VOLUME 58, NUMBER | JANUARY-MARCH 2011 re NN RA Soo ae “ats %, , | 3 PY COR ae Are i - aa Pa @ Je - ¢ a * P nt . wane * ay 4S A: be # Nisiipee oN RAR EDAPHIC ECOLOGY AND GENETICS OF THE GABBRO-ENDEMIC SHRUB CEANOTHUS RODERICKII (RHAMNACEAE) Dylan-O..Burce and Paul S: MONS: ities RieSeriavisecanreveteavertanigivanieeen l POLLINATION BIOLOGY OF DARLINGTONIA CALIFORNICA (SARRACENIACEAE), THE CALIFORNIA PITCHER PLANT George A. Meindl and Michael R. MeSler..........ccccccccccccceeeeettteeeeeeeeeeeeeeees 22 A MORPHOMETRIC ANALYSIS OF VARIATION BETWEEN ELYMUS ALASKANUS AND ELYMUS VIOLACEUS (POACEAE): IMPLICATIONS FOR RECOGNITION OF TAXA Kristen Harrison and Richayd J. Hep. Feed NS oc even hans dias 32 CHROMOSOME COUNTS AND TAXONOMY OF MENTZELIA THOMPSONII (LOASACEAE) Joshua M. Brokaw, Michael D. Windham, and Larry Hufford.............06+- 50 A NEw SPECIES OF MENTZELIA (LOASACEAE) FROM MONO COUNTY, CALIFORNIA Joshua M. Brokaw and Larry HuffOr[d........cccccccccccccccccsssseseseeeeeeesenneaeeaanees a) RUA GINUN ee seh aces sareeveiea bcc hu neeie dass ett barodcd eo Ma ee LR ccetes es ice La sees PUM ie Wee Davee eal oa eee 64 EUS) 8 E10) 1) bene eR EN, ok, SAN eRe ee eT: ae ie nee Reet 66 MAbDRONO (ISSN 0024-9637) is published quarterly by the California Botanical Society, Inc., and is issued from the office of the Society, Herbaria, Life Sciences Building, University of California, Berkeley, CA 94720. Subscription information on inside back cover. Established 1916. Periodicals postage paid at Berkeley, CA, and additional mailing offices. Return requested. POSTMASTER: Send address changes to MADRONO, Kim Kersh, Membership Chair, Uni- versity and Jepson Herbarium, University of California, Berkeley, CA 94720-2465. kersh@berkeley.edu. Corresponding Editor—TIMOTHY LOWREY Copy Editor—RICHARD WHITKUS Museum of Southwestern Biology Department of Biology MSC03 2020 Sonoma State University University of New Mexico 1801 E. Cotati Avenue Albuquerque, NM 87131-0001 Rohnert Park, CA 94928-3609 madrono@unm.edu whitkus @ sonoma.edu Book Editor—JON E. KEELEY Noteworthy Collections Editors—DIETER WILKEN, MARGRIET WETHERWAX Board of Editors Class of: 2011—JAMIE KNEITEL, California State University, Sacramento, CA KEVIN Rice, University of California, Davis, CA 2012—GRETCHEN LEBUMN, San Francisco State University, CA ROBERT PATTERSON, San Francisco State University, CA 2013—Er!c ROALSON, Washington State University, WA KRISTINA SCHIERENBECK, California State University, Chico, CA 2014—BRANDON PRATT, California State University, Bakersfield, CA ToM WENDT, University of Texas, Austin, TX CALIFORNIA BOTANICAL SOCIETY, INC. OFFICERS FOR 2011—2012 President: V. Thomas Parker, Department of Biology, San Francisco State University, San Francisco, CA 94132, parker @sfsu.edu First Vice President: Andrew Doran, University and Jepson Herbaria, University of California, Berkeley, CA 94720, andrewdoran @ berkeley.edu Second Vice President: Marc Los Huertos, Division of Science & Environmental Policy, California State University, Monterey Bay, Seaside, CA 93955, mloshuertos@csumb.edu Recording Secretary: Mike Vasey, Department of Biology, San Francisco State University, San Francisco, CA 94132, mvasey @sfsu.edu Corresponding Secretary: Heather Driscoll, University Herbarium, University of California, Berkeley, CA 94720, hdriscoll @berkeley.edu Treasurer: Thomas Schweich, California Botanical Society, Jepson Herbarium, University of California, Berkeley, CA 94720, tomas @schweich.com The Council of the California Botanical Society comprises the officers listed above plus the immediate Past President, Dean Kelch, Jepson Herbarium, University of California, Berkeley, CA 94720, dkelch @berkeley.edu; the Membership Chair, Kim Kersh, University and Jepson Herbaria, University of California, Berkeley, CA 94720, kersh @ berkeley. edu; the Editor of Madrono; and three elected Council Members: Chelsea Specht, Department of Plant and Microbial Biology, University of California, Berkeley, CA 94720-2465, cdspecht@berkeley.edu; Ellen Simms, Department of