Penstemon luculentus R.L.Johnson & M.R.Stevens, nom. nov. replaces Penstemon fremontii var. glabrescens Dorn & Lichvar. The varietal name glabrescens was not elevated because it was already occupied by Penstemon glabrescens Pennell, a different species. This new arrangement is supported by molecular and morphological evidence. An analysis of genetic diversity in populations of both varieties of P. fremontii Torr. & A. Gray (glabrescens and fremontii) from the Piceance Basin, Colorado, using SSR (simple sequences repeats) or microsatellites markers, revealed significant genetic differentiation between the two. Penstemon fremontii var. glabrescens was also genetically different from P. gibbensii Dorn and P. scariosus var. garrettii (Pennell) N.H. Holmgren. The combination of hirtellous stems, glabrous leaves, non-glandular inflorescence, and long anther hairs distinguish P. luculentus from other morphologically similar species.

Colorado, Rio Blanco, Piceance, White River Shale, Penstemon

While investigating Penstemon scariosus Pennell (1920) and its varieties, the authors encountered two herbarium specimens from Rio Blanco County, Colorado (BRY81341, BRY81345) that had hirtellous stems, a trait not found in P. scariosus. Further investigation led us to determine that the specimens had been misidentified and that they correctly belonged to Penstemon fremontii var. glabrescensDorn and Lichvar (1990) under existing taxonomic circumscription. Similarly, we encountered several herbarium specimens labeled as P. gibbensii Dorn (1982) from Rio Blanco County, Colorado (BRY112313, BRY112314, BRY112315, BRY112316) that also belonged to P. fremontii var. glabrescens. All but one of these specimens was collected prior to the publication of P. fremontii var. glabrescens and they had not been annotated since to reflect this newer taxonomy. The original determinations of these specimens reflect the observed similarity of P. fremontii var. glabrescens to P. scariosus and P. gibbensii, rather than with P. fremontii Torr. & A. Gray in Gray (1862)sensu stricto.

Though var. glabrescens was recognized at the varietal level within P. fremontii, uncertainty as to its placement within this taxon has been expressed. In the most recent treatment of the Colorado Flora: Western Slope, Weber and Wittmann (2012) state, “In our opinion, this variety is not closely related to P. fremontii and it might be better placed, as a species, closer to the peripheral P. scariosus and P. gibbensii.” The similarity of var. glabrescens to P. gibbensii and P. scariosus was also mentioned in its original publication and morphological comparisons made with these taxa (Dorn and Lichvar 1990), although there was no indication with which of the four varieties of P. scariosus those comparisons were made.

Penstemon gibbensii can be easily distinguished from P. fremontii var. glabrescens by the abundant glandular pubescence present on the inflorescence axis (including sepals and corolla) and distal portions of the stem as compared to the later. The glandular hairs often extend from the distal stem region to mid-stem or below, though becoming less dense proximally. Penstemon scariosus only occasionally has glandular hairs (in some varieties) with hairs sparse and never extending onto the proximal portion of the stem. Variety glabrescens is most easily distinguished from P. fremontiisensu stricto by its glabrous leaves and longer-haired anthers versus P. fremontii that has hirtellous leaves and shorter anther hairs. Variety glabrescens is most easily distinguished from P. scariosus by its hirtellous stem, P. scariosus having glabrous stems.

In this paper, we re-evaluate some morphological characteristics between P. fremontii and P. fremontii var. glabrescens. We also make comparisons against P. scariosus var. garrettii (Pennell 1920) N.H. Holmgren in Cronquist et al. (1984) because it represents a variety of P. scariosus that is geographically proximate and of similar floral characteristics. We also compare the genetic structure within and between P. fremontii varieties fremontii and glabrescens, P. gibbensii, and P. scariosus var. garrettii from the same region using simple sequence repeat (SSR; i.e., microsatellite markers). These markers are useful in inferring genetic exchange among biological populations (Balloux and Lugon-Moulin 2002). It is our opinion that P. fremontii var. glabrescens is a distinct taxon and should be elevated as a unique species.

A minimum of one herbarium voucher and four tissue samples were collected at each accession site (Table 1). These samples were collected either during July 2013 or June 2014. DNA extractions were from lyophilized or silica gel dried leaf tissue collected, in situ (Table 1), using the method detailed by Todd and Vodkin (1996). We used the same PCR parameters and ten of the fluorescently labeled primers (Table 2.) reported by Anderson et al. (2016) to run each DNA sample. Furthermore, we followed their protocol using Geneious 8.0.5 (Kearse et al. 2012) to score the output generated from the ABI 3730xl (Applied Biosystems, Carlsbad, CA, USA) at Brigham Young University’s DNA Sequencing Center (Provo, UT, USA) for the population genetic structure study (Fig. 3A, B).

