The taxonomic history of Claytonia lanceolata Pursh (western springbeauty, lanceleaf springbeauty) is fraught with confusion. This geographically widespread species was considered the most morphologically variable of all tuberous, perennial Claytonia L. (Montiaceae) by Davis (1966). This sentiment was echoed by Chambers (1993) and later by Stoughton and Jolles (2013). Stoughton and Jolles (2013) suggested that molecular data taken in combination with morphological and natural history observations would help elucidate evolutionary history and clarify long-standing taxonomic problems associated with this enigmatic taxon (See Davis 1966, but also Miller and Chambers 2006). Studies relying on a small number of ‘exemplar’ specimens inevitably underestimate the overall range of morphological variation for a geographically widespread taxon. This seems to be the primary reason why application of the name C. lanceolata s. s. (i.e. sensu Pursh 1814; s. s.) has been uncertain.

Pursh's (1814) original description of C. lanceolata encompasses only a fraction of the morphological variation more recently observed in the species by Miller and Chambers (2006), and Pursh (1814) did not provide measurements or proportions of any kind beyond the general shape of a few key plant parts (i.e. cauline leaves, sepals, and petals). Pursh (1814) designated a type specimen from the Rocky Mountains and provided an illustration of a plant meeting the vague description of C. lanceolata. Otherwise, he provided no information about the existence or location of additional, representative collections but did suggest that material collected from eastern Siberia matched C. lanceolata perfectly (Pursh 1814). Presumably, the material from Siberia referenced by Pursh (1814) is now considered C. tuberosa Pallas ex Willdenow or C. czukczorum Volkova. These species exhibit similar morphology to C. lanceolata but are not treated as conspecific (Miller and Chambers 2006). Recent taxonomic circumscriptions for C. lanceolata do not include plants from Siberia. Unfortunately, most of the infraspecific taxa previously ascribed to C. lanceolata s. l. were originally described from few specimens (in some cases only the type collection). As a first step toward understanding this taxonomically enigmatic group across all of western North America, and due to the complexity of the situation,we limit our focus in this paper to taxa occurring in California and treated as C. lanceolata by Miller and Chambers (2006).

Morphological variation has been inadequately captured by type specimens and paratypes in taxonomic diagnoses resulting in a proliferation of taxonomic names associated with C. lanceolata (Stoughton and Jolles 2013). These taxa have been taxonomically treated in different ways for a variety of regional floras (Table 1). For example, C. lanceolata var. rosea (Rydberg) R. J. Davis was described as a unique species by Rydberg (1904) from three specimens collected at the type locality in Colorado. This example represents a case where further systematic study clarified at least one small part of the clouded taxonomy of C. lanceolata s. l. Although C. rosea Rydberg was an available name early in the 20th century (Rydberg 1904), it was not until the extensive field observations and cytological work of Halleck (1963) and Halleck and Wiens (1966) that C. rosea was generally accepted as distinct from C. lanceolata. Similarly, taxonomic issues persist in California concerning infraspecific taxa previously treated as part of C. lanceolata. First, two taxa previously treated as varieties of C. lanceolata (C. l. var. sessilifolia (Torrey) A. Nelson and C. l. var. peirsonii Munz and Johnston; Table 1) have type localities in California but have recently been treated as synonyms of C. lanceolata. A third taxon, C. obovata Rydberg, synonymized with C. lanceolata by Davis (1966), has a type locality in Oregon but Californian populations in Mendocino County are specifically mentioned in the protologue by Rydberg (1932).

Claytonia lanceolata, as circumscribed by Miller and Chambers (2006), ranges in distribution from Canada to near Mexico and encompasses the variation associated with all three of these taxa (C. obovata, C. lanceolata var. peirsonii, and C. lanceolata var. sessilifolia), but currently no infraspecific taxa are recognized. In California, plants conforming to Miller and Chambers' (2006) interpretation of C. lanceolata (Consortium of California Herbaria 2015; T. Stoughton pers. obs.) can be found on a broad spectrum of substrates with vast compositional differences falling into three main categories: (1) those rich in mafic minerals (e.g. gabbro, peridotite, and serpentinite), (2) those rich in alkali minerals (e.g. gneiss, granite, rhyolite, schist, and slate), and (3) siliciclastic or carbonate-dominated (meta)sedimentary rocks (e.g. greywacke, limestone, marble, sandstone, and shale). Further complicating the matter, C. lanceolata s. l. grows in California at a wide range of elevations, fromca. 575mto near 2750m(Consortiumof California Herbaria 2015), and it is found in a wide range of habitat types from transmontane pinyon-juniper and oak woodland habitats to montane or cismontane mixed-conifer forest, subalpine forest, and mesic meadow habitats (Chambers 1993; Stoughton and Jolles 2013; Consortium of California Herbaria 2015).

Table 1.

Taxa of the Claytonia lanceolata species complex as treated by earlier authors and in the present treatment. An ‘X’ indicates that a taxon was accepted under the name listed in the leftmost column. No entry indicates that the taxon had not yet been described when the work was published. Not treated indicates that a taxon was either not accepted, or unaddressed (e.g. in a regional flora where that taxon does not occur).

Two key factors appear to have had a substantial influence on application of the name C. lanceolata to plants in California and elsewhere in western North America. First, most members of the C. lanceolata species complex (Table 1) appear above ground and vanish quickly in a number of weeks (Miller and Chambers 2006; Stoughton and Jolles 2013). This ephemeral snowmelt ecology likely contributes to the modest number of collections for the group, which in turn limits material available for taxonomists. At least in southern California, plants representing C. lanceolata s. l. generally flower and completely disappear during the spring season well before the ‘peak’ period of botanical collecting (T. Stoughton pers. obs.). The dearth of C. lanceolata s. l. specimens in the pooled records of 35 herbaria available through the Consortium of California Herbaria (only 226 records, including duplicates, as of 14 Nov 2015) suggests that an early blooming period hinders full understanding of the taxa, geographic distribution, and overall range of morphological variation. Claytonia perfoliata Donn ex Willdenow, a lower elevation, annual species that is part of an equally perplexing group of closely related species, has more than nine times as many collections served on the Consortium of California Herbaria (maximum of 2000 records returned, including duplicates, as of 14 Nov 2015). Second and more substantially, the succulent nature of Claytonia poses a challenge to preserving diagnostic morphological characteristics in this group. Claytonia plants have relatively high saturated water content (Ogburn and Edwards 2012) which causes specimens to wilt quickly following removal from the ground. The fluids in Claytonia tissues cause newspaper and blotters used for collecting to stick to the plants, obliterating many morphological features during the drying process (T. Stoughton pers. obs). Other alternatives for preserving morphological features, such as pickling Claytonia tissues in spirits, tend not to be common practice for most general collectors and can often be precluded by site accessibility issues (e.g. space requirements when backpacking to multiple, remote sites).

In this study, we revisit and address taxonomic uncertainty for the Californian members of the C. lanceolata species complex using morphometric analyses of key morphological characters. We provide a revised taxonomic treatment for C. lanceolata s. l. in California using these data in combination with preliminary molecular phylogenetic data and information about natural history (including substrate association). Our revised taxonomic treatment includes recognition of C. lanceolata and C. obovata in addition to three new species, C. panamintensis T. R. Stoughton, C. peirsonii (Munz and Johnston) T. R. Stoughton (including four subspecies), and C. serpenticola T. R. Stoughton.

Materials and Methods

Sampling—Going forward in this manuscript, we use ‘Claytonia lanceolata s. l.’ to refer specifically to the treatment of C. lanceolata by Miller and Chambers (2006), whereas C. lanceolata s. s. refers to the concept of Pursh (1814) and ‘C. lanceolata species complex’ refers to the taxa listed in Table 1. Over the course of four field seasons (between February and July, 2011–2014), hundreds of plants resembling C. lanceolata s. l. were examined and collected in California (Fig. 1), Oregon, Idaho, and Montana as part of a broader study on phylogenetic relationships of tuberous, perennial Claytonia (T. Stoughton unpubl. data). To better understand variation in plants of C. lanceolata s. l. (sensu Miller and Chambers 2006), specimens previously identified as C. lanceolata from throughout the western U. S. A. held at BABY, CAS/DS, HSC, NSMC, NY, OSC, RSA/POM, UC/JEPS, UNLV, and UNR herbaria were examined.

Phylogenetic Analysis—To investigate patterns of genetic differentiation in C. lanceolata s. l. (i.e. as treated by Miller and Chambers 2006), we conducted Bayesian phylogenetic inference using gene sequences available on GenBank ( http://www.ncbi.nlm.nih.gov/genbank/; see Benson et al. 2005). GenBank sequences were supplemented with new sequences generated at Rancho Santa Ana Botanic Garden (RSABG) for C. lanceolata s. l. collected from populations both within and outside of California (Appendix 1). Samples used for this study were freshly collected material dried in silica from the 2011–2014 field seasons. Exhaustive geographic sampling was not conducted for C. lanceolata s. l., given its wide distribution and the diversity of habitat types it occupies, but samples from all Californian taxa we recognize in this paper were analyzed. This set includes material from the San Bernardino Mountains, San Gabriel Mountains, and Panamint Mountains of southern California, the Klamath-Siskiyou region (including North Coast Ranges) of northern California, and the Sierra Nevada. Additional samples of C. lanceolata s. l. were included from the mountains of Idaho, Nevada (on GenBank), Oregon, and British Columbia, Canada to represent the extremes of its geographic range. Pre-existing sequence data on GenBank sampled from across the genus Claytonia (O'Quinn and Hufford 2005; M. Hershkovitz unpubl. data) were added for comparisons and to provide phylogenetic context for the gene sequences developed at RSABG. This phylogenetic comparison included more than half (5/9) of the tuberous, perennial Claytonia species recognized by Miller and Chambers (2006), but in no way represents an attempt to resolve the phylogeny of Claytonia. Rather, we include a phylogenetic estimate as corroboration of taxonomic concepts, which are based principally on morphology.

Fig. 1.

Map of locations in California and Oregon visited during 2011–2014 field seasons and referenced in text. Georeferenced specimens from the Consortium of California Herbaria (CCH 2015) are gray circles labeled as C. lanceolata s. l.

Genomic DNA was isolated at RSABG from leaf material of 12 C. lanceolata s. l. samples (classified in Vegetative Morphology analysis below) using a protocol modified from the CTAB-method of Doyle and Doyle (1987). Two or more individuals per taxon were used to sample multiple exemplars in the C. lanceolata species complex (15 total samples). Nine samples of other tuberous, perennial Claytonia were included as outgroups with respect to C. lanceolata s. l. Thirty-one total samples were used for the phylogenetic analysis, including more distant outgroups from Claytonia and Lewisia (Appendix 2).