Intergrative Biology, 3060 Valley Life Sciences Bldg., #3140, University of California, Berkeley, CA 94720, esimms @berkeley.edu. Staci Markos, University and Jepson Herbaria, University of California, Berkeley, CA 94720, smarkos @ berkeley.edu. Graduate Student Representatives: Ben Carter, Department of Integrative Biology and University Herbarium, University of California, Berkeley, CA 94720, bcarter@berkeley.edu. Webmaster: Susan Bainbridge, Jepson Herbarium, University of California, Berkeley, CA 94720-2465, sjbainbridge @ berkeley.edu. This paper meets the requirements of ANSI/NISO Z39.48-1992 (Permanence of Paper). MADRONO, Vol. 58, No. 1, pp. 1-21, 2011 EDAPHIC ECOLOGY AND GENETICS OF THE GABBRO-ENDEMIC SHRUB CEANOTHUS RODERICKI (RHAMNACEAE) DYLAN O. BURGE AND PAUL S. MANOS Duke University Department of Biology, Box 90338, Durham, NC 27705 dylan.o.burge@gmail.com ABSTRACT Edaphic-endemic plant taxa are often interpreted as recently derived entities that evolved in situ, with genetic divergence driven by substrate specialization. However, little is known about the evolution of specific edaphic-endemic taxa, particularly the role that soil conditions may play in their initial divergence and continued persistence. Our study focuses on Ceanothus roderickii, a strict specialist on soils derived from a single outcrop of the geological material gabbro located in south- western E] Dorado County, California. In order to elucidate the evolutionary history of C. roderickii we sequenced the third intron of the low-copy nuclear gene nitrate reductase (NIA) for individuals representing four populations of C. roderickii and a wide taxonomic and geographic sampling of closely related plants, including 37 populations of Ceanothus cuneatus and a single representative from 16 other taxa. Analysis of NIA shows that C. roderickii is closely related to C. cuneatus var. cuneatus, a widely distributed taxon found on a diversity of soils. Ceanothus cuneatus var. cuneatus is paraphyletic and comprises two major geographic groups, one coastal and one interior, the latter containing C. roderickii. Thirteen soil chemistry variables were assayed in 42 populations of C. cuneatus representing the wide geographic range of this species, and in 10 populations of C. roderickii. Analysis of these data indicates that evolution of C. roderickii was associated with specialization to nutrient-deficient forms of gabbro-derived soil. Soil chemistry associations of C. cuneatus var. cuneatus and C. roderickii are most divergent where the species come into close contact on gabbro, with C. cuneatus var. cuneatus occupying comparatively nutrient-rich forms of gabbro-derived soils, a result that is consistent with reinforcement. Key Words: Ceanothus, edaphic, evolution, gabbro, NIA, Pine Hill intrusive complex. Edaphic factors—those pertaining to the sub- strate or soil—have long been interpreted as potential drivers of plant diversification (Stebbins 1942; Kruckeberg 1986; Rajakaruna 2004). This idea derives from the strong association of many so-called ‘edaphic endemic’ taxa with particular soil or substrate conditions (Mason 1946; Gankin and Major 1964; Kruckeberg 1986, 2002). In California, for example, approximately 10% of native vascular plants at the level of species and below are endemic to soils derived from serpenti- nite parent material (Kruckeberg 1986; Hickman 1993). Edaphic endemics are often classified either as relicts (paleoendemics) or as recently derived entities (neoendemics) that evolved in situ, with substrate specialization accompanying genetic di- vergence (Raven and Axelrod 1978). Recent work by Baldwin (2005) provided the first phylogenetic evidence for recent divergence of an edaphic endemic taxon, discovering that the serpentinite- endemic herb Layia discoidea D. D. Keck ‘“‘budded off’ from within a less specialized species less than 1 mya. It is not clear, however, whether this pattern is common to the large number of other edaphic endemics in California and elsewhere. By combining detailed genetic surveys with analyses