Figure 3. 

A Plot of the second order difference (ΔK) of K values (2–8) tested in STRUCTURE analysis identifying K = 3 as the optimal number of populations based on the accessions of Penstemon luculentus, P. fremontii, P. scariosus var. garrettii, and P. gibbensii tested. As the K values tested were from 2 to 8, the first difference in K values (ΔK) starts at K = 3 B Bar plot of inferred ancestry coefficients from STRUCTURE analysis results for with K = 3 using 248 samples from 32 accessions. Each number on the x axis represents the accessions ID# in Table 1.

To understand the population genetic structure of the accessions we sampled (Table 1), we used STRUCTURE 2.3 (Falush et al. 2003; Pritchard et al. 2000). The optimal number of genetically distinct clusters or groups (K) was determined by testing K values from 2 to 8 (1 was not tested as multiple clusters were expected) and plotting the second order difference (ΔK) between each K value (Fig. 3A) according to Evano et al. (2005). Analyses consisted of 10 iterations using a burnin period of 50,000 reps with 1,000,000 MCMC reps following burnin, admixture assumed, and sampling locations used as priors. Genetic diversity was partitioned using an analysis of molecular variance (AMOVA) implemented in GenAlEx 6.501 (Peakall and Smouse 2012) to compute pairwise F_ST and R_ST values between taxa (Table 3). The AMOVA was implemented using 999 permutations to calculate P-values for each F_ST or R_ST value. Both pair-wise matrices were then used in GenAlEx to conduct principal coordinate analyses (PCoA) to visualize the differences between taxa (Fig. 4).

Figure 4. 

Plots of eigenvectors of the first two coordinates of principal coordinate analysis based on pairwise R_ST (top graph) or F_ST (bottom graph) values computed from genotypes of ten SSR markers on all taxa. Numbers in parentheses on each axis indicate the percent variation explained by each coordinate.

We made morphological comparisons, using field-collected plants and herbarium specimens obtained from the Stanley L. Welsh Herbarium (BRY) and Rocky Mountain Herbarium (RMS). We took multiple measurements from 38 herbarium sheets of P. fremontii var. glabrescens (≡ P. luculentus) including the holotype and four paratypes, and 20 sheets each of P. fremontiisensu stricto and P. scariosus var. garrettii. Sheet selection was based on the specimen completeness (i.e. only entire plant(s), not partial plants) and the specimen’s pressed condition. Accurate floral measurements required corollas to have dried completely pressed without shrinkage. Sheets of P. fremontii and P. scariosus var. garrettii were selected from the same or adjacent counties to Rio Blanco Co. in Utah and Colorado. Small measurements were taken from digital images with an Olympus SZX-16 dissecting microscope and processed using CellSens Standard 1.8 imaging platform (Olympus Corporation). Because of size similarities between measured plant characteristics, data were plotted as box percentile plots (Fig. 5) with the boxes delimiting the 75th and 25th percentiles and whiskers delimiting the 10th and 90th percentile. Outliers were shown as circles outside the whiskers. We did not have enough material to include P. gibbensii.

Figure 5. 

Box percentile plots showing variations among plant characteristics between P. fremontii, P. luculentus, and P. scariosus var. garrettii. Boxes delimit the 75th and 25th percentiles. The whiskers delimit the 10th and 90th percentile with outliers shown as circles outside the whiskers. The horizontal bar shows the 50th percentile and the horizontal triangle is the mean.

We first analyzed the SSR data, between, and within specimens of P. luculentus, P. fremontii, P. scariosus var. garrettii, and P. gibbensii (Table 1) using STRUCTURE. The results revealed that the best K value for these taxa was K = 3 and at that K value, P. luculentus distinctly differed in population genetic composition from any of the other morphologically similar species (Fig. 3A, B). All eight sites (64 specimens) of P. luculentus sampled across the plant’s range were similar in genetic composition. Varying levels of admixture were detected among sites of P. scariosus var. garrettii. Some sites genetically resemble P. gibbensii and P. fremontii with inferred ancestry coefficients of all specimens of 0.9 or greater for the P. gibbensii and P. fremontii group (blue in Fig. 3B). However, some sites were genetically distinct from all other species with inferred ancestry coefficients of all specimens of 0.9 or greater for their own P. scariosus var. garrettii group (red in Fig. 3B). Still other sites contained specimens that varied in their relatedness to either of these two groups. Penstemon fremontii showed greater genetic similarity to P. gibbensii and P. scariosus var. garrettii than with P. luculentus. This genetic similarity may be due to several factors, such as a possible common ancestor or historical recombination between species. The elucidation of the factors involved in creating these genetic relationships is beyond the scope of this work and requires further research.