Sequence data of the internal transcribed spacers (ITS1, ITS2) flanking the nuclear ribosomal 5.8S exon (nrITS) were obtained from all samples following manufacturer's instructions for Phusion® High-Fidelity DNA Polymerase (New England Biolabs, Ipswich, Massachussetts). The nrITS gene region was amplified on a GeneAmp PCR System 9700 (Applied Biosystems, Foster City, California) in a single reaction using the plant specific primer pair ITS5a (Prince and Kress 2006) and 26SR (Prince 2010). Reaction conditions for all amplifications were: 98°C for 30 s; 35 cycles of 98°C for 10 s, 48°C for 20 s, and 72°C for 20 s; an extension cycle for 7 min at 72°C followed the 35 cycles, and samples were held at 4°C until removed and stored at -20°C. Amplification products were then purified using a polyethylene glycol (PEG) precipitation. Purified PCR products were processed using a BigDye Terminator (Applied Biosystems) cycle sequencing reaction with the same primers (albeit at lower concentrations) following themanufacturer's instructions. Cycle sequencing products were purified using Sephadex columns (Amersham Biosciences, Piscataway, New Jersey) and analyzed on an automated sequencer (3130xl, Applied Biosystems).

Geneious v. 6.1.2 (Biomatters Ltd., Auckland, New Zealand) was used to edit, assemble and align sequences, resulting in an alignment 639 base pairs long. Bayesian inference (BI) was conducted using MrBayes v. 3.1.2 (Huelsenbeck and Ronquist 2001), and posterior probability values (PP) are reported for all nodes ≤ 0.95. One replicate of BI was conducted using flat priors. The BI analysis was run for ten million generations, with four chains, and sampled every 1000 generations. The final average standard deviation of split frequencies was less than 0.01. Inspection of the resulting parameter files using Tracer v. 1.6. (Rambaut et al. 2014) indicated that effective sample sizes were sufficiently high to meet statistical assumptions. The first 25% of trees were discarded from the final tree set as burn-in; remaining trees were used to determine the PP distribution for the nrITS gene region. Information on vouchers and GenBank numbers are included in Appendix 1, and data are available from the Dryad Digital Repository:  http://dx.doi.org/10.5061/dryad.qm344 (Stoughton et al. 2017).

Geomorphic Context—Because C. lanceolata s. l. is known to occur on a broad diversity of rock types, substrate affinity was identified from parent rock material that was collected from around Claytonia plants at selected field sites in California and southern Oregon (Appendix 1). Slope aspect, geomorphic landform, elevation, associated species, and other local site information was also recorded in the field at these sites and is included on herbarium labels for all collections.

Vegetative Morphology—Five morphological characters (stem length [measured from proximal end of stem at tuber apex to attachment point of the cauline leaf pair], cauline leaf length [measured from stem-petiole attachment to leaf apex], cauline leaf width [measured at widest point of blade], cauline leaf length/width ratio, and peduncle length [measured from attachment point of cauline leaf pair to base of proximal-most pedicel]) were measured and used in morphometric analyses comparing all of the plants occurring in California treated as C. lanceolata by Miller and Chambers (2006). Characters were selected because they exhibit variation within and among species and are easily measured; the drying process does not appear to distort leaves and stems as much as tubers and floral parts (T. Stoughton pers. obs.). Measurements were taken from 19 pre-existing RSA/POM herbarium specimens collected in California and from an additional 89 specimens (including duplicates) collected in California and Oregon contributed by this study (108 total herbarium specimens; Appendix 1). Measurements provided in the current study were made from dried herbarium specimens specifically for the enhanced identification and curation of specimens already existing in herbaria.

To best capture variation within a single population, measurements were taken from all measurable plants on a given herbarium specimen and averaged for each character to obtain one set of character means per specimen. Herbarium specimens were then identified using the taxonomic concepts presented in this study (see Taxonomic Treatment below), a limited number of samples were assayed using molecular phylogenetic data (see Phylogenetic Analysis above), and the taxonomically sorted morphometric data were ultimately used for principal components analysis (PCA) and discriminant analysis (DA) to characterize variation. Prior to multivariate analyses, morphological characters were examined for independence, multi normality and outliers. In a few cases, outlying measurements were given the maximum or minimum quartile value for that taxon/character combination depending on which extreme of the first and third quartiles they fell, and missing data was filled with average values for that taxon/character combination. Peduncle length was correlated with stem length and was therefore omitted from multivariate analyses. The remaining characters were log-transformed to achieve normality. Subsequently, both PCA and DA were conducted in R v. 3.1.2 (R Core Team 2014) using the STATS::princomp and MASS::lda functions (Ripley 1996; Venables and Ripley 2002), respectively. The stats::predict function in R was also used to determine how effectively specimens were classified in the DA using our a priori taxonomic categorization by dividing the number of correctly predicted samples by the total number of samples for that taxon. A χ² test was used to determine whether the DA diagnosed the taxonomic identity of our samples with greater accuracy based upon the morphological characters measured compared with random assignment. These analyses (i.e. both the PCA and DA) were conducted including all samples listed in Appendix 2 in a single analysis (eight taxa, n = 108) and separately for each geographic region (i.e. north California [three taxa, n = 42] and south California [5 taxa, n = 66]).

Results and Discussion

Phylogenetic Analysis—Multiple unique lineages attributed to C. lanceolata are uncovered in the nrITS gene tree with strong PP support (i.e. PP = 1.0 for C. lanceolata, C. obovata, C. panamintensis, and C. peirsonii clades), all of which are circumscribed in the Taxonomic Treatment below. However, monophyly of C. lanceolata s. l. (i.e. as treated by Miller and Chambers 2006) cannot be rejected based on the PP distribution of our BI analysis (Fig. 2). Results are inconclusive (regarding monophyly of C. lanceolata) due largely to an unresolved backbone separating major lineages within a clade that includes all tuberous, perennial Claytonia sampled in this study.

Claytonia caroliniana Michaux, C. virginica L. and C. tuberosa are inferred as paraphyletic in our BI analysis, but these results are not surprising considering the historic work on C. virginica by Doyle (1981, 1983, 1984a, b), Doyle et al. (1984), and Doyle and Doyle (1988). These results also support a recent effort by Yatskievych et al. (2013) to incorporate natural history information in recognizing a new species of Claytonia from Arkansas that is part of this primarily arctic species complex (theC. virginica complex, including C. tuberosa and C. caroliniana). Furthermore, monophyly of sect. Claytonia L. is called into question based on the position of two C. serpenticola samples together in a trichotomy that includes C. acutifola Pallas ex Willdenow and the rest of the tuberous, perennial Claytonia (Fig. 2). Claytonia acutifolia is a caudicose perennial and has been traditionally treated in another section. Miller and Chambers (2006) suggested abandonment of the sectional classification in Claytonia until further study can elucidate the morphological and anatomical transitions to perennial habit.

Fig. 2.

Bayesian 70% consensus phylogeny of Claytonia lanceolata s. l. estimated using data collected from the nrITS gene region. Branches are unlabeled where PP support is ≥ 0.95, support values are specified for branches with PP ≥ 0.70 but < 0.95, and branches are collapsed where PP support is<0.70. Numbers associated with each terminal correspondwith a list of GenBank accession numbers in Appendices 1 and 2. Highlighted taxa were treated as C. lanceolata by Miller and Chambers (2006).

Use of nrITS in this study may be problematic, considering the possibilities of concerted evolution and multiple copies (especially in polyploid plants, of which there are many in Claytonia), but we do not believe its use detracts significantly from our main conclusions. We used different PCR primers than O'Quinn and Hufford (2005) and M. Hershkovitz (unpubl. data), yet samples corresponding to our morphological concept of C. lanceolata s. s. are resolved as a clade with samples from these other studies, providing no evidence that different copies of nrITS were amplified. The results of this preliminary analysis do not permit new inferences regarding relationships among tuberous Claytonia, but this was not a specified goal of this study. Instead, our phylogenetic results fully complement our morphological analyses of Californian taxa (see below). Given our discovery of strongly supported lineages (Fig. 2), at minimum the results are suggestive that the current treatment of C. lanceolata by Miller and Chambers (2006) should be reconsidered. The phylogenetic status of sect. Claytonia (implying a single origin of the tuber), and relationships of tuberous species within that putative clade, are the subjects of our continuing research using high-throughput sequencing.

Geomorphic Context—Both in the field and in herbaria we observed edaphic differences (habitat differences) that are correlated with morphological differences among taxa in both northern and southern California. Several clear ecological differences were observed within C. lanceolata s. l. in northern California and southwestern Oregon where plants grow on mafic rocks (e.g. gabbro, serpentinite, peridotite) and siliciclastic or carbonate-dominated sedimentary substrates (e.g. greywacke, limestone, shale) in comparison to sites where populations occur in the same vicinity on more alkali-rich lithologies (e.g. granite, rhyolite). Ecological differentiation was also observed between populations of C. lanceolata s. l. in southern California where it grows on alkali-rich lithologies (e.g. gneiss, granite, rhyolite, schist) in comparison to sites in the same vicinity where populations are found on siliciclastic or carbonate-dominated (meta)sedimentary rocks (e.g. limestone, marble, sandstone, shale, slate). In general, a combination of character differences were observed among plants growing in northern California (i.e. the Klamath-Siskiyou region; Fig. 3A–B), those in the Northern and Central Sierra Nevada regions (Fig. 3C), and those from southeastern California (Fig. 3D–H). Observed morphological differentiation among populations on different substrates reflects changes in betalain pigmentation, inflorescence architecture, and morphology of cauline leaves, subterranean stems, and flowers (Fig. 3). Ecological differences correspond with changes in parent rock material, soil type, and water availability, all of which are generally reflected by vegetation associations.

Morphological and ecological differences among taxa are summarized in the Taxonomic Treatment below. Explicit consideration is given as to whether a taxon's occurrence on a given substrate in a given region indicates a physiological/ ecological requirement, or rather mere happenstance (e.g. a particular substrate type is common in the region where a taxon is known). With so few populations identified, particularly for the southern Californian taxa, edaphic constraints on species distributions are still unclear for most of the C. lanceolata species complex.

Vegetative Morphology—PCA of the morphological variables we measured for herbarium specimens indicates that the first three components summarize 99.7% of the total variation in the dataset with the first component explaining 57.7% (Fig. 3J; Table 2). The highest loadings in the first component indicate that cauline leaf length/width ratio (0.737) and width (-0.655) contribute most to the total variation (Table 2). This is not surprising given that cauline leaf morphology has been given a great amount of attention in nearly all treatments of the C. lanceolata species complex (Table 1). Plotted scores for samples of all Californian taxa (8 taxa, n = 108; Fig. 3I) showed differentiated clustering of samples assignable to each taxon we circumscribed, with the vast majority of overlap occurring among taxa in different regions of California rather than among taxa in the same geographic area. In other words, the PCA indicates that morphological similarity for the characters measured is shared mostly among northern and southern Californian taxa but not among the collective taxa within either geographic area. One exception is the substantial overlap between two taxa in southern California belonging to the C. peirsonii species complex (Fig. 3E–H, and see Taxonomic Treatment below).

