinkish-lavender; tube base expanding and investing the fruit apex. Stamen filaments unequal, 0.3-1.3 mm long, inserted unequally 0.3-1.6 mm below corolla sinuses, anthers ± included in throat to exserted less than half the length of the corolla lobes; pollen white (uncommonly blue), apertures pantoporate, acolpate; sexine seimitectate, reticulate, heterobrochate. Ovary three-chambered, stigmatic lobes three, included in to slightly exserted from corolla throat; capsule ~ 2.6-3.8 mm long, dehiscing circumcisally around the base with valves splitting upward. Seeds generally 4-9 per locule, medium brown, ovoid-angular, mucilaginous when wet. Nuclear gene loci showing diploid PCR amplification patterns.
Habitat, distribution, and phenology. Navarretia modocensis occurs in volcanic influenced soils in forest openings and sagebrush slopes from (390-)800-1700 meters predominately in the Modoc Plateau of northeastern California and adjacent southern Oregon, but extending south to the western flank of the northern Sierra Nevada/ southern Cascade Range in Butte County, California, and with a long-distance disjunct occurrence in San Benito County, California. Flowering occurs primarily in (April-)June-July.
Conservation status. Navarretia modocensis has many occurrences throughout its range and is often abundant when encountered. It is a species of Least Concern following IUCN (2012) Red List version 3.1 criteria.
Etymology. From the Latin -ensis, origin or place, combined with Modoc, in reference to the Modoc Plateau on which this taxon predominantly (but not exclusively) occurs.
Representative specimens examined ( -908393], likely belongs here also, representing a second disjunct population for this species. Pollen is usually white in N. modocensis, but blue pollen has been observed. Navarretia modocensis is the species most likely to be visually confused with N. vividior. Though N. modocensis can have larger inflorescence heads, calyces, and flowers than N. vividior, the range of measurements in these features overlap. We are fairly confident (on fresh flowers) that N. modocensis has a yellow corolla tube on fresh flowers, and that populations with blue pollen are uncommon, but we are less certain that N. vividior always has yellow on its tube, or that populations with white pollen do not exist (given white results from the absence of pigment; we have collected species in several genera, including Navarretia that are characterized by blue pollen but occasionally have populations with white pollen). The more robust features of N. modocensis contrast more sharply with N. aeroides, which has smaller flowers, smaller inflorescence heads, thinner branches, and is visually much more glandular and less villous in its inflorescence heads. The smaller dimensions and very different corolla coloration patterns also readily distinguish N. divaricata and N. torreyella from N. modocensis. Diagnosis. A species similar to Navarretia vividior, but distinguished by being diploid, rather than allotetraploid, and being less robust in all respects; N. aeroides has smaller inflorescence heads that are conspicuously stipitate-glandular throughout (sometimes inconspicuously villous proximally), thinner branches, tends toward smaller corollas, and has stem trichomes mostly 0.5mm or less rather than mostly ± 1 mm.

Navarretia aeroides
Description. Taprooted annual herbs to 9(-12) cm tall and 15(-22) cm wide, sometimes larger, often wider than tall. Primary stem erect, terminating in an inflorescence head 1-2(-4) cm above the cotyledons; generally greatly exceeded by secondary stems, with tertiary, and quaternary stems present on larger plants; higher order branches arise from axils of proximal inflorescence bracts, axils of leaves subtending the primary head, or less commonly, leaves within 1 cm of an inflorescence head; branches ascending to spreading and ± leafless, except for leaves subtending higher order branches or within 1 cm of a head; stem and branches reddish-brown, glandularpubescent or sparingly so, the trichomes mostly less than 0.5 mm long; distal-most branches filiform, generally no more than 0.3 mm in diameter. Cotyledons two, linear, entire, united at base. Leaves somewhat finely stipitate-glandular proximally, less so distally; leaves at the lowermost nodes opposite, linear-filiform, and widened at the point of stem attachment, the proximal nodes often congested with overlapping leaf bases. More distal leaves alternate, entire, or more commonly with 1-3(-5) paired or unpaired linear lateral lobes 1-8 mm long attached along the proximal 3-5(-15) mm of the leaf, with an elongated, linear terminal segment. Inflorescences head-like, generally ≤ 10 mm diameter (exclusive of bract tips; ~15 mm with bract tips), mostly less than 10-flowered, sometimes more, ± conspicuously glandular. Inflorescence bracts < 10(-12) mm long, ± palmatifid to subpalmatifid; outermost 1-2 bracts with a short achlorophyllous base and 2-3 pairs of lateral lobes flanking an elongate terminal lobe, the distal pair of lateral lobes sometimes shorter and reflexed somewhat out of plane relative to the other lobes; bract bases become larger and clasping centripetally with lateral lobes reduced to a single pair departing from near the apex of the bract base flanking the central terminal lobe, all bract lobes chlorophyllous, entire, long tapering acute. Bracts sparsely glandular-villous abaxially, more densely glandular-villous adaxially and proximally along the lobes just above the rachis, with the stipe of each gland diminishing in length toward the bract tips. Flowers actinomorphic, calyces mostly 4.0-6.0(-8.5) mm long, tube ~ 1.3-2.2-(2.5) mm; costae entire, long tapering acute, strongly unequal to subequal with typically two costae longer than the other three; costae narrowing proximally, the shorter ones narrower at base than the intercostal membrane and the longer ones subequal with the membrane; calyx tube achlorophyllous, glandular-puberulent on the intercostal membrane and proximal costae, gland stipes lengthen on the costae at the junction with the intercostal membrane, diminishing in length toward the chlorophyllous costae tips; intercostal membrane v-shaped at sinus. Corolla generally ± equal to the calyx costae at anthesis but exceeding the calyx as fruit matures, narrowly funnelform, glabrous, 4.2-6.0 mm long, lobes 0.75-1.3 mm long × 0.5-0.9(-1.0) mm wide, tube white, transitioning to a bluish throat and lobes in some populations (drying bluish purple, with the distal tube showing hints of magenta or somewhat brownish) or remaining white in others (drying with white or light blue lobes and throat with brownish or magenta distal tube); tube base expanding and investing the fruit apex. Stamen filaments unequal, 0.2-0.55 mm long, inserted unequally 0.4-0.9 mm below corolla sinuses, anthers included in throat to slightly exserted; pollen blue or white, generally matching corolla lobe coloration, apertures pantoporate, acolpate; sexine seimitectate, reticulate, heterobrochate. Ovary threechambered, stigmatic lobes three, included in to slightly exserted from corolla throat; capsule ~ 2.4-3.4 mm long, dehiscing circumcisally around the base with valves splitting upward. Seeds generally 4-8 per locule, medium brown, ovoid-angular, mucilaginous when wet. Nuclear gene loci showing diploid PCR amplification patterns.
Habitat, distribution, and phenology Navarretia aeroides prefers (reddish) clay soils in forest openings from 400-1350(-1900) meters elevation. Occurrences are widely scattered in the Sierra Nevada from Mariposa County, California in the south to Plumas County in the North, and in the Trinity mountains of the North Coast Range, California. This species flowers primarily June-July.

Conservation status.
Many historical collections of this species are sufficiently general in their locality descriptions in areas now populated that our efforts to relocate them, compounded by recent drought years, have been unsuccessful. On the other hand, all but two of our collections were made serendipitously, in the course of looking for other species, suggesting our present knowledge of occurrences is incomplete. Following IUCN (2012) Red List version 3.1 criteria, this species is most accurately characterized as data deficient, though it may be Vulnerable based on the fragmented nature of a limited number of occurrences.
Etymology. From the Latin aeroides, like the sky or sky-blue, in reference to the typical color of the corolla. Notes. Navarretia aeroides are mephic when fresh. In 2015, the population first collected by Barbe, Fuller, & Howell in the mountains east of Quincy, California, was found to have been sprayed with 2,4-D (and blue indicator dye), along with N. propinqua, in an area designated for ORV use, perhaps having been mistaken for immature thistle. To date, occurrences in the Trinity Mountains can be distinguished morphologically (white corollas with magenta streaking in the throat and white pollen) from occurrences in the Sierra Nevada (blue to light blue corollas and blue pollen), yet we have resisted recognizing this difference at the subspecific level. As in any species with colored corollas, occasional white flowered individuals are observed in the Sierra Nevada among a sea of blue flowered individuals. The paratype of Gilia atrata M.E.Jones from Colfax, California [POM-75128 scan!] belongs here.