To gain an improved understanding of the relationships between P. luculentus, P. fremontii, P. scariosus var. garrettii, and P. gibbensii, we analyzed the SSR allele results using AMOVA (analysis of molecular variance). The analysis revealed that, based on F_ST, molecular variance was partitioned as 15% among taxa, 26% among individuals across taxa, and 59% within individuals of the same taxa, with an overall F_ST of 0.149 (P-value = 0.001). For the AMOVA analysis based on R_ST, molecular variance was partitioned as 11% among taxa, 78% among individuals, and 11% within individuals, with an R_ST value of 0.106 (P-value = 0.002). All pair-wise F_ST and R_ST values were statistically significant except for the R_ST value of P. gibbensii and P. scariosus var. garrettii (Table 3). Analysis with both F_ST and R_ST indicated that P. luculentus has a unique genetic composition as compared to the other taxa which is illustrated in the graphs of the first two coordinates of the PCoA analyses (Fig. 4). These results support the validity of P. luculentus being recognized as a unique species distinct from P. fremontiisensu stricto. The F_ST analysis suggests that P. scariosus var. garrettii and P. fremontii are more closely related than either are to P. gibbensii, while the R_ST analysis suggests that P. scariosus var. garrettii and P. gibbensii are more similar. This discrepancy suggests that microsatellite mutations, which are modeled in the stepwise mutation model of R_ST (reviewed by Balloux and Lugon-Moulin 2002), contribute to genetic differentiation among the taxa examined. The determination of the mutation rates of each SSR locus is beyond the scope of this study, but should be considered in future analyses with these loci.

Morphological comparisons revealed overlap in the size of many plant characters between P. luculentus, P. fremontii, and P. scariosus var. garrettii. Even though there was overlap in the range of measured characteristics, the means do reveal segregating features (Fig. 5). Overall, P. luculentus had more flower stems, a smaller caulescent leaf width, a smaller corolla, and a smaller anther cell length but was found to be intermediate in caulescent leaf length. While P. luculentus was similar to P. fremontii in sepal and anther hair length, these characters were much shorter than those found in P. scariosus var. garrettii.

While P. luculentus has similar morphologically characteristics to P. fremontii, and P. scariosus var. garrettii, there are distinctions that can reliably segregate these taxa. Distinguishing characteristics are more apparent when comparing these taxa in situ. The combination of hirtellous stems, glabrous leaves, non-glandular inflorescence, and long anther hairs can be used to segregate P. luculentus from other related taxa. Differences in other morphological characters are subtler, largely observed as differences in the means of their measurements, and are not reliably diagnostic.

Molecular evidence suggests that P. luculentus is distinct from P. fremontiisensu stricto. It is also distinct from P. scariosus var. garrettii and P. gibbensii. While P. luculentus is not sympatric with P. scariosus var. garrettii, it is well within the geographic range of P. fremontii. We observed P. luculentus and P. fremontii, growing naturally, within 100 m of each other with no apparent hybridization between them. Although we did not observe the two taxa growing interlaced, it is possible that they could co-occur in some areas of the Piceance Basin. Despite both P. luculentus and P. fremontii commonly occurring in the Piceance Basin, there was no morphological evidence that these taxa are exchanging alleles even though they are blooming simultaneously. The results of our study of both the SSR and morphometric data indicate that P. luculentus should be elevated to species status.

This study was initiated as a collateral discovery while determining the extent of where Penstemon scariosus is geographically found. We would like to gratefully acknowledge the funding supported by a BLM grant L14AC00346 “Molecular Characterization of White River Beardtongue, Penstemon scariosus var. albifluvis” to MRS, RLJ and LAJ and the Department of Plant and Wildlife Sciences, Brigham Young University. We are also grateful for plant specimen loans of P. fremontii, and P. fremontii var. glabrescens from the Rocky Mountain Herbarium (RMS), University of Wyoming.