Regional analyses show comparable (but greater) separation of discrete clusters, particularly for the taxa in southern California. In the northern California regional PCA (three taxa, n = 42, Fig. 3L), the first three components summarize 99.8% of the total variation with the first component explaining 66.6% and the strongest loading contributed by cauline leaf length/ width ratio (0.819; Table 2). In the southern California regional PCA (five taxa, n = 66, Fig. 3M), the first three components summarize 99.6% of the total variation with the first component explaining 60.9% and the strongest loading contributed by cauline leaf width (-0.743; Table 2). Furthermore, we found, as expected, that the four lineages identified within the C. peirsonii species complex from southern California (treated as subspecies of C. peirsonii, see Taxonomic Treatment below) cluster closely together but are distinct from C. panamintensis. The results from both PCA analyses suggest strong morphological divergence on a regional scale but convergence at broader levels, as revealed by the analysis of all Californian taxa together (Fig. 3I). Morphological similarity in this group is likely the result of retained ancestral similarity, hybridization, or some other factor(s) not discussed here.

Fig. 3.

Morphology of Claytonia lanceolata s. l. in California. A–C. Northern Californian taxa. A. C. obovata. B. C. serpenticola. C. C. lanceolata. D–H. Southern Californian taxa. D. C. panamintensis. E. C. peirsonii subsp. peirsonii. F. C. peirsonii subsp. bernardinus. G. C. peirsonii subsp. californacis. H. C. peirsonii subsp. yorkii. I–M. Results from morphometric analyses. I. PCA scatter and J. scree plot showing proportions of explained variance for all taxa accepted in California analyzed together; K. DFA scatter plot for the same, and PCA scatter plots for regional analyses of L. northern and M. southern Californian taxa analyzed separately. Confidence interval (95%) ellipses are included on all scatter plots.

Table 2.

Results of the principal components analyses for Californian Claytonia lanceolata s. l. North taxa and South taxa respectively refer to separate geographic analyses run on plants in northern and southern California, whereas All taxa refers to an analysis including all taxa we treat as part of this group in California. Bold is used to draw the reader's eye to cumulative variation explained by three principal components and to signify those characters with the highest loadings on each of the components.

For the DA of all Californian taxa (eight taxa, n = 108), three linear discriminants were important for maximal differentiation of species (LD1 = 0.7394, LD2 = 0.2212, LD3 = 0.0389) explaining 99.9% of the total variation in the morphological dataset. Factors that best discriminated among taxa were the length (-5.417) and width (3.813) of the cauline leaves. The graph of optimized discriminants (Fig. 3K) indicates that all factors taken in combination separate discrete clusters of samples for most taxa. Essentially, these results are comparable to those of the PCA in showing distinct clustering of specimens assignable to each taxon we circumscribed with the majority of overlap occurring among species in different portions of California (i.e. northern or southern California) rather than among species within the same geographic area.

Additionally, taxonomic predictions from the DA were fairly accurate (78.7%, 85 of 108 correctly predicted; Table 3) but indicate that classification accuracy based on the morphological characteristics we measured varies among taxa (C. lanceolata, 12/24 = 50% correct; C. obovata, 11/12 = 91.7% correct; C. panamintensis, 6/6 = 100% correct; C. peirsonii subsp. bernardinus, 12/12 = 100% correct; C. peirsonii subsp. californacis, 9/12 = 75% correct; C. peirsonii subsp. peirsonii, 29/ 29 = 100% correct; C. peirsonii subsp. yorkii, 3/7 = 42.9% correct; C. serpenticola, 3/6 = 50% correct). For the northern California regional DA (three taxa, n = 42), cauline leaf length (-4.108) discriminates best among taxa and predictive accuracy was high (C. obovata, 12/12 = 100% correct; C. lanceolata, 21/24 = 87.5% correct) for all but one taxon, C. serpenticola (3/6 = 50% correct). In comparison, for the southern California regional DA (five taxa, n = 66), cauline leaf width (5.646) and length (-5.470) discriminate best among taxa and predictive accuracy was also high (C. panamintensis, 6/6 = 100% correct; C. peirsonii subsp. bernardinus, 12/12 = 100% correct; C. peirsonii subsp. californacis, 11/12 = 91.7%correct; C. peirsonii subsp. peirsonii, 29/29 = 100% correct) for all but one taxon, C. peirsonii subsp. yorkii, (5/7 = 71.4% correct). It should be noted that these prediction accuracies all increase (or stay the same) in individual regional analyses compared with the analysis for all Californian taxa (Table 3). This result suggests that actual misclassification rates in the field using the treatment provided below should be lower than we observed using the DA for four continuous morphological characters. Furthermore, additional discontinuous characters facilitate easy recognition of the two taxa identified least successsfully by the DA. The alternate leaf arrangement and venation of C. serpenticola readily distinguish it from C. lanceolata and the limited betalain pigmentation of C. peirsonii subsp. yorkii is diagnostic for that taxonwith respect to othermembers of the C. peirsonii complex. Results of the x2 test from the predictive DA of all Californian taxa strongly reject the null hypothesis of statistical independence between taxon and measured morphological characters for samples included in this study (All Californian taxa: X-squared = 435.2407, df = 49, p < 0.001; North Californian taxa: X-squared = 47.9062, df = 4, p < 0.001; South Californian taxa: X-squared = 228.8473, df = 16, p < 0.001).As with the PCA discussed above, these findings suggest both strong morphological differentiation on the regional scale and convergence across California.

The inferences made from the data we collected have clear taxonomic implications for California: 1) recognition of species status for a southern Californian taxon formerly subsumed within C. lanceolata s. l. and further delimitation of four subspecies, three of these being newly described taxa, 2) reinstatement with modified circumscriptions for previously recognized taxa in California treated as C. lanceolata by Miller and Chambers (2006), and 3) circumscription of two newly discovered species from the Klamath-Siskiyou region of California and Oregon and the Transmontane habitats of California and Nevada, respectively. Using a taxonomic concept originally proposed by Cronquist (1978), we argue that observed and statistically significant morphological differences are sufficient to recognize populations in southern California and southern Nevada discussed by Stoughton and Jolles (2013) as two new species of Claytonia, C. panamintensis, and C. peirsonii. Drawing from discourse concerning species concepts communicated by several authors over the years (e.g. Baum 2009; Johnson and Cairns-Heath 2010; Stoughton et al. 2014), information about the natural histories included below and molecular divergence included above provides additional support for the recognition of four morphologically and genetically diagnosable subspecies of C. peirsonii endemic to the North American Desert Southwest.

Table 3.

Results of the descriptive discriminate analyses for Californian Claytonia lanceolata s. l. North taxa and South taxa, respectively, refer to separate geographic analyses run on plants in northern and southern California, whereas All taxa refers to an analysis including all C. lanceolata s. l. taxa we treat as part of this group in California. Percent of incorrect samples predicted as geographic impossibility refers to situations when northern Californian taxa are misclassified as southern Californian taxa, or vice versa. Measurements denoted by ¹ indicate group means.

We here present a new taxonomic circumscription of tuberous, perennial Claytonia in California based on these results and information included in the Taxonomic Treatment below. We cannot overemphasize that the diagnostic characters presented in the current treatment are best observed from several plants while in the field (Stoughton and Jolles 2013). Claytonia specimens that will be most useful for subsequent taxonomic work should include an image of a live plant and should be made by quickly and carefully pressing plants in the field. Finally, we invite researchers who locate and collect these plants to communicate with us (images and vouchers ideal) toward further advancing our knowledge of this challenging group.

Taxonomic Treatment

1. Claytonia panamintensis T. R. Stoughton, sp. nov.—TYPE: U. S. A. California: Inyo Co., upper Johnson Canyon, Panamint Mountains, 6 April 2013, T. R. Stoughton 1711 (holotype: RSA!; isotypes: UCR!, JEPS!).

Perennial; tuberous root 0.5–2.5 cm in diameter, globose, sometimes with a distinctly narrowed area at tuber apex where the aerial stems emerge; rhizomes absent. Stems (7-)11–14 (-20) cm long; foliage dark green adaxially, often at least weakly beet-red abaxially. Basal leaves absent to 1 (rarely more than 1), absent at flowering time, up to 15 cm long, petioles tapered or sometimes indistinct from blades; blades ca. 25 × ca. 2.5 mm, lanceolate, apex acute to acuminate. Cauline leaves 2–4, opposite at least proximally, 1-nerved, (20-)30–60(-80) × (2.5-)4–9 mm, elliptic to oblanceolate, distinctly petiolate, petioles generally ≥ 5 mm long. Inflorescences 1–3, terminal and often also axillary, pedunculate, unibracteate, bracts 1–3 mm long, membranous. Flowers 16–20 mm in diameter, developing buds nodding; sepals 2–4 × 1–4 mm; petals 8–10 × 2.5–4.5 mm, white, generally with at least faintly-colored pink veins, generally with a yellow-colored blotch at base of blade above claw, apices entire to irregularly toothed distally. Ovules 6. Seeds unknown. Chromosome number unknown. Figures 3D, 4C.

Distribution—Found in southeastern California in the Panamint Mountains (Fig. 1) east at least to the Spring Mountains of southern Nevada, on north-facing, stony and talus slopes comprised of primarily (meta)sedimentary substrates (e.g. marble, shale/slate, sandstone) mixed with decomposing organic material. Found most often in openings of Pinus-Juniperus and Quercus woodland habitats; 1,825 m +/- ca. 500 m elevation.

Notes—Claytonia panamintensis is only superficially similar to the broad interpretation of C. lanceolata by Miller and Chambers (2006). Plants treated here as C. panamintensis are distinctly petiolate and therefore should not be confused for the sessile-leaved C. lanceolata s. s. Claytonia panamintensis differs from the entire C. peirsonii species complex (discussed below) in having nodding flower buds (as opposed to the spreading to erect buds present in C. peirsonii) and an elongate peduncle. Claytonia panamintensis also differs from C. lanceolata s. l. in California by its ecological setting (C. panamintensis is associated with transmontane habitats and a mixture of variable sedimentary or metasedimentary substrates), shape, arrangement, and venation of the cauline leaves, longer stems, and a racemose inflorescence that retains an elongate peduncle with little to no internodal elongation among pedicels by the time of fruiting.

Given its distinctly petiolate cauline leaves, betalain pigmentation on abaxial surfaces of the foliage, and near sympatric distribution, C. panamintensis may be easily confused with C. umbellata S. Watson in the Great Basin. Claytonia panamintensis can be readily distinguished from C. umbellate because it lacks a taproot extending from below the tuber (consistently present in C. umbellata), it tends to have larger (longer) leaves and flowers than C. umbellata, and the shape of their leaf blades are different. Tuberous, perennial Claytonia plants in the Panamint Mountains (i.e. C. panamintensis) were previously identified as C. lanceolata (Miller and Chambers 2006) or C. lanceolata var. peirsonii (Stoughton and Jolles 2013; i.e. C. peirsonii in this treatment), the latter ofwhich has sessile to only short-petiolate or indistinctly petiolate leaves (Stoughton and Jolles 2013). The distribution (Fig. 1), habitat, DNA data (Fig. 2), and morphology (Figs. 3D, 4C) of these Claytonia in the Panamint Mountains indicate that C. panamintensis is a distinct species worthy of recognition. Claytonia panamintensis should be sought out in additional areas throughout southeastern California, particularly in (meta)sedimentary habitats of the Pinus-Juniperus and Quercus belts in mountainous areas adjacent the Panamint Mountains, and east toward the Spring Mountains in southern Nevada. Collections held at CAS/DS, NSMC, and UNLV from the Bristol Range and Mormon Mountains in Lincoln Co., Nevada, approach C. panamintensis in gross morphology but may represent a distinct taxon and therefore are not included as representative here. More field and molecular work are needed to address these outstanding questions.