Representative specimens examined (paratypes). UNITED STATES OF AMER-
Navarretia aeroides is a smaller-featured plant than either N. vividior or N. modocensis, though its corolla overlaps in size with N. vividior. The more conspicuously glandular inflorescence heads (in side-by-side comparisons) contrasts with all of the other species detailed here, and corolla coloration, fresh and dried, readily distinguishes this taxon from N. divaricata and N. torreyella. Diagnosis. A species similar to Navarretia divaricata, but distinguished by having three equal stigmatic lobes and three fully developed fruit valves, and generally larger corollas with a deep maroon distal tube and throat abruptly transitioning to nearly white or less commonly pink lobes, the lobes drying lighter than the much darker throat and tube.

Navarretia torreyella
Description. Taprooted annual herbs to 7(-10) cm tall and 14(-20) cm wide, sometimes larger, generally wider than tall. Primary stem erect, terminating in an inflorescence head 1-2(-3) cm above the cotyledons; generally greatly exceeded by secondary stems, with tertiary, and quaternary stems present on larger plants; higher order branches arise from axils of proximal inflorescence bracts, axils of leaves subtending the primary head, or less commonly, leaves within 1 cm of an inflorescence head; branches ascending to spreading and ± leafless, except for leaves subtending higher order branches or within 1 cm of a head; stem and branches reddishbrown, sparsely minutely glandular pubescent to glabrescent, less often villous, the trichomes generally less than 0.5 mm long; distal-most branches filiform, generally no more than 0.3 mm in diameter. Cotyledons two, linear, entire, united at base. Leaves somewhat finely stipitate-glandular proximally, less so distally; leaves at the lowermost nodes opposite, linear-filiform, and widened at the point of stem at- tachment, the proximal nodes often congested with overlapping leaf bases. More distal leaves alternate, entire, or more commonly with 1-3 paired or unpaired linear lateral lobes 1-5 mm long attached along the proximal 3(-15) mm of the leaf, with an elongated, linear terminal segment. Inflorescences head-like, generally ≤ 10 mm diameter (exclusive of bract tips; ~15 mm with bract tips), mostly less than 15 flowered, villous proximally, obscurely glandular. Inflorescence bracts < 10(-12) mm long, palmatifid to subpalmatifid; outermost 1-2 bracts with a short achlorophyllous base and 2-3 pairs of lateral lobes flanking an elongate terminal lobe, the distal pair of lateral lobes sometimes shorter and reflexed somewhat out of plane relative to the other lobes; bract bases become larger and clasping centripetally with lateral lobes reduced to a single pair departing from near the apex of the bract base flanking the central terminal lobe, all bract lobes chlorophyllous, entire, long tapering acute. Bracts sparsely villous abaxially, more densely villous adaxially and proximally along the lobes just above the rachis, with the distal 1/2 of each lobe more or less glabrous or with a few minute, stipitate glands. Flowers actinomorphic, calyces mostly 4.5-7.8 mm long, tube ~ 1.5-2 mm; costae entire, long tapering acute, strongly unequal to subequal with typically two costae longer than the other three; costae narrowing proximally, the shorter ones narrower at base than the intercostal membrane and the longer ones subequal with the membrane; calyx tube achlorophyllous, minutely glandular-puberulent on the intercostal membrane with the costae at least somewhat villous, the trichomes longest along the costae at the junction with the intercostal membrane, the free portion of the costae glabrous to very sparely, obscurely, and minutely glandular distally; intercostal membrane v-shaped at sinus. Corolla generally ± equal to the calyx costae at anthesis but exceeding the calyx as fruit matures, narrowly funnelform, glabrous, 4.0-6.0(-6.8) mm long, lobes 0.7-1.5 mm long × 0.5-1.0 mm wide, proximal tube white, distal tube and throat maroon, abruptly transitioning to white or whitish to uncommonly pink lobes; tube base expanding and investing the fruit apex. Stamen filaments unequal, 0.25-1.2 mm long, inserted unequally to subequally 0.2-1.2 mm below corolla sinuses, included in throat to exserted less than half the length of the corolla lobes; pollen white (uncommonly light blue), apertures pantoporate, acolpate; sexine seimitectate, reticulate, heterobrochate. Ovary three-chambered, stigmatic lobes three, included in to slightly exerted from the corolla throat; capsule ~ 2.5 mm long, dehiscing circumcisally around the base with valves splitting upward. Seeds generally 2-5 per locule, medium brown, ovoid-angular, mucilaginous when wet. Nuclear gene loci showing diploid PCR amplification patterns.
Habitat, distribution, and phenology. Navarretia torreyella occurs on basalt flats, pyroclastic rubble, and clay soils from 1000-2100 meters elevation, in Butte, El Dorado, Nevada, Placer, Plumas, and Tehama Counties, California. Depending on latitude and elevation, it blooms from (May) June-July (September), beginning its flowering a little earlier than N. divaricata, and about the same as N. crystallina, when these taxa co-occur.
Conservation status. Navarretia torreyella has many occurrences throughout its range and is typically abundant when it is encountered. It is a species of Least Concern following IUCN (2012) Red List version 3.1 criteria.
Etymology. In honor of John Torrey for the plant he recognized, before others, as distinct at the species level.

Representative specimens examined (paratypes). UNITED STATES OF
Key to the taxa formerly treated as Navarretia divaricata 1 Corollas 3.5-5 mm, lobes white or the tips tinged pink to lavender when fresh, drying pink (generally much darker than throat and tube), tube and lower throat yellowish when fresh, similar when dried (sometimes streaked with red); stigmas minute with 2 of 3 lobes fused almost to tips, fruit with 1

Methods
Herbarium specimens were examined directly or via high resolution scanned images of herbarium sheets; scans are indicated as such in specimen citations. Specimen collections were examined from UC, JEPS, DAV, AHUC, RSA, POM, CHSC, IDS, BRY, HSC, CAS, ORE, OSC, and WTC. Single or small groups of particular specimens of interest were provided by GH, NDG, NY, RENO, US, and UBC. Specimens examined also included material gathered from our own field work (deposited at BRY, JEPS, or both). Working iteratively between herbarium specimens, field work, and laboratory examination, taxonomic hypotheses were refined following the unified species concept (de Queiroz 2007). For Navarretia divaricata only, the distribution map (Fig. 2) was compiled by augmenting examined specimens with data obtained from CDA, CIC, DS, HJAEFB, ID, RM, SOC, SRP, WS, and YM as searchable via the Consortium of California Herbaria (http://ucjeps.berkeley.edu/consortium/) and the Consortium of Pacific Northwest Herbaria (http://pnwherbaria.org). Digital photographs were used to verify determinations of the online records when available, particularly with questionable occurrences in disjunct locations; questionable occurrences were not mapped. Morphologically, specimens were grouped under the framework of population aggregate analysis/specimen aggregate analysis (Davis andNixon 1992, Snow et al. 2003). Specimens were examined for qualitative or quantitative features useful for distinguishing among taxa. Digital calipers were used for larger features, while smaller features were measured from digital images taken with an Olympus SZX-12 dissecting microscope using CellSens software (Olympus Soft Imaging Solutions Corp.). All corolla measurements were taken from flowers post-anthesis with expanding to fully mature fruits, after first boiling for ~ 30 seconds and keeping the tissue rehydrated in Pohl's solution (Pohl 1965).
DNA sequences were examined in the context of exclusivity (Brower 1999), with an emphasis on shared patterns across independent loci given that any single locus may violate this principle in recently diverging groups due to lineage sorting or other recognized biological processes. DNA sequence data was initially obtained from 66 accessions representing 10 species as defined herein. This sampling included 11 accessions of Navarretia divaricata and 43 accessions representing the various taxa formerly recognized as Navarretia divaricata subsp. vividior from localities representing the geographic distribution of each taxon. This dataset was simplified a posteriori to include just six accessions of each taxon in the Navarretia divaricata complex and a single representative of five related species, selecting accessions that represent the range of genetic variation observed while eliminating much redundancy in the dataset. Focusing on species delimitation rather than phylogenetic reconstruction, the five additional species included in the analyses are morphologically most similar to the focal taxa: N. crystallina, N. filicaulis (Torr. ex A.Gray) Greene, N. miwukensis, N. peninsularis, and N. prolifera. A concatenated chloroplast matrix was constructed from the 5' trnK intron and 5' portion of matK (Johnson andSoltis 1995, Johnson andJohnson 2006), trnL-trnL-trnF intergenic spacer and intron (Taberlet et al. 1991), trnS-trnG inter-genic spacer (Hamilton 1999), and rpl16 regions (Small et al. 1998). A matrix from the nuclear ribosomal ITS1, 5.8s, and ITS2 region (White et al. 1990, Porter 1996 was also constructed. For these matrices, amplification primers as described in the papers just cited were used in PCR reactions consisting of 30 cycles of 95°C for 1 min, 52°C for 1 min, and 72°C for 1 min. PCR amplicons were cleaned using PrepEase PCR purification plates (Affymetrix, Santa Clara, California, USA) prior to sequencing with BigDye vs 3.1 (Applied Biosystems/Thermo Fisher Scientific, Waltham, Massachusetts, USA). Sequenced products were cleaned with Sephadex and run on an AB 3730 xl DNA sequencer in the DNA Sequencing Center at Brigham Young University. Three additional matrices from low copy nuclear genes were also constructed: pistillata ) and two paralogs of isocitrate dehydrogenase (Weese and Johnson 2005;Johnson and Johnson 2006). To simplify data acquisition and cloning, new taxon-specific primers were designed for this study to amplify 600-800 bp segments of these low copy regions that included introns and exons: PIdv1F 5'-TGGGTACTCAT-AGGTTGGTTGA-3'; Pidv1R 5'-TGCAAGGAGAGACTTACCTGA-3'; idhAdv1F 5'-AGCAATCAAGTGTGCGACAA-3'; idhAdv1R 5'-AGCGGCCACTTCTTCT-GATA-3'; idhBdv354F 5'-CTGCAGATGAAGCTCGTATGG-3'; idhBdv1082R 5'-CGTAAGCTGTGGTCATCGAA-3'. For these low copy nuclear regions, a touchdown PCR protocol was employed starting with a 62°C annealing temperature and decreasing 1°C each cycle until reaching 52°C, then repeating 29 additional cycles with a 52°C annealing. When multiple copies were evident from direct sequencing attempts that made chromatograms unreadable (nuclear regions only), PCR amplicons were first purified using PrepEase PCR purification plates, resuspended in 30 µl water, then cloned via TOPO TA for sequencing kits (Invitrogen/ThermoFisher Scientific, Waltham, Massachusetts, USA), with generally eight colonies per cloning reaction subsequently re-amplified via PCR with standard M13 primers and these amplicons sequenced. Sequences have been deposited in GenBank (Appendix 1). Sequence matrices were aligned by eye using AliView (Larsson 2014). Unweighted parsimony analyses using PAUP* 4.0b10 (Swofford 2003) with 10,000 random addition replications, collapsing branches with minimum length of zero, and assessing support with 100,000 bootstrap replications using fast stepwise addition were performed for each matrix and the results from the separate analyses compared by eye.

Results
The concatenated cpDNA matrix consisted of 35 terminals and 4428 nucleotide characters, of which 51 are parsimony informative. Parsimony analysis of this matrix recovered a single topology of 111 steps ( Fig. 8A; CI = 0.96; RI = 0.99).
The nrITS sequence matrix consisted of 37 terminals with two populations of N. vividior represented by distinct sequences recovered after cloning the original PCR fragments; the remaining four populations of N. vividior provided clean reads from direct sequencing of the original PCR products and were not cloned. This matrix was 631 nucleotide characters in length of which 36 were parsimony informative. Parsimony analysis of this matrix recovered 206052 topologies of 66 steps, which narrowed to eight unique topologies after collapsing branches with a minimum length of zero and then filtering to retain only the shorted trees (still of 66 steps; Fig. 8B; CI = 0.82; RI = 0.95).
The idh-A, idh-B, and PI matrices each contained 41 terminals, with each of the six populations of N. vividior represented by two distinct sequences recovered following cloning of the original PCR products. Some cloned fragments were clearly chimeras of the two primary sequences with a single break point in each fragment that varied in location among fragments, indicating these chimeras were generated within the PCR reactions. The idh-B matrix contained 729 nucleotide characters of which 55 were parsimony informative. Parsimony analysis of this matrix recovered a single topology of 113 steps ( Fig. 8C; CI = 0.95; RI = 0.99). The idh-A matrix contained 880 nucleotide characters of which 117 were parsimony informative. Parsimony analysis of this matrix recovered 108 topologies of 220 steps, which reduced to 13 after condensing to remove branches with a minimum length of zero and filtering to retain only the shortest topologies ( Fig. 8D; CI = 0.87; RI = 0.96). The PI matrix contained 646 nucleotide characters of which 35 were parsimony informative. Parsimony analyses recovered 32 topologies of 83 steps which reduced to nine after condensing to remove branches with a minimum length of zero and filtering to retain only the shortest topologies ( Fig. 8E; CI = 0.81; RI = 0.95).

DNA-based inferences.
In all DNA sequence-based topologies, using markers representing the chloroplast genome and four putatively unlinked nuclear loci (Fig. 8), Navarretia divaricata forms an exclusive group well separated from any of the material previously recognized as N. divaricata subsp. vividior. Though sequences representing the latter taxon form an exclusive group in chloroplast and nuclear ITS sequences, this larger exclusive group is divided into smaller exclusive groups that also appear, with two exceptions, as exclusive groups in each of the three low-copy nuclear gene data sets (i.e., idh-B, idh-A, and PI; Fig. 8). The two exceptions involve N. vividior as defined herein, and N. torreyella. Sampled populations of N. vividior all contained two homeologs in the low copy nuclear gene data sets: one that clusters with N. modocensis, and one that clusters with N. aeroides. This is an expected pattern for a plant of allopolyploid origin. The placement of N. vividior in the cpDNA tree (Fig. 8A) solely near N. aeroides indicates N. aeroides (or its ancestral lineage) was the maternal parent in the formation of N. vividior, whereas N. modocensis (or its ancestral lineage) was the paternal parent. Gene conversion in the nrITS region in some populations of N. vividior has fixed, or nearly fixed this locus in favor of the paternal parent (Fig. 8B), which often happens over time in allopolyploids (e.g., Johnson and Johnson 2006). Allopolyploidy in N. vividior is consistent with the chromosome count of 2n = 36 recorded by Crampton for plants collected not far from the type locality (Crampton 494B [AHUC-038375! Figure 8. Representative most parsimonious, unrooted trees inferred from analysis of DNA sequence data. Base substitutions are reconstructed along interior branches, followed by bootstrap support values. Shaded regions around terminal branches circumscribe individuals of the same taxon, using colors for Navarretia vividior, N. modocensis, N. aeroides, and N. torreyella that correspond with the colors of symbols used in Fig.  4. Branches not found in all most parsimonious topologies for each region are represented by dashed lines. A Single topology inferred from concatenated cpDNA sequences B One of eight topologies inferred from nrDNA ITS sequences C Single topology inferred from nuclear idh-B sequences D One of 13 topologies inferred from nuclear idh-A sequences E One of nine topologies inferred from nuclear PI sequences. . In the idh-A data set (Fig. 8D), Navarretia torreyella failed to form an exclusive group, a pattern that may be interpreted as incomplete lineage sorting in this single locus; the two groups do not correlate with geography and though we did not test it specifically, we suspect that both alleles likely reside within single populations given the geographic proximity of plants from which divergent alleles were sampled. In summary, DNA sequence data support the recognition of five distinct species in what has heretofore been considered a single species with two subspecies.
Comparative morphology with similar species. With the exception of Navarretia filicaulis, all of the species included in this study share a common branching architecture. They also share features of flower and fruit that distinguish them from other Navarretia. The branching architecture consists of a short primary stem terminating in a head-like inflorescence with elongate, more or less leafless secondary stems arising from leaf or outermost bract axils at the base of the inflorescence. Each secondary stem terminates in a head-like inflorescence with tertiary, and even quaternary stems similarly arising from the axils of the outer inflorescence bracts of the higher-order inflorescences, or from leaves that occasionally appear within one cm of an inflorescence (Figs 1, 3, 5-7). When well branched, this pattern gives rise to plants that are typically wider than tall. All species can produce depauperate plants consisting of a single head, and N. divaricata sometimes produces plants that are noticeably taller than wide-perhaps when vegetation is dense. Following pollination, it is common in these species for the corolla to detach from the receptacle but remain vested around the upper half of the enlarging fruit, with the corolla base stretched greatly. Spent corollas thus typically adhere to the fruit through maturation rather than being pushed off by the expanding fruit as is common in other Navarretia. The fruit detaches circumcisally about its base with the valves separating more or less from the base to the apex. These flower and fruit characteristics are shared by N. filicaulis that may even, uncommonly, branch divaricately from the lowermost inflorescence bracts. Navarretia prolifera (two subspecies) are distinguished from the species elaborated here by possessing larger flowers with long-exserted stamens. Navarretia crystallina and N. miwukensis, recently described, vary in their inflorescence architecture by having flowers inserted directly on a common receptacle (Johnson et al. 2016). Navarretia peninsularis, at one time placed as a variety of N. divaricata (Jepson 1943), is distinguished by having broader terminal lobes to its leaves and bracts and wider calyx costae. Similarity in overall plant size, habit, and diminutive flowers, combined with the absence of analytical study beyond that conducted by Jepson and Bailey (Jepson 1943), have undoubtedly contributed to the view that N. divaricata as heretofore defined represents a single species.
Recognition of near-cryptic species. Morphology and one's perception has been the guiding force of species delimitation for centuries. Though some may not admit it, many botanists in the field are sympathetic to Cronquist's (1978) view that "species are the smallest groups that are consistently and persistently distinct, and distinguishable by ordinary means" (often interpreted as a 10× hand lens). That some plants can distinguish their own pollen versus that from a genetically different individual in the same population demonstrates that organisms can perceive what our eyes cannot. It should not, then, be surprising that different species (i.e., independent evolutionary lineages with distinct trajectories) may be distinguishable genetically by the organisms themselves and biologists with appropriate tools, despite little or no discernable morphological differentiation. This may be inconvenient for field taxonomists, and recognition of such species may be challenging by human eye, yet the taxonomic recognition of biological reality is still merited (Judd et al. 2007;Carolan et al. 2012;Jörger and Schrödl 2013). Withholding taxonomic recognition for cryptic or nearly cryptic species may simplify the placing of a name on an entity by sight, yet also compromises our understanding of diversity via inaccurate estimates of species diversity, superficial understanding of diversification patterns and processes, and inaccurate assessments of species abundance and ecological preferences.
Over several years of field work without observing hybrid swarms or convincing intermediate forms, we suspected that Navarretia vividior was a species distinct from N. divaricata. In the process of sampling broadly across these species' ranges, the distinctiveness of N. torreyella soon came into focus both morphologically and molecularly. Though similar in many respects, it would be inaccurate to characterize N. divaricata, N. vividior, and N. torreyella as truly cryptic-they can be readily diagnosed via observable morphological differences that can be articulated in a dichotomous key. Either N. modocensis or N. aeroides could replace N. vividior in the above three-way species comparison with little to no editing of the dichotomous key (depending on feature choice). The challenge, morphologically, is in distinguishing between N. vividior, N. modocensis, and N. aeroides. Among these latter three species, N. aeroides is most distinctive and may be appropriately considered "near-cryptic"; it is discernable from N. modocensis in features relating to the corolla and inflorescence size, and inflorescence glandularity. Nevertheless, the morpho-space distinguishing N. aeroides from N. modocensis is narrow. Consequently, their allopolyploid derivative, N. vividior, has a limited morphospace for intermediacy by which it can be distinguished from its parents, and in that limited space, it lies closer to N. modocensis, its paternal parent, than to N. aeroides, but overlaps with both. Navarretia vividior and N. modocensis are the closest of several morphologically similar species pairs we have investigated (e.g., Johnson et al. 2013Johnson et al. , 2016 to being truly cryptic. It is not surprising that the distinctiveness of these taxa has been overlooked in the past, and we recognize that difficulty exists in differentiating these species, particularly in their pressed and dried condition on herbarium sheets. Geographic distribution and syntopy. The taxonomy proposed herein alters our understanding of species distributions and abundance. In its broad distribution that largely encompasses the ranges of the other taxa (compare Figs 2, 4), Navarretia divaricata is known to co-occur with N. modocensis and N. torreyella in many locations, and with N. aeroides and N. vividior in at least some locations. It can also be found with many other Navarretia, including N. leptalea, N. breweri, N. capillaris, N. linearifolia, N. propinqua, N. intertexta, N. crystallina, N. miwukensis, and N. prolifera in portions of its range.
The distribution of N. vividior is narrowed to predominantly the North Coast Range, with the exception of apparent occurrences in the foothills of the northern Sierra Nevada/southern Cascade Range in Butte County, California where it may co-   (Fig. 4). In Trinity County, California, it co-occurs with N. aeroides in at least two locations and at times with N. divaricata. We have also collected N. vividior growing with N. atractyloides, N. mellita, N. squarrosa, N. intertexta, N. lecucocephala, N. subuligera, and N. linearifolia. Navarretia modocensis' distribution extends beyond the Modoc plateau (Fig. 4). It overlaps with a portion of the range of N. torreyella (Fig. 4), though we have not found them co-occurring (both have been collected in the Butte Meadows area). We also have yet to find N. modocensis co-occurring with N. aeroides or N. vividior, though such locations might exist in Butte County, California. We have collected N. modocensis with N. atractyloides, N. breweri, N. divaricata, N. filicaulis, N. intertexta, N. propinqua, N. subuligera, N. sinistra, and N. linearifolia in the principle portion of its range, and the disjunct occurrence in San Benito county was co-occurring with N. mellita.
The distribution of Navarretia aeroides covers considerable geographic area (Fig. 4), yet it has been seldom collected, with over half of the occurrences known only from historical collections over 60 years old. Occurrences appear to be highly fragmented and localized. Awareness of this taxon will undoubtedly aid efforts to better define its abundance. All occurrences in the North Coast Range that we are aware of have been pale (whitish) flowered individuals with white pollen that contrast sharply with the larger and hairier-headed N. vividior with its blue flowers and blue pollen when the two co-occur. Besides N. vividior, we have observed N. aeroides co-occurring with N. atractyloides and N. intertexta in the North Coast Range, and with N. divaricata, N. filicaulis, N. propinqua, N. prolifera subsp. lutea, N. miwukensis, N. leptalea subsp. leptalea, in addition to growing near N. torreyella in the Sierra Nevada. The latter occurrence, observed in July, found N. aeroides on one side of a dirt 4WD road in bloom with N. divaricata and N. prolifera subsp. lutea, while N. torreyella occurred completely senesced in monoculture on the opposite side of the road.
Though N. torreylla has the smallest geographic distribution of the species considered here (Fig. 4), it occurs abundantly both in number of individuals per occurrence and number of occurrences within its range. It grows intermixed with N. divaricata in many locations and herbarium sheets of mixed collections originally determined as either N. divaricata subsp. divaricata or N. divaricata subsp. vividior exist. Navarretia propinqua, N. prolifera subsp. lutea, N. crystallina, and N. leptalea subsp. leptalea also co-occur at times, and at least one observation exists of it growing near N. aeroides as described above.
Butte County, California collections. Several collections housed at CHSC are intriguing because they compare favorably with either N. modocensis or N. vividiorand N. vividior is otherwise found only on the west side of the central valley. Occurring at somewhat lower elevations, in a hotter environment in the transition between the northern Sierra Nevada and southern Cascade Range, some of these occurrences flower earlier than typical for both N. modocensis and N. vividior. Because of their parent/offspring relationship and the difficulty at times in distinguishing these two species on herbarium sheets, we have attempted to relocate these populations over the past several years with limited success because of drought, private property access, and possibly invasive species density. For example, in 1980, Schlising & Azevedo 3699 col-lected N. modocensis at an ecotone that we visited in 2015; the area was extremely dry and though N. filicaulis was found in abundance, a thorough search found no sign of N. modocensis. A return visit in 2017, following closer to normal precipitation in the preceding Fall and Winter months (http://www.usclimatedata.com) found the area to be lusher, and both N. modocensis and N. filicaulis present in abundance BRY). On the other hand, no sign of N. modocensis could be found at the site of Jokerst et al. 462 in 2017, yet the species was described as common when collected in 1979 growing with N. tagetina, which was found in abundance in 2017. Oswald 578 documents what we have determined to be N. vividior along the north rim of Upper Bidwell Park in 1983 following a burn. In 2017, N. intertexta, N. pubescens, N. tagetina, and N. viscidula were observed in abundance hiking the entire length of the north rim trail without trace of N. vividior, but many areas of possible habitat were covered with dense, near monotypic stands of Centaurea solstitialis L. growing through considerable plant litter from previous years. Efforts to contact land-owners for property access permission may enable occurrences that have not been recollected (to our knowledge) in the last 30-40 years to be relocated and assessed to see if any represent mixed populations along Cohasset Road (Schlising 3435, Ikeda 383, Oswald 1999), near Paradise (Ahart 1906), and along Ponderosa Way (Taylor 1393, Oswald 3773).