of edaphic conditions experienced by edaphic endemics and their close relatives, it may be possible to discern general trends in the evolution of edaphic endemics, and how these trends relate to soil conditions. Here we focus on the Cerastes subgenus of Ceanothus, which contains 10 edaph- ic-endemic taxa restricted to California and Baja California, Mexico (Table 1). Our goal is to discern the evolutionary history of a single species of edaphic endemic Cerastes, and relate this history to the substrate conditions experienced by this taxon and its closest relatives. In the Sierra Nevada foothills of western El Dorado County, California, soils weathered from mafic rocks of the Pine Hill intrusive complex (~100 km’, Fig. 1; Springer 1971) support approximately 600 vascular plant species (Wilson et al. 2009, Appendix 1), representing more than 10% of the California flora (5867 species; Hick- man 1993), including several endemics and taxa of limited or disjunct distribution (Wilson 1986; Hunter and Horenstein 1991; Wilson et al. 2009). Endemics of the Pine Hill intrusive complex include Ceanothus roderickii W. Knight, Fremon- todendron californicum (Torr.) Coville subsp. decumbens (R. M. Lloyd) Munz, Galium califor- nicum Hook. & Arn. subsp. sierrae Dempster & Stebbins, and Wyethia reticulata Greene. The first three of these plants are federally-listed Endangered Species (USFWS 1996). The Pine Hill intrusive complex is composed primarily of the rock gabbro, with minor 2 MADRONO [Vol. 58 TABLE 1. CEANOTHUS, SUBGENUS CERASTES TAXA. Taxon = taxa from Fross and Wilken (2006). Sampled = populations sampled for genetic and/or soil analyses (Appendix 1). Geographic distribution = distribution of taxa in North America (North CA: region of CA from the latitude of Point Conception, north; South CA: region of CA from the latitude of Point Conception, south; BC, Mexico: Mexican state of Baja California). Soil = geological parent material(s) for soils on which taxon occurs (Fross & Wilken 2006). 725 populations of C. cuneatus var. cuneatus sampled for genetics and soil, 8 for genetics only, and 13 for soil only. ° Three populations of C. roderickii sampled for genetics and soil, 1 for genetics only, and 7 for soil only. ‘Also found on gabbro-derived soils. ‘ Parent material classified as metavolcanic (““Mzv’’) by Jennings (1977). Taxon Sampled Geographic distribution Soil Ceanothus arcuatus McMinn 0 North CA various C. bolensis S. Boyd & J.E. Keeley 0 BC, Mexico basalt C. crassifolius Torr. var. crassifolius 0 South CA; BC, Mexico various C. crassifolius Torr. var. planus Abrams 0 South CA various C. cuneatus Nutt. var. cuneatus 46° West US; BC, Mexico various C. cuneatus Nutt. var. dubius J.T. Howell ] North CA various C. cuneatus Nutt. var. fascicularis (McMinn) Hoover 1 North CA various C. cuneatus Nutt. var. ramulosus Greene l North CA various C. cuneatus Nutt. var. rigidus (Nutt.) Hoover l North CA various C. divergens Parry subsp. confusus (J.T. Howell) Abrams 1 North CA various C. divergens Parry subsp. divergens 0 North CA various C. divergens Parry subsp. occidentalis (McMinn) Abrams 1 North CA various C. ferrisiae McMinn ] North CA serpentinite C. fresnensis Abrams ] North CA various C. gloriosus J.T. Howell var. exaltatus J.T. Howell l North CA various C. gloriosus J.T. Howell var. gloriosus 1 North CA various C. gloriosus J.T. Howell var. porrectus J.T Howell l North CA granite C. jepsonii Greene var. albiflorus J.T. Howell 1 North CA serpentinite C. jepsonii Greene var. jepsonii 1 North CA serpentinite C. maritimus Hoover l North CA various C. masonii McMinn 1 North CA various C. megacarpus Nutt. var. insularis (Eastw.) Munz 0 South CA various C. megacarpus Nutt. var. megacarpus 0 South CA various C. ophiochilus S. Boyd, T.S. Ross & Arnseth 0 South CA pyroxenite® C. otayensis McMinn 0 South CA; BC, Mexico basalt C. pauciflorus DC. 0 Mexico various C. perplexans Trel. 0 South CA; BC, Mexico various C. pinetorum Coville l North CA various C. prostratus Benth. l North CA various Ceanothus pumilus Greene | North CA serpentinite C. purpureus Jeps. l North CA volcanic C. roderickii W. Knight 11° North CA gabbro C. sonomensis J.T. Howell l North CA various C. verrucosus Nutt. 0 South CA; BC, Mexico various C. vestitus Greene 0 West US; Mexico various amounts of pyroxenite and diorite (Springer (the Rescue Series; Rogers 1974). Gabbro con- 1980; hereafter referred to collectively as “‘gab- tains less iron, Mg, and potentially plant-toxic bro’), which weather to form reddish-brown transition elements (e.g., Cr, Co, Ni) than are sandy loams with very stony to clayey variants found in ultramafic rocks such as serpentinite —= Fic. 1. Sampling and soil map for the Pine Hill region, El Dorado Co., California. Polygons for gabbro or serpentinite derived soils adapted from GIS data layers in Soil Survey Geographic (SSURGO) database for El Dorado Area, California (U.S. Department of Agriculture, Natural Resources Conservation Service 2007). Extremely stony gabbro soil: shallow soils derived from gabbro parent material, corresponding to “Rescue extremely stony sandy loam’ (Rogers 1974, RgE2). Other gabbro soil: deeper soils derived from gabbro parent 2011] BURGE AND MANOS: EDAPHIC ECOLOGY OF CEANOTHUS RODERICKII of Coloma Bass Lake ; . . be af : —— — 38° 40'N . ise" US Hwy 50 XA C. roderickii, sampled See O C. cuneatus var. cuneatus, sampled © C. roderickii distribution © Extremely stony gabbro soil CE) Other gabbro soil Serpentine soil Major road (1.0 km) 121° 0' W g | % material, corresponding to ‘Rescue sandy loam” (Rogers 1974, ReC, ReB, & ReD) and ‘“‘Rescue very stony sandy loam” (Rogers 1974, RfC, RfD, & RfE). Serpentine soil: very shallow, rocky soils derived from serpentinite parent material, corresponding to ‘Serpentine rock land’ (Rogers 1974, SaF). Ceanothus roderickii distribution adapted from Hinshaw (2008) in consultation with G. Hinshaw and L. Fety (March, 2010). Sampling locations indicated by Stars or open circles (Table 2). 4 MADRONO (Alexander 1993, unpublished). As a result, soils derived from gabbro usually contain elevated levels of Mg relative to soils derived from less mafic rocks such as diorite, but have lower Ca to Meg ratios than soils weathered from ultramafic rocks such as serpentinite (Goldhaber et al. 2009). Endemism on the Pine Hill intrusive complex, as well as the presence of species normally restricted to serpentinite-derived soils, has been attributed to the similar properties of gabbro and serpentinite rock (Wilson 1986). However, anal- yses of soils from the Pine Hill intrusive complex have not identified soil parameters that predict plant distributions (Hunter and Horenstein 1991; Alexander, unpublished), leading Hunter and Horenstein (1991) to conclude that endemism on these soils may be attributed to the island-like topographic position of the complex in the otherwise low-lying foothills of the central Sierra Nevada. However, these results may be con- founded by plant demography, especially in the case of C. roderickii, which is dependant on fire for significant recruitment (Boyd 2007). Ceanothus roderickii is a member of the Cerastes subgenus of Ceanothus, a group of 24 species that is almost entirely restricted to the California Floristic Province (CFP) of western North America (Fross and Wilken 2006). Mem- bers of Cerastes possess a suite of morphological and physiological adaptations for drought resis- tance (Ackerly et al. 2006) and are associated with chaparral vegetation. However, the group is both morphologically and ecologically diverse, with an array of growth forms and a broad spectrum of habitat associations, sometimes including specialized edaphic ecology (McMinn 1942; Nobs 1963; Fross and Wilken 2006). Ceanothus roderickii is a decumbent shrub spreading horizontally via arching branches that usually root adventitiously when nodes contact soil (Knight 1968), a trait that allows this species to reproduce clonally during fire-free intervals when recruitment from the seed-bank is limited (Boyd 2007). A close relationship between C. roderickii and the widespread Ceanothus