Fig. 4.

Illustrations of Californian plants. A. C. peirsonii subsp. californacis. B. C. peirsonii subsp. yorkii. C. C. panamintensis. D. C. serpenticola. E. C. obovata. F. C. lanceolata.

The specific epithet, panamintensis, refers to the type locality for the species in the Panamint Mountains. The suggested common name for the species is Panamint spring beauty.

Representative Specimens Examined—U. S. A. California, Inyo Co.: Panamint Mountains, upper Johnson Canyon, 17 Mar 2014, T. Stoughton 1844 (RSA, JEPS); Johnson Canyon, 1.8 km NW of Hungry Bills Ranch, 23 May 2001, D. York 2630 (RSA812089). Nevada, Clark Co.: Spring Mountains, 1.75 mi E of Red Rock Summit, 19 Apr 1974, J. Holland 141 (UNLV7576); Northern Red Rock Canyon Recreation Area, 28 Apr. 1990, A. Leary 3938 (UNLV32952); Wilson Pass Rd, NE slope below ridge, 2 Apr 1985, A. Pinzl 6409 (NSMC8104, UNLV19821).

2. Claytonia serpenticola T. R. Stoughton, sp. nov.—TYPE: U. S. A. California: Trinity Co. (near Humboldt Co. line), North Coast Ranges, Lassics Botanical Area, north slope of Signal Peak (Mount Lassic), 28 May 2013, T. R. Stoughton 1759. (holotype: RSA!; isotype: JEPS!).

Perennial; tuberous root 0.5–2.0 cm in diameter, globose; rhizomes absent. Stems (1.5-)3–7(-10) cm long; foliage dark to light green ab/adaxially. Basal leaves absent to 1 (rarely more than 1), absent at flowering time, (2-)7–12 cm long, petioles tapered or indistinct from blades; blades (1-)2–5 × (1-) 2–5 mm, linear-lanceolate to oblanceolate, apex acute to acuminate. Cauline leaves 2–4, alternate to subopposite, 1-nerved, (10-)12–34(-45) × (1-)2–4.5(-8) mm, linear to lanceolate or narrowly elliptic, sessile, base of blade sometimes attenuate but not distinctly petiolate. Inflorescences 1–3, terminal and often also axillary, pedunculate, unibracteate, bract 1–2 mm long, membranous. Flowers 10–16mm in diameter, developing buds nodding to spreading; sepals 3–6 × 1–5 mm; petals 5.5–8 × 1.5–4 mm, white to pinkish, with faintly colored pink veins, generally with a yellowcolored blotch at base of blade above claw, apices entire. Ovules 6. Seeds 1–6, ca. 2 mm long, smooth to very weakly tuberculate; elaiosomes generally < 1 mm, + / - equal to funicular notch at maturity. Chromosome number unknown. Figures 3B and 4D.

Distribution—Found in the Klamath-Siskiyou region of northwestern California in the North Coast Ranges up to the Siskiyou Mountains in southern Oregon (Fig. 1), on xeric, stony slopes (generally north-facing) comprised of mafic substrates (e.g. gabbro, peridotite, serpentinite), sometimes mixed with sedimentary rocks (e.g. shale), and in soils derived from these parent materials. Most often found in openings of mixed-conifer and subalpine forest habitats from 1,000 to ca. 2,450 m.

Notes—Claytonia serpenticola is morphologically similar to the broad interpretation of C. lanceolata by Miller and Chambers (2006). However, C. serpenticola differs in its ecological setting (it is primarily associated with mafic rocks), by its alternate leaf arrangement, and by the shape, number, and venation of its cauline leaves. At least in northern California, C. serpenticola is diagnosable by its propensity to have inflorescences that emerge from the axils of its cauline leaves. This diagnostic feature has been observed to occur only rarely in populations of C. lanceolata or C. obovata, and only in areas where these taxa grow in sympatry with C. serpenticola (T. Stoughton pers. obs.). Some morphological variation was observed within and among populations identified here as C. serpenticola, perhaps reflecting morphological evidence of hybridization with close congeners, particularly in areas of sympatry with C. obovata. Nevertheless, C. serpenticola is a diagnosable taxon worthy of recognition at the rank of species. Claytonia serpenticola is morphologically most similar to C. peirsonii subsp. bernardinus, based on the results of our morphological study, sharing lance-linear leaves with a single vein (and a propensity for development of axillary inflorescences), but the two do not co-occur geographically and the latter typically has purple pigmentation on the abaxial surfaces of its leaves. The distinctive, beet-red colored pigmentation of C. peirsonii is absent from the abaxial surfaces of the leaves of C. serpenticola. Our molecular results also suggest these two are highly divergent at the nrITS gene region (Fig. 2), if they are at all closely related.

The discovery of a distinct lineage of Claytonia in the Klamath-Siskiyou region occurred during preliminary molecular analyses when DNA sequences for C. lanceolata generated by M. Hershkovitz (unpubl. data on GenBank) did not resolve as sister to other sequences acquired by O'Quinn and Hufford (2005). Upon examining the specimen sampled by M. Hershkovitz, B. W. Klipfel 318 (HSC41230), the diagnostic feature of this lineage became immediately evident: these plants have alternate cauline leaves and axillary (as opposed to strictly solitary/terminal) inflorescences. The more common condition for C. lanceolata s. l. is to have an opposite pair of cauline leaves and Pursh (1814)mentions specifically in the original description that C. lanceolata has a solitary inflorescence.

The alternate phyllotaxis of C. serpenticola is significant not only because it is an easy morphological character to use and makes the taxon an outlier for the C. lanceolata species complex, but also because the leaf blades of C. serpenticola otherwise might be considered to resemble in shape and size those of geographically co-occurring C. lanceolata (see treatment below). Claytonia serpenticola differs also from C. lanceolata by having single-nerved cauline leaves with pinnate venation. In contrast, C. lanceolata has tri-nerved cauline leaves with (sometimes inconspicuous) parallel venation.

The specific epithet, serpenticola, refers to a propensity for this species to be associated with serpentine soils. The suggested common name for the species is Serpentine spring beauty.

Representative Specimens Examined—U. S. A. California: Humboldt Co., 3/4 mi N of Alder Spring near Lassics Lookout, 30 Jun 1982, D. K. Imper s.n. (HSC79323); Horse Mountain, 8 May 1974, B. W. Klipfel 318 (HSC41230); Horse Mountain, 30 May 1975, B. W. Klipfel 648 (HSC44425); 2 miles south of intersection with Forest Service Road 5N09, 2 Jun 2008, C. Witte 663 (HSC98784); Shasta Co., ca. 2 miles N of Slate Mountain, 22 May 1993, D. W. Taylor 13465 (JEPS101969); Siskiyou Co., E slope of Kangaroo Mtn., above Toehead Lake, 3 Jul 1972, J. P. Smith 5708 (HSC27336); Ridge between Durney Lake and Little Crater Lake, 17 Apr 1976, J. Whipple 1030 (HSC92978); Trinity Co., Black Lassic along W side, 13 Jun 1972, D. E. Anderson 5546 (HSC31833);Nside of Red Lassic, 14 Jun 1972, D. E. Anderson 5583 (HSC31422); Big Flat, 6 Jun 1967, W. J. Ferlatte 502 (HSC4434); E side of Packers Peak, 28 Jun 1967, W. J. Ferlatte 544 (HSC4433); Head of Union Creek, 7 Jul 1967, W. J. Ferlatte 654 (HSC4435); Swift Creek at SE end of Parker Meadow, 28 May 1979, D. L. Goforth 144 (HSC64957); Red Lassic, 20 May 1973, T. W. Nelson 914 (HSC32862); Along Mule Ridge SE of Red Lassic, 30 May 1973, T. W. Nelson 992 (HSC32904); E side of Black Lassic along S side of branch off Shanty Creek, 23 May 1976, T. W. Nelson 2547 (HSC39535); Shores of Diamond Lake, 28 Jun 1974, J. O. Sawyer 2635 (HSC38970); Upper Bowerman Meadows on trail to Lake Anna, 14 Jun 1981, J. O. Sawyer 4132 (HSC75606); Big Bear Lake trail, N of Eagle Creek Campground, 21 Apr 1973, J. P. Smith 6241 (HSC28802); Mt. Eddy trail above Deadfall Lakes, 31 May 2014, T. R. Stoughton 1899 (RSA, JEPS); Lassics Botanical Area, north slope of Signal Peak (Mount Lassic), 1 Jun 2014, T. R. Stoughton 1900 (RSA, JEPS); Deadfall Creek drainage in the vicinity of Deadfall Lakes, 11 Jun 1976, J. Whipple 959 (HSC92977); Oregon: Jackson Co., Observation Peak, Siskiyou Mtns, 16 Jun 1994, K. L. Chambers 5816 (OSC209262); Red Mountain W. from Siskiyou Peak, 14 June 1899, J.B. Leiberg 4068 (ORE29125 at OSC); Observation Peak, Siskiyou Mtns, 28May 2014, T. R. Stoughton 1896 (RSA, JEPS).

3. Claytonialanceolata Pursh, Flora Americae Septentrionalis: 175. 1814.—TYPE: U. S. A. Idaho: Clearwater Co., headwaters of Kooskooski, 27 Jun 1806, Lewis s.n. (holotype: PH, [photos] POM! OSC!).

Claytonia caroliniana Michaux var. sessilifolia Torrey, Pacific Railroad Survey Reports 4(5): 70. 1857. Claytonia lanceolata Pursh var. sessilifolia (Torrey) A. Nelson, Bulletin of the Torrey Botanical Club 27: 259. 1900. Claytonia sessilifolia (Torrey) Henshaw, Mountain Wild Flowers of America: 28. 1906.—TYPE: U. S. A. California: Sierra Co., Downieville, 2 May 1854, Bigelow s. n. (holotype: NY!; isotypes: GH, US).

Claytonia chrysantha Greene, Leaflets of Botanical Observation and Criticism 2: 45. 1910. Claytonia lanceolata f. chrysantha (Greene) St. John, Research Studies of the State College of Washington 1: 97. 1929. Claytonia lanceolata subsp. chrysantha (Greene) Ferris in Abrams, Illustrated Flora of the Pacific States 2: 122. 1944. Claytonia lanceolata var. chrysantha (Greene) C. L. Hitchcock, Vascular Plants of the Pacific Northwest 2: 229. 1964. Claytonia caroliniana var. chrysantha (Greene) Boivin, Phytologia 16: 323. 1968.—TYPE: U. S. A. Washington: Whatcom Co., Mt. Baker, E side of Deming Glacier, moist sandy slopes, 5,500 ft, 6 Aug 1909, M. W. Gorman 2809 (holotype: NDG; isotypes: ORE! [at OSC], WS).

Claytonia lanceolata var. idahoensis R. J. Davis, Brittonia 18: 294. 1966.—TYPE: U. S. A. Idaho: Bannock Co., at Justice Park, 14 mi S of Pocatello, R. J. Davis 6029 (holotype: [photo] ISC!; isotypes: ASU, BRY, DS!, NY!, UTC).

Perennial; tuberous root 0.2–2.5 cm in diameter, globose or sometimes with a narrowed area at tuber apex where the aerial stems emerge; rhizomes absent. Stems (3-)4.5–13.5(-16) cm long; foliage generally green throughout, sometimes beet-red on abaxial surfaces in the Sierra Nevada. Basal leaves absent to 1 (rarely to 6), generally absent at flowering time, 3.5–9.5 cm long, petioles tapered or indistinct from blades; blades 10–35 × (1.5-)3.5–7(-12) mm, generally lanceolate to elliptic to oblanceolate, apex acute to obtuse. Cauline leaves 2(3), opposite, 3-nerved (sometimes lateral veins indistinct), (10-)17–43(-70) × (1.5-)3–7(-12) mm, lance-linear to lanceovate to narrowly elliptic, sessile (rarely winged-petiolate), blades sometimes widest above middle but generally not appearing petiolate. Inflorescences generally 1(2), terminal (axillary inflorescences rarely present), pedunculate, unibracteate, bracts 1–5 mm long, membranous. Flowers 10–18 mm in diameter, developing buds nodding; sepals 1.5–6 × 1–4 mm; petals 5–9 × 1.5–4.5 mm, white to pinkish, with faintly to strongly-colored pink veins, with a yellow-colored blotch at base of blade above claw, apices strongly to weakly notched or entire. Ovules 6. Seeds 1–6, 1.5–2 mm long, smooth; elaiosomes < 1 mm, + / - equal to funicular notch at maturity. Chromosome number unknown, likely n = 8 based on reports by Lewis (1967) made from samples collected in Washington (Kittitas Co.) and Oregon (Baker Co.). Figures 3C, 4F.

Distribution—Found from Northern California in the Central Sierra Nevada and Klamath Regions (Fig. 1), northward To Canada (Alberta, British Columbia, and Saskatchewan) primarily on the east side of the Cascade Ranges, and eastward in the mountains to Colorado, New Mexico, and Wyoming. In California, on mostly north-facing, stony slopes comprised of more alkali-rich substrates (e.g. granite, rhyolite) and in soils derived from a combination of these parent rocks and some organic material. Most often found in protected, mesic openings of mixed conifer or subalpine forest habitats from ca. 600 to ca. 2,750 m.

Notes—Our concept of C. lanceolata is morphologically similar to C. peirsonii subsp. bernardinus and C. serpenticola, but differs in the shape, venation, and arrangement of its leaves, and by its ecological setting (C. lanceolata is primarily associated with mesic, montane and subalpine habitats). It differs also by its larger overall plant and smaller flower size, often (at least weakly) notched petals, and by having an elongate inflorescence that often exhibits elongation of the peduncle and internodal elongation among pedicels by the time of fruiting.

Claytonia lanceolata is most likely to bemistaken for C. serpenticola in the field because the two taxa tend to have lance-linear cauline leaf blades that are similar in size. The alternate phyllotaxis and propensity for axillary infloresence development in C. serpenticola make it distinguishable from C. lanceolata in most cases. However, these features can be complicated in C. serpenticola by aborted axillary inflorescence development, resulting in solitary inflorescences, and by the occasional presence of cauline leaf pairs appearing to be sub-opposite rather than distinctly alternate. In California, C. lanceolata has a geographic range that is primarily east of C. serpenticola. Aside from phyllotaxis, the primary character distinguishing these two taxa is the venation of the cauline leaf pair: C. lanceolata is tri-nerved (sometimes appearing as single-nerved without use of a microscope or hand lens) whereas C. serpenticola is distinctly single-nerved. Claytonia lanceolata also does not appear to occur on the same substrate types as C. serpenticola, at least in California.

Discovered in the early years of territorial and botanical exploration, C. lanceolata var. sessilifolia was originally described as a variety of the eastern C. caroliniana from material collected in the Sierra Nevada during surveys for the Pacific Railroad. This taxon was originally characterized as having single-nerved cauline leaves, but closer inspection of this character reveals that the tri-nerved nature of the cauline leaf pair is often obscured in dried herbarium specimens because the midrib is thicker than the lateral veins in Californian populations of C. lanceolata (especially those in the Sierra Nevada). After being transferred to C. lanceolata as a variety, C. sessilifolia (Torr.) Henshaw was treated as a distinct species only a few years later but venation of the cauline leaf pair was not further addressed for the basionym. Davis (1966) later taxonomically treated C. lanceolata var. sessilifolia (Table 1), but he apparently confused some specimens of the tri-nerved C. obovata with this taxon. For example, Wiggins 12393 (DS327493) collected from Siskiyou Co., California, was listed as a representative specimen for C. lanceolata var. sessilifolia by Davis (1966). The two are easily distinguished except in cases where diagnostic characteristics have been overlooked or obscured by their poor preservation on herbarium specimens. The cauline leaf blades of C. obovata are generally less than three times longer than wide and obtuse to rounded at the apex whereas those of C. lanceolata in California are generally greater than three times as long as wide and have acute to acuminate (rarely obtuse) apices. Due to the wide range of morphological variation observed in C. lanceolata, and for the lack of molecular resolution or exhaustive sampling of lineage diversity in the species, we conservatively lump infraspecific taxa contained within C. lanceolata until outstanding questions can be addressed in the context of the full distribution of the species and its closest relatives.

Representative Specimens Examined—U. S. A. California: Alpine Co., Meiss Ridge, 24 May 2014, T. R. Stoughton 1884 (RSA, CAS, JEPS, UCR, UNR); El Dorado Co., Echo Lakes Divide, 7 Jul 1925, F. W. Peirson 6314 (RSA65732); Humboldt Co., Near Bret Hole, 19 Jul 1980, M. A. Baker 3035 (HSC73821); Grogan's Hole near N Trinity Mountain, 27 May 1973, B. W. Klipfel 30 (HSC48838); S of Trinity Summit Station, 15 Jun 1971, J. O. Sawyer 2385 (HSC24076); Trinity Summit Area, Trinity Alps, 30 May 2013, T. R. Stoughton 1762 (RSA, JEPS); Lassen Co., 0.5 mi W of Predonyer Pass, Highway 36, 23 May 1952, M. S. Baker 12429 (RSA74140);Madera Co., Iron Lake Ridge, 2 Aug 1938, M. V. Hood s.n. (RSA464475); 1.4 km NE of Iron Mountain, 10 Jul 1998, D. York 2191 (RSA696304); Modoc Co., Ca. 1 km NW of Bald Mountain, E side of the Warner Mountains, 24 Jun 2002, B. Bartholomew 8839 (CAS1032034); Deep Creek, 15 Jun 1946, A. M. Alexander 4773 (POM275200A); Sierra Co., Nine woods near Yuba Pass, 3 Jul 1927, R. C. Bacigalupi 1601 (POM161630); Gold Lake, 22 Jun 1927, H. A. Barker 720 (RSA8013); 15 mi west of Portola, 5May 1929, M. Canby 128 (POM220759); Siskiyou Co., Baldy Mountain ca. 6 mi W of Happy Camp, 2 Jul 1952, P. A. Munz 17896 (RSA78818); Ridge west of Baldy Mountain, 6 mi W of Happy Camp, 2 Jul 1952, P. A. Munz 17908 (RSA78830); Along Happy Camp-O'Brien Rd above Louse Creek, May 1977, P. Stekel 239B (HSC48274); Poker Flat, 9 Jul 1959, R. Van Deventer s. n. (HSC61811, HSC61812); Tuolumne Co., Eagle Peak, 20 Jun 1936, R. F. Hoover 1354 (UC766152); Canyon opening N head of Emigrant Lake, 22 Jul 1941, R. F. Hoover 5513 (UC765758); Oregon: Douglas Co., Nof Crater Lake National Park, 15 Jul 2012, D. Jolles 512 (RSA, JEPS); N of Crater Lake National Park, 26 May 2014, T. R. Stoughton 1890 (RSA, JEPS); Trail to Abbott Butte Lookout, 27 May 2014, T. R. Stoughton 1895 (RSA, JEPS); Klamath Co., near Crater Lake National Park, 16 Jul 2012, D. Jolles 515 (RSA, JEPS); Crater Lake National Park, 29 Jun 1962, J. Rockwell 113 (HSC17656); near Crater Lake National Park, 26 May 2014, T. R. Stoughton 1889 (RSA); Linn Co., Monument Peak, 17 Jun 1948, H. M. Gilkey s. n. (OSC65157); Ca. 10 mi W of the junction of FS Rd 11 and State Highway 22, 2 Jun 2009, R. G. Halse 7635 (OSC223100).

4. Claytonia obovata Rydberg, North American Flora 21: 299. 1932.—TYPE: U. S. A. Oregon: Douglas Co., Abbott Butte, 1,600 m, 6 July 1899, J. B. Leiberg 4262 (holotype: [photo] US!; isotype: ORE! [at OSC]).

Perennial; tuberous root 0.7–2(-3) cm in diameter, globose or sometimes with a narrowed area at tuber apex where the aerial stems emerge; rhizomes absent. Stems (2-)3.5–8.5(-13) cm long; foliage dark green adaxially, beet-red abaxially. Basal leaves absent to 1 (rarely more than 1), absent at flowering time, 6–11(-16) cm long, petioles tapered or indistinct from blades; blades 10–25(-40) × (2-)4–9 mm, widely lanceolate to lance-ovate or narrowly elliptic, apex acute to obtuse. Cauline leaves 2(3), opposite, 3-nerved (sometimes more), (5-)7.5–22 (-30) × (3-)4–9.5(-15) mm, blades lance-ovate to elliptic to ovate or obovate, sessile, base generally asymmetric, rarely narrowed at base but not distinctly petiolate. Inflorescences 1 (2), terminal (rarely also axillary), sessile to short-pedunculate, unibracteate, bracts 1–3 mm long, membranous. Flowers 10–15 mm in diameter, developing buds spreading to ascending; sepals 2–6 × 1–5 mm; petals 5–7.5 × 1.5–5 mm, white to pinkish, with faintly colored pink veins, generally with a yellow blotch at base of blade above claw, apices entire. Ovules 6. Seeds 1–6, 2–2.5 mm long, smooth; elaiosomes <1 mm, + / - equal to funicular notch at maturity. Chromosome number unknown. Figures 3A, 4E.

Distribution—Found in the Klamath-Siskiyou region of northwestern California in the North Coast Ranges to southwestern Oregon at Abbott Butte (Fig. 1), generally on north-facing, stony and talus slopes comprised of siliciclastic or carbonate-dominated (meta)sedimentary rocks (e.g. greywacke, limestone, shale), or mafic substrates (e.g. gabbro, peridotite, serpentinite), and in soils derived from these parent materials. Most often found in openings of subalpine forest habitats from ca. 1,200 to ca. 2,200 m.

Notes—Claytonia obovata is morphologically similar to the broad interpretation of C. lanceolata by Miller and Chambers (2006), but differs primarily in its ecological setting (C. obovata is primarily associated with cismontane, subalpine habitats) and by the presence of sunken, red veins on the cauline leaves that diverge near the base of the leaf blade and generally converge near the apex. It differs also by the shape of the cauline leaf pair, its smaller overall plant size, and in having a sub-umbellate inflorescence that generally lacks a peduncle and exhibits little to no internodal elongation among pedicels by the time of fruiting. Claytonia obovata most notably resembles C. peirsonii subsp. peirsonii but the two do not co-occur geographically and should not be confused. Claytonia obovata lacks raised (appearing thickened) veins of any kind on the adaxial surfaces of the cauline leaves, which are always present in C. peirsonii subsp. peirsonii.

Davis (1966) did not recognize C. obovata, but described visiting the type locality at Abbott Butte, Oregon, where he encountered plants matching the descriptions of C. obovata as well as C. multicaulis Nelson (currently treated as a synonym of C. multiscapa), C. chrysantha Greene, and C. lanceolata. Essentially, Davis (1966) recounted seeing plants showing a wide range of intermediate characters at Abbott Butte. Davis (1966) mentioned also growing plants in the greenhouse that maintained a broad cauline leaf morphology at flowering and subsequently used his findings to justify not recognizing C. obovata.We interpret Davis' observations differently: the broad leaves of C. obovata are not likely the result of phenotypic plasticity in a polymorphic C. lanceolata, but instead the expression of fixed genetic changes in populations assignable to C. obovata, a closely related species. It is true that the type specimen and live plants from Abbott Butte do not have obovate cauline leaves (i.e. ovate with the narrower end positioned proximally), as Davis (1966) points out, but Rydberg (1932) also cites plants from the North Coast Ranges of California at Hull Mountain in the original description of C. obovata. The plants at Hull Mountain tend to have obovate cauline leaves, although sometimes only weakly so (T. Stoughton pers. obs.). We have been in the field at both Abbott Butte and Hull Mountain (and localities in between, Fig. 1) to observe these tuberous, perennial Claytonia in their native habitats. Davis (1966) gives no indication of visiting other sites where C. obovata grows or seeing material from Hull Mountain in Mendocino Co., California. As for the possible hybrid swarm at Abbott Butte mentioned above, we observed a vast and variable (perplexing!) population of C. lanceolata there and only scattered remnants of a few plants at lower elevations that resemble the type specimen of C. obovata. More than 40 yr later, we did not observe any yellow-flowered plants assignable to C. chrysantha, reported to be present by Davis (1966). We have, however, observed what appears to be natural interspecific hybridization involving C. obovata and C. serpenticola at one sympatric locality near Mount Shasta in northern California, in addition to what appears to be hybridization between C. obovata and C. lanceolata at Abbott Butte. We suspect a similar scenario to what has been observed by Doyle and Doyle (1988) with closely related tuberous Claytonia on the east coast, or possibly a similar situation to what was observed by Stewart and Wiens (1971) with close relatives in the Mountain West, in addition to what Davis (1966) mentioned he observed at Abbott Butte. We do not believe that a low frequency of hybridization with close relatives is grounds for sinking a diagnosable taxon that maintains morphological and genetic coherence across a broad range (T. Stoughton et al. unpubl. data).

We found scant collections in existing herbaria (not listed as representative) but several plants collected from the same general area in the Central Sierra Nevada (Tuolumne Co., Bald Mountain, L.R. Heckard 4741) approach both C. obovata and C. peirsonii in their leaf morphology. We consider these plants to represent a range extension for C. obovata but suspect these populations represent a new, pedunculate subspecies of either C. obovata or C. peirsonii that may be worthy of later taxonomic recognition. Additional field and molecular research are required to address outstanding questions concerning these geographically and morphologically outlying populations.

Representative Specimens Examined—U. S. A. California: Mendocino Co., Anthony Peak, east of Lookout, 14 Jul 1944, M. S. Baker 10808 (CAS378247); Near Monkey Rock, 27 May 2012, H. A. Bartosh 588 (RSA); Windy Point, 0.2 rd mi north of Hull Mountain, 4 Jul 1967, R. Gankin 907 (RSA202422); Below Hull Mountain Lookout, 25 Jun 1969, Hamann & Dearing s. n. (HSC17659); Leech Lake Mountain, northeast of Covelo, 20 Jun 1952, G. D. Hanna & G. C. Gester s. n. (CAS515388); Mt. Hull, Jun 1894, C. A. Purpus 867 (UC82747); Along road to summit, Anthony Peak, 17 Jun 1980, G. L. Smith 5942 (CAS927960); Ridge NWof Hull Mountain, 5 June 1974, S. Smith s. n. (JEPS74662); near Calamese Rock, 26 May 2013, T. R. Stoughton 1751 (RSA); Road to Hull Mtn. S of Monkey Rock, 26 May 2013, T. R. Stoughton 1757 (RSA, JEPS); ridge E of Hull Mtn, 26 May 2013, T. R. Stoughton 1758 (RSA, JEPS); ridge below and E of lookout tower on Anthony Peak, 1 Jun 2014, T. R. Stoughton 1902 (RSA, JEPS); 0.5 mi N of Hull Mountain, 22 Apr 1972, D. Toren s.n. (CAS567326); Forest Road 20n02 between Calamese Rock and Hells Half Acre, 21 Jun 1983, C. R. Wheeler 3429 (CAS911596); Windy Gap, on spur ridge to N of Hull Mountain proper, 24 Jun 1956, H-M. Wheeler s. n. (JEPS14708); Siskiyou Co., 0.5 mi E of Highway 3 on the dirt road heading E at Scott Mountain Pass, 30 May 1981, D. Renwick 162 (HSC73412); Nfacing slopes in the Eddy Creek drainage, 19 Jun 1976, J. Whipple 1068 (HSC92962); 0.3 miles west of Carmen Lake, 28 May 1950, I. L. Wiggins 12393 (DS327493); Tehama Co., Anthony Peak, east side, 14 Jul 1944, J. T. Howell 19764 (CAS322563); Trinity Co., Along the Limestone Ridge Trail 0.5 mi S of Rattlesnake Lake, 29 May 1974, E. Carter 773 (CAS1129584); 0.5 mi W of Cory Peak, 29 May 2014, T. R. Stoughton 1897 (RSA, JEPS); hills on S side of upper Deadfall Lakes, 30 May 2014, T. R. Stoughton 1898 (RSA, JEPS); Oregon: Curry Co., Chetco Divide Trail, between Red Mountain and Chetco Peak, 5 May 1980, V. Stansell s. n. (OSC158534).

5. Claytonia peirsonii (Munz & Johnston) T. R. Stoughton, stat. et comb. nov.—Claytonia lanceolata Pursh var. peirsonii Munz & Johnston, Bulletin of the Torrey Botanical Club 49: 352. 1923. Claytonia caroliniana var. peirsonii (Munz & Johnston) Boivin, Phytologia 16: 323. 1968.—TYPE: U. S. A. California: San Bernardino Co., near Kelley's Cabin, Ontario Ridge, San Antonio [San Gabriel] Mountains, 29 May 1920, F. W. & Mabel Peirson s.n. (holotype: POM!; isotype: NY!).

Distribution—Found in montane to transmontane habitats of the Transverse Ranges (San Gabriel and San Bernardino Mountains) to the southern Sierra Nevada in southern California (Fig. 1), generally on north-facing, stony and talus slopes comprised of igneous, sedimentary, and metamorphic rocks, and growing in soils derived from these parent rock materials. Most often found in openings of subalpine forest down to pinyon-juniper and oak woodland habitats from ca. 1,350 to ca. 2,600 m in elevation.

Notes—Claytonia peirsonii, which we also refer to as the C. peirsonii species complex, is a highly variable taxon that includes multiple unnamed lineages (Figs. 3E–H, 4A–B) discussed in some detail by Stoughton and Jolles (2013). It was originally described by Munz and Johnston (1923) as a variety of C. lanceolata based upon only a few specimens, and persisted through multiple taxonomic re-circumscriptions of C. lanceolata until it was subsumed as a synonym (along with all other varieties of C. lanceolata in California) by Chambers (1993).

We maintain a conservative taxonomic approach which recognizes the collective of southern populations identified as C. lanceolata var. peirsonii by Stoughton and Jolles (2013), excluding those treated as C. panamintensis (see above), as a unique and variable lineage recognized here at the rank of species with at least four easily recognizable subspecies. Based on evidence presented in this paper, we assign for the first time additional infraspecific ranks to populations assignable to C. peirsonii in the southern Sierra Nevada and San Bernardino Mountains. The nominate taxon is the most widespread and morphologically polymorphic of all of the subspecies and is known from the most populations, all of which are restricted to the San Gabriel Mountains. As reported by Stoughton and Jolles (2013), plants of this species flower early in the spring months (February–May) and therefore may be present but still unaccounted for in the conifer and oak belts of the higher mountains in the eastern Mojave Desert (i.e. the Clark, Kingston, and New York Mountains) where in-depth floristic studies are lacking.

5a. Claytonia peirsonii subsp. peirsonii

Perennial; tuberous root 0.3–2.1 cm in diameter, globose, sometimes with a narrowed area at tuber apex where the aerial stems emerge; rhizomes absent. Stems (2.5-)5–18(-25) cm long; foliage dark green adaxially, generally with heavy, beet-red coloration abaxially. Basal leaves absent to 2 (rarely more than 2), absent at flowering time, (8-)10–15(-19) cm long, petioles tapered or indistinct from blades; blades 10–35 × (2-) 5–7 mm, elliptic to oblanceolate, apex acute to obtuse. Cauline leaves 2–4, opposite at least proximally, 1-nerved (sometimes more than one but lateral veins not all developed fully or converging at leaf apex), (6-)14–35(-42) × (3-)3.5–11.5(-18) mm, blades ovate or lance-ovate to elliptic, sometimes rounded, rarely lanceolate or oblanceolate, generally sessile to distinctly short-petiolate or with blade attenuate into indistinct petiole, base often asymmetric, adaxial surfaces generally with reddish, sunken 1° veins and strongly thickened (appearing raised), green-colored 2° veins. Inflorescences 1–3, terminal and often also axillary, sessile to short-pedunculate, unibracteate, bracts 1–3 mm long, membranous. Flowers 14–22 mm in diameter, developing buds erect to ascending or spreading; sepals 2–7 × 1.5–6 mm; petals 7–11 × 2–6 mm, white with lightly colored pink veins, generally with a yellow to orange-colored blotch at base of blade above claw, apices entire (rarely weakly notched). Ovules 6. Seeds 1–6, 1.5–2.5 mm long, smooth to very weakly tuberculate; elaiosomes generally<1 mm, + / - equal to funicular notch at maturity. Chromosome number unknown. Figure 3E.

Distribution—Found in Southern California in the San Gabriel Mountains (Fig. 1), generally on north-facing, stony and talus slopes comprised of primarily alkali-rich lithologies (e.g. gneiss, granite, schist) mixed with moderate to high amounts of decomposing organic material from the surrounding forest. Most often found in openings and under closed canopy of mixed conifer and subalpine forests down to pinyon-juniper and oak woodland habitats from ca. 1,575 to ca. 2,600 m.

Notes—Claytonia peirsonii subsp. peirsonii is morphologically similar to C. obovata and the broad interpretation of C. lanceolata by Miller and Chambers (2006) in some respects. However, it differs in its ecological setting (C. peirsonii subsp. peirsonii is generally associated with alkali-rich rocks in subalpine and transmontane habitats), betalain pigmentation on the abaxial surfaces of its leaves, and the presence of raised 2° veins and a single (sometimes branched), sunken, red 1° vein on the adaxial surfaces. It differs also by the shape, arrangement, and venation of its cauline leaves, smaller overall plant size, and in having a sub-umbellate inflorescence that exhibits an abbreviated peduncle (often absent) with little to no internodal elongation among pedicels by the time of fruiting. Claytonia peirsonii subsp. peirsonii is also morphologically similar to the other subspecies of C. peirsonii but can be distinguished by the strongly thickened 2° veins on the adaxial surfaces of the leaves (often weak in other varieties, if present), wider cauline leaves, and it does not co-occur geographically with any other subspecies in the San Gabriel Mountains. We have observed a great deal of morphological variation within and among populations identified here as C. peirsonii subsp. peirsonii, particularly among populations in the high elevation, eastern San Gabriel Mountains and those disjunct to the west of Mount Baldy mostly at lower elevations (Fig. 1). Conservatively, we treat these as the same taxon until further morphological and molecular research can be conducted. Claytonia peirsonii subsp. peirsonii is considered to be fairly uncommon and should be sought after in additional areas in southeastern California, particularly in subalpine and transmontane habitats near and in the San Gabriel Mountains.

The subspecific epithet, peirsonii, is the nominate epithet and refers to the original collectors of Claytonia lanceolata var. peirsonii. The suggested common name for the subspecies is Peirson's spring beauty.

Representative Specimens Examined—U. S. A. California: Los Angeles Co., Just below Pacific Crest Trail to the E of Little Jimmy Campground near Windy Gap, 13 May 2012, T. R. Stoughton 1539 (RSA812097, UCR, JEPS); Near the Devil's Punchbowl County Park, 1 Apr 2013, T. R. Stoughton 1706 (RSA812100, UCR, JEPS, CAS); near Devil's Punchbowl County Park, 15 Feb 2014, T. R. Stoughton 1837 (RSA); PCT near Devil's Punchbowl County Park, 15 Feb 2014, T. R. Stoughton 1838 (RSA); San Bernardino Co., Ontario Ridge, 25 Jun 1980, T. P. Krantz s. n. (RSA509347); N aspect of Thunder Mountain, E of ski run, 5 May 1987, O. Mistretta s. n. (RSA505073); 0.5 mi N of Telegraph Peak, 5 May 1987, O. Mistretta s. n. (RSA505075, RSA505076); Head of Delker Canyon along trail, 13 May 1987, O. Mistretta s. n. (RSA505074); Devil's Backbone Ridge, 19 May 1920, F. W. Peirson 2151 (JEPS5840, RSA65672); Mt. Baldy Ski area, Gold Ridge north of Thunder Mountain, below FS Road 7W04, 11 Jun 2001, V. Soza 1274 (RSA679587); N slope of Thunder Mountain on ridges between ski runs, 8 May 2012, T. R. Stoughton 1532 (RSA812094, UCR, JEPS, CAS); In cirque N of Telegraph Peak, 8 May 2012, T. R. Stoughton 1533 (RSA812095, UCR, JEPS); N slope of Timber Mountain, switchbacks along trail to summit, 8 May 2012, T. R. Stoughton 1536 (RSA812096, UCR, JEPS); Ontario ridge trail just W of and below Bighorn Peak near Kelly Camp, 30 May 2012, T. R. Stoughton 1549 (RSA812098, UCR); Thunder Mountain, 18 May 2014, T. R. Stoughton 1883 (RSA).

5b. Claytonia peirsonii subsp. bernardinus T. R. Stoughton, subsp. nov.—TYPE:U. S. A. California: San Bernardino Co., San Bernardino Mountains, north slope of Bertha Ridge (ridge between Bear and Holcomb Valleys), 20 April 2012, T. R. Stoughton 1514 with S. Eliason & E. Williams (holotype: RSA!; isotypes: UCR!, JEPS!).

Perennial; tuberous root 0.6–1.7 cm in diameter, globose; rhizomes absent. Stems (1.5-)2.5–5.5(-8) cm long; foliage dark green adaxially, often at least weakly beet-red abaxially. Basal leaves not observed on specimens. Cauline leaves 2–4, opposite at least proximally, 1-nerved, (7-)13–35 (-45) × (1-)1.5–4(-5) mm, linear to lanceolate, sessile, adaxial surfaces generally with reddish, sunken 1° veins, sometimes with greenish, weakly thickened (appearing raised) 2° veins. Inflorescences 1–3, terminal and often also axillary, sessile to short-pedunculate, unibracteate, bracts 1–3 mm long, membranous. Flowers 13–22 mm in diameter, developing buds erect to ascending or spreading; sepals 2–6 × 1.5–4mm; petals 6.5–11 × 2–5mm, white, with faintly colored pink veins, generally with a yellow-colored blotch at base of blade above claw, apices entire (rarely weakly notched). Ovules 6. Seeds unknown. Chromosome number unknown. Figure 3F.

Distribution—Found in southern California in the San Bernardino Mountains (Fig. 1), on north-facing, stony and talus slopes comprised of carbonate-dominated (meta) sedimentary substrates (e.g. limestone, marble) often mixed with decomposing organic material from the surrounding forest. Mostly found in openings of a mixture of pinyonjuniper and white fir-limber pine associations around 2,350 m in elevation + / - ca. 500 m.

Notes—Claytonia peirsonii subsp. bernardinus is morphologically similar to C. panamintensis, C. serpenticola, and to the broad interpretation of C. lanceolata by Miller and Chambers (2006), but differs generally by its ecological setting (C. peirsonii subsp. bernardinus is associated with variable, carbonated ominated sedimentary substrates in transmontane habitats), betalain pigmentation on the abaxial surfaces of its leaves, and by the presence of raised 2° veins on the adaxial surfaces. It also differs by shape, arrangement, and venation of its cauline leaves, smaller overall plant size, and in having a sub-umbellate inflorescence that exhibits an abbreviated peduncle (often absent) with little to no internodal elongation among pedicels by the time of fruiting. Claytonia peirsonii subsp. bernardinus is probably most easily confused with C. serpenticola, the two sharing similarly shaped cauline leaves, but they do not co-occur geographically and C. serpenticola lacks the betalain pigmentation that C. peirsonii subsp. bernardinus produces on the abaxial surfaces of its leaves. Claytonia panamintensis can be distinguished from C. peirsonii subsp. bernardinus by its nodding flower buds and distinctly petiolate cauline leaves. In contrast, the leaves of C. peirsonii subsp. bernardinus are unambiguously sessile and its flower buds tend to be erect to ascending in orientation during early development. The weakly thickened (appearing raised) secondary veins, if present, and lanceolate to lance-linear shape of the cauline leaf pair in C. peirsonii subsp. bernardinus, make it easily distinguished from all other subspecies of C. peirsonii.

We have observed a great amount of morphological variation within the population identified here as C. peirsonii subsp. bernardinus, but even in sympatric situations C. peirsonii subsp. bernardinus is readily distinguished from C. peirsonii subsp. californacis by the shape of the cauline leaf pair alone. No intergradation among these subspecies has been observed in the San Bernardino Mountains, which is the only place we currently know these taxa to occur.

Stoughton and Jolles (2013) suggested in their review of C. lanceolata that some populations in the San Bernardino Mountains of C. lanceolata var. peirsonii had considerably narrower leaves than plants in the San Gabriel Mountains and elsewhere, which they illustrated for comparison in figure 10 of their paper. In fact, the plant on the far right side of this figure, and the leaf in the lower left corner, serve perfectly for an illustration of C. peirsonii subsp. bernardinus: they were drawn from images of plants at the type locality (Bertha Ridge) for this taxon. The rest of the plants in Stoughton and Jolles' (2013) figure 10 are representative of C. peirsonii subsp. peirsonii and are illustrated from images of plants in the San Gabriel Mountains near (and at) the type locality for C. peirsonii subsp. peirsonii. Stoughton and Jolles (2013) treated all of these populations as conspecific due to their having access to a limited amount of material and the rather ambiguous protologue in which C. lanceolata var. peirsonii was first described (Munz and Johnston 1923). Claytonia lanceolata var. peirsonii was originally diagnosed only by its characteristically ‘subumbellate’ inflorescence architecture and by its southern Californian distribution (Munz and Johnston 1923; Stoughton and Jolles 2013). All of the subspecies of C. peirsonii are unified by their sub-umbellate inflorescence architectures (i.e. having little to no peduncle or internodal elongation among pedicels) and their geographic distribution in the mountains of southern California.

Claytonia peirsonii subsp. bernardinus is morphologically most similar to C. serpenticola (Fig. 3) from northern California and genetically most closely related to C. peirsonii subsp. californacis (Fig. 2) from the San Bernardino Mountains, based upon our current understanding and evidence presented in this manuscript. Claytonia peirsonii subsp. bernardinus is considered to be fairly uncommon (known currently from a single population). It should be sought out in additional areas in southeastern California, particularly in transmontane, carbonate-dominated habitats near and in the San Bernardino Mountains, and in similar habitats in the White Mountains of Inyo Co., based on shared species that occur on carbonate substrates in these two mountain ranges.

The subspecific epithet, bernardinus, refers to the type locality for this subspecies in the San Bernardino Mountains. The suggested common name for the subspecies is San Bernardino spring beauty.

Representative Specimens Examined—U. S. A. California: San Bernardino Co., N slope of Bertha Ridge, ridge between Holcomb & Bear valleys, 22 Apr 2011, T. R. Stoughton 1077 (RSA812091, UCR, JEPS, CAS); N slope of Bertha Ridge, ridge between Holcomb & Bear valleys, 20 Apr 2013, T. R. Stoughton 1717 (RSA812102, UCR, JEPS); N slope of Bertha Ridge, ridge between Holcomb & Bear valleys, 29 Mar 2014, T. R. Stoughton 1850 (RSA).

5C. Claytonia peirsonii subsp. californacis T. R. Stoughton, subsp. nov.—TYPE: U. S. A. California: San Bernardino Co., San Bernardino Mountains, Holcomb Valley, head of Furnace Canyon, 20 April 2012, T. R. Stoughton 1516 with S. Eliason & E. Williams (holotype: RSA!; isotypes: CAS!, UCR!, JEPS!).

Perennial; tuberous root 0.5–1.5 cm in diameter, globose; rhizomes absent. Stems (2-)3–8(-12.5) cm long; foliage dark green adaxially, often at least weakly beet-red abaxially. Basal leaves absent to 1 (rarely more than 1), absent at flowering time, up to 9 cm long, petioles tapered or indistinct from blades; blades 5–25 × 0.5–3 mm, linear-lanceolate to narrowly elliptic, apex acute to obtuse. Cauline leaves 2–4, opposite at least proximally, 1-nerved, (7-)10–23(-30) × (1-)2–5.5(-9) mm, oblanceolate to elliptic (rarely ovate-elliptic), indistinctly to distinctly short-petiolate (rarely appearing sessile), adaxial surfaces generally with sunken 1° veins that are reddish in color, often with greenish, weakly thickened (appearing raised) 2° veins. Inflorescences 1–3, terminal and often also axillary, sessile to short-pedunculate, unibracteate, bract 1–3 mm long, membranous. Flowers 13–22 mm in diameter, developing buds erect to ascending or spreading; sepals 2–6 × 1–4 mm; petals 6.5–11 × 2.5–5.5 mm, white (rarely light pink), generally with faintly-colored pink veins and a pink (sometimes yellow to orange) blotch at base of blade above claw, apices entire. Ovules 6. Seeds unknown. Chromosome number unknown. Figures 3G, 4A.

Distribution—Found in southern California in the San Bernardino Mountains (Fig. 1), on north-facing, stony and talus slopes comprised of carbonate-dominated (meta)sedimentary substrates (e.g. limestone, marble) mixed with decomposing organic material from the surrounding forest. Mostly found in openings of a mixture of pinyon-juniper and white firlimber pine associations around 2,300min elevation + / - ca. 500 m.

Notes—Claytonia peirsonii subsp. californacis is morphologically similar to C. panamintensis and to the broad interpretation of C. lanceolata Pursh by Miller and Chambers (2006). It differs generally by its ecological setting (C. peirsonii subsp. californacis is associated with variable, carbonatedominated sedimentary substrates in transmontane habitats), betalain pigmentation on the abaxial surfaces of its leaves, and by the presence of raised 2° veins on the adaxial surfaces. It differs also by the shape, arrangement, and venation of its cauline leaves, its smaller overall plant size, and in having a sub-umbellate inflorescence that exhibits a short peduncle (often absent) with little to no internodal elongation among pedicels by the time of fruiting. Claytonia peirsonii subsp. californacis is probably most easily confused with other subspecies of C. peirsonii, particularly subsp. peirsonii and subsp. yorkii because of the shape of their cauline leaves, but these do not co-occur geographically and therefore should not be confused. Furthermore, C. peirsonii subsp. yorkii lacks the betalain pigmentation of C. p. subsp. californacis, and C. p. subsp. peirsonii has more strongly thickened (appearing raised) 2° venation on the adaxial surfaces of the cauline leaf pair. Claytonia panamintensis can be distinguished from C. peirsonii subsp. californacis by its nodding flower buds and distinctly petiolate cauline leaves, although the cauline leaves of C. peirsonii subsp. californacis are occasionally distinctly short-petiolate (or indistinctly so). The combination of heavy betalain pigmentation, weakly thickened (appearing raised) 2° veins, and the oblanceolate to elliptic shape of the cauline leaves of C. peirsonii subsp. californacis distinguishes it from all other subspecies of C. peirsonii.

Claytonia peirsonii subsp. californacis was first collected during follow-up surveys in 2012 preceding a paper by Stoughton and Jolles (2013) regarding the discovery of Claytonia lanceolata var. peirsonii in the San Bernardino Mountains. The first author was searching in suitable habitat for new populations of C. lanceolata var. peirsonii in the San Bernardino Mountains one year after it was first collected there on Bertha Ridge (Stoughton and Jolles 2013). It was noted in the field that plants at Furnace Canyon in Holcomb Valley (i.e. those treated here as C. peirsonii subsp. californacis) did not perfectly resemble the plants collected earlier in the day at Bertha Ridge (i.e. those treated here as C. peirsonii subsp. bernardinus). So as not to cloud the taxonomic issue, Stoughton and Jolles (2013) argued for treatment of all populations in southern California and southern Nevada as C. lanceolata var. peirsonii pending further research. All of these plants approach the vaguely described C. lanceolata var. peirsonii, mostly due to the lack of a circumscription complete with measurements (until now). Although it was suspected that plants now referable to as C. peirsonii subsp. californacis were unique the first time they were collected, the molecular and morphological data presented above have fortified the case for recognizing this taxon as a unique subspecies in the C. peirsonii species complex.

Claytonia peirsonii subsp. californacis is morphologically most similar to C. peirsonii subsp. yorkii (Fig. 3) from the southern Sierra Nevada and genetically most closely related to C. peirsonii subsp. bernardinus (Fig. 2) from the San Bernardino Mountains, based on our current understanding and evidence presented here. Claytonia peirsonii subsp. californacis is considered rare, known only from a single population. It should be sought out in additional areas in southeastern California, particularly in transmontane, carbonate-dominated habitats in and adjacent to the San Bernardino Mountains.

The subspecific epithet, californacis, refers to the type locality (Furnace Canyon) for this subspecies in the San Bernardino Mountains. The suggested common name for the subspecies is Furnace spring beauty.

Representative Specimens Examined—U. S. A. California: San Bernardino Co., Head of Furnace Canyon, Apr 2014, R. O'Quinn s. n. with D. Jolles (RSA); Ridge running directly E from the Head of Furnace Canyon, N Holcomb Valley, 20 Apr 2013, T. R. Stoughton 1720 (RSA812103, UCR, JEPS); Ridge running directly E from the Head of Furnace Canyon, N Holcomb Valley, 29 Mar 2014, T. R. Stoughton 1852 (RSA, UCR, JEPS); Ridge running directly E from the Head of Furnace Canyon, N Holcomb Valley, 20 Apr 2014, T. R. Stoughton 1882 (RSA).

5D. Claytonia peirsonii subsp. yorkii T. R. Stoughton, subsp. nov.—TYPE: U. S. A. California: Kern Co., southern Sierra Nevada, north slope of Cross Mountain, 30 March 2013, T. R. Stoughton 1705 with D. Jolles (holotype: RSA!; isotypes: UCR!, JEPS!).

Perennial; tuberous root 0.6–2.5 cm in diameter, globose, occasionally with a narrowed area at tuber apex where the aerial stems emerge; rhizomes absent. Stems (3-)6.5–9.5(-15) cm long; foliage generally green on ab/adaxial surfaces. Basal leaves absent to 1 (rarely more than 1), absent at flowering time, up to 11 cm long, petioles tapered or indistinct from blades; blades 10–15 × 1–2 mm, lanceolate to oblanceolate or narrowly elliptic, apex acute to obtuse. Cauline leaves 2–4, opposite at least proximally, 1-nerved, (4-)14–24(-40) × (2-) 3–7(-8.5) mm, oblanceolate to narrowly elliptic, sessile to indistinctly petiolate (sometimes distinctly short-petiolate), adaxial surfaces of blades generally with sunken primary veins that are reddish in color (pigmentation sometimes lacking) and weakly thickened (raised) secondary veins that are green in color. Inflorescences 1–3, terminal and often also axillary, sessile to short-pedunculate, unibracteate, bract 1–3 mm long, membranous. Flowers 14–20 mm in diameter, developing buds erect to ascending or spreading; sepals 2–7 × 1.5–5 mm; petals 7–10 × 3–6 mm, light pink (sometimes white), generally with faintly-colored pink veins and a pink or yellow to orangecolored blotch (rarely absent) at base of blade above claw, apices entire or appearing weakly notched. Ovules 6. Seeds unknown. Chromosome number unknown. Figures 3H, 4B.

Distribution—Found in California in the transmontane habitats of the southern Sierra Nevada (Fig. 1), on north-facing, stony and talus slopes comprised of alkali-rich, igneous extrusive substrates (e.g. rhyolite) mixed with decomposing organic material from the surrounding forest. Mostly found in openings of a mixture of Pinus sabiniana and Quercus chrysolepis associations from around 1,450 m in elevation + / - ca. 500 m.

Notes—Claytonia peirsonii subsp. yorkii is morphologically similar to C. panamintensis and to the broad interpretation of C. lanceolata by Miller and Chambers (2006). It differs generally by its ecological setting (C. peirsonii subsp. yorkii is associated with alkali-rich, igneous extrusive rocks in transmontane habitats) and by the presence of raised 2° veins on the adaxial surfaces of its leaves. It differs also by shape, arrangement, and venation of its cauline leaves, its smaller overall plant size, and in having a sub-umbellate inflorescence that exhibits an abbreviated peduncle (often absent) with little to no internodal elongation among pedicels by the time of fruiting. Claytonia peirsonii subsp. yorkii is probably most easily confused with other subspecies of C. peirsonii, particularly subsp. peirsonii and subsp. californacis, because of the shape of their cauline leaves. Claytonia peirsonii subsp. yorkii can be readily distinguished from these two by weak to lacking betalain pigmentation on the abaxial surfaces of its leaves. All other subspecies of C. peirsonii have moderate to heavy betalain pigmentation on the abaxial surfaces of their leaves. Claytonia panamintensis can be distinguished from C. peirsonii subsp. yorkii by its nodding flower buds and distinctly petiolate cauline leaves, although the cauline leaves of C. peirsonii subsp. yorkii are sometimes (indistinctly) short-petiolate.

Claytonia peirsonii subsp. yorkii was first collected by Dana York in 1998 (D. York 2079, RSA), after which York collected it again (at the same and only known location) two more times in 2001 (D. York 2530, 2555). York noted in the field that plants collected at Cross Mountain in the southern Sierra Nevada (i.e. those treated here as C. peirsonii subsp. yorkii) were growing across a steep slope in loose talus of rhyolite and putatively numbered in the hundreds. York identified these plants as C. lanceolata var. peirsonii. Also discussed by Stoughton and Jolles (2013), these plants lack an expanded peduncle and internodal elongation among pedicels. They do not co-occur geographically with but are nearly a perfect match for the San Bernardino Mountain endemic C. peirsonii subsp. californacis, aside from a few minor subtleties in overall morphology and ecology. The two subspecies are known from different substrates: C. peirsonii subsp. yorkii is known from collections only on rhyolite while C. peirsonii subsp. californacis is only known from collections on marble-dominated scree slopes. Claytonia peirsonii subsp. yorkii is considered to be rare, known only from a single, moderately sized population divided between two subpopulations spaced no more than 0.4 air km from each other. This taxon should be sought out in additional areas in southeastern California, particularly in and around xeric habitats of the southern Sierra Nevada where alkali-rich, igneous extrusive rocks are exposed and associated species are known to occur.

The subspecific epithet, yorkii, refers to the first collector of this subspecies in the southern Sierra Nevada, Dana York. The suggested common name for the subspecies is York's spring beauty.

Representative Specimens Examined—U. S. A. California: Kern Co., Jawbone Canyon, N slope just below summit of Cross Mountain, 23 Mar 2014, T. R. Stoughton 1847 (RSA, UCR);Near the summit of Cross Mountain located in Jawbone Canyon, 21 Mar 1998, D. York 2079 (CAS1119715, RSA685507); Ca. 24 km NW of California City, Jawbone Canyon, upper slope of Cross Mountain, 8 Apr 2001, D. York 2530 (RSA683083, RSA812090); Ca. 24 km NW of California City, Jawbone Canyon, upper slope of Cross Mountain, 28 Apr 2001, D. York 2555 (CAS1127300).