cuneatus Nutt. was proposed by Knight (1968). However, this author also speculated on the possibility of a close relationship between C. roderickii and Ceanothus fresnensis Abrams or Ceanothus pros- tratus Benth., the only other decumbent members of Cerastes known from the central Sierra Nevada. Sequence data from nuclear ribosomal DNA suggest that C. roderickii 1s closely related to C. cuneatus var. cuneatus (Hardig et al. 2000). Ceanothus cuneatus is among the most widely distributed members of Ceanothus, occupying forest, woodland, and chaparral habitats at low to moderate elevations in far western North America from Baja California, Mexico to north- [Vol. 58 western Oregon (Fig. 2), almost entirely within the CFP (Fig. 2). Ceanothus cuneatus comprises five varieties (Table 1), four of which are narrowly distributed (Fross and Wilken 2006). The most widely distributed variety, C. cuneatus var. cuneatus, 1s a characteristic component of chaparral and woodland communities in the foothills and mountains of the CFP and is known to grow on soils derived from a variety of geological parent materials (Fross and Wilken 2006; Table 1). Ceanothus cuneatus var. cuneatus is the only member of Cerastes other than C. roderickii known to occur in the Pine Hill region of El Dorado Co., California. This study was designed to elucidate the evolutionary history of C. roderickii and relate this history to the substrate specificity of the taxon and its closest relatives. Specifically, this study aimed to 1) test the hypothesis that C. roderickii is most closely related to and possibly derived from within C. cuneatus var. cuneatus, 2) characterize the soil chemistry associations of C. roderickii relative to those of C. cuneatus var. cuneatus, and 3) identify specific chemical prop- erties of C. roderickii soils that may have provided selective pressure during evolution of the species. MATERIALS AND METHODS Genetic Sampling Genetic sampling of Ceanothus populations was designed to represent the geographic range and edaphic tolerances of the focal taxa C. roderickii and C. cuneatus var. cuneatus, and to encompass related plants (Tables | and 2, Fig. 2, Appendix 1). DNA from 57 plants was studied, representing 22 of the approximately 35 Cerastes taxa (species, subspecies, and varieties) currently recognized (Table 1; Fross and Wilken 2006). All individuals sampled for the present work were collected by D. Burge, with the exception of a sample of Ceanothus pinetorum Coville obtained by D. Wilken (DHW 16736, Table 2, Appendix 1). Individuals from four populations of C. roderickii were sampled to represent the geo- graphic distribution of this species (Table 2, Fig. 1, Appendix 1). Individuals from 33 popula- tions of C. cuneatus var. cuneatus were sampled to represent the extensive geographic range of this taxon and the variety of edaphic conditions that it experiences over this area (Table 2, Fig. 2). Individuals representing populations of 20 addi- tional Cerastes taxa were sampled for genetic analysis based on a large-scale phylogenetic study of Ceanothus (Burge et al. in press). In this large- scale study, which is based on more than 140 Cerastes populations from across the geographic range of the group, individuals included in the present study (Table 2, Appendix 1) form a 2011] BURGE AND MANOS: EDAPHIC ECOLOGY OF CEANOTHUS RODERICKII = e : : Portland Sampling Sites O C. cuneatus var. cuneatus @ Other Cerastes Other Map Features €) C. cuneatus distribution 150 km Boundary of CFP e@ Major cities — 40°N Figure 2 (8-14, 16-20, 66-76) Las Vegas Los Angeles Ss FiG. 2. Sampling map for western North America. Soil and/or genetic sampling locations indicated by open circles (Table 2). Global distribution of C. cuneatus indicated by gray shading (data provided by the participants of the Consortium of California Herbaria; March, 2010). Boundary of the California Floristic Province (CFP) adapted from Myers et al. (2000). [Vol. 58 ~ MADRONO CTZTEOPTINH :IZEOPTINH yoHURIs epeT][ sein “vO Iv 9¢COPCNH eJIUBS[OA BgOT] -eUeYysy, “WO OV ZTIEOPZINH ‘I IEOr7NH STUBSTOA BE60OI -FOHNS “VO 6¢ 60€0P7INH :80€0P7NH yoyurTjuadias BEBO “PLISIS “WO Be TEEOPTINH :ITEECOrP7NH orydiowrejour BICIT ‘Biseyg “WO Le CVCOVCINH AI’ USUIIpas 8666 ‘OdsIqO sin] ues ‘WD OC 9PE0P7NH ouuRss BHZ6 :O83Iq URS “VO SE OOLOVCNH ~AIBJUSUIpas BOLO ‘OUIpIeUlog URS “VD Ve ClL 0.95) relationships with isolates from other taxa, including other varieties of C. cuneatus and other species of Cerastes (Fig. 3). Out of the 30 Ceanothus plants from which two NIA sequence types were recovered (Table 2, 2011] BURGE AND MANOS: EDAPHIC ECOLOGY OF CEANOTHUS RODERICKII 1] : : 2. Ys eee OF © Fs, 7249228885 4? < o ve &£ & aS 2S 8 Oh Of ee Fo eee Rete he 4° @ ¥ § 6 F > oF P 4 OA @) ice) 5 8 & > O OQ 4 < ne} i?) Variable region motif TTTTAAACAAAA-TTA Figure key TTTTTAAAAAAA-TTA TTTTAAAAAAAA-TTA TTTAAAAAAAAA-TTA 4h C. roderickii TTTTTAAAAAAAA-TA @ Oihericaimeies TTTTAAAAAGAA-TTA CPC TARAA AAT — 0.09 Substitutions/site TT TTAAAAAAA--TTA TTT TAAAATAAAATTA 1101-c1 A_ Isolate code and motif type TTTTAACAAAAA-TTA TTTTTAAAAAAAATTA TTTTAAAAAAA--TTG TTTTTAAAAGAAATTA TTTTTTAAA-AAATTA O C. cuneatus var. cuneatus Z2ErAONQGOHTOWAUVUAD Pp Fic. 3. Bayesian 50% majority-rule consensus phylogram for nitrate reductase. Heavy bars indicate posterior probability >0.95. Phylogram is manually rooted based on root position inferred from expanded Ceanothus phylogeny (see Materials and Methods). Highly-variable region motifs from NIA (see Materials and Methods) shown below phylogram; motif types mapped on phylogram using letter codes. ““A, B, C’’: groups and clades discussed in text; numbers on branches indicate posterior probabilities. All NIA isolates from DOB collections, with exception of 16736-cl from D.H. Wilken 16736 (Table 2; Appendix 1). 12 MADRONO Appendix 1), 21 have these isolates in conflicting positions on the phylogeny (Fig. 3; PP > 0.95). Of the remaining nine plants from which two NIA sequence types were recovered, five have both isolates as members of a single well- supported clade (PP > 0.95), and four have isolates that are neither strongly supported as members of the same clade, nor in conflicting phylogenetic positions (Fig. 3). Gene Genealogy Among the 87 NIA isolates included in the analysis, TCS identified 82 unique sequences. Three of these are represented by more than one NIA isolate (Fig. 4), one comprising four isolates from individuals of C. cuneatus var. cuneatus collected in the southern Sierra Nevada (1150a- cl, 1149a-cl, 1136a-cl, and 1134a-c2), a second comprising two isolates from individuals of C. cuneatus var. cuneatus collected in the northern Sierra Nevada of California and Cascade Ranges of Oregon (1164a-cl and 1168a-c3), and a third represented by two isolates from individuals of C. roderickii collected in different populations (1087a-c3 and 824b-cl). All remaining sequence types are unique. The gene genealogy inferred by TCS is reticulate, with 22 loops (ambiguities) as reconstructed (Fig. 4). Highly-Variable Region Motifs Among the 87 NIA isolates utilized in the study, a total of 14 motif types were identified for the first highly-variable region (Materials and Methods; Fig. 3). The ““N” motif (Fig. 3) is unique to C. roderickii and was present in all 16 isolates (four per individual plant) obtained from individuals of this species, as well as 16 isolates obtained from 4 additional individuals of the species collected in different populations or sub- populations (unpublished data). Nine motif types are found in Ceanothus cuneatus var. cuneatus (Fig. 3). Seven of these types are unique to C. cuneatus var. cuneatus, including three known from just a single NIA isolate each (F, G, and I). The remaining two motifs recovered in C. cuneatus var. cuneatus (C and D) are shared with other varieties of C. cuneatus or other Cerastes species (Fig. 3). None of the motifs from C. cuneatus var. cuneatus 18 unique to a _ well- supported group of C. cuneatus var. cuneatus isolates in the NIA tree (Fig. 3), although the “B” motif is found predominantly in Clade B (Fig. 3; PP 0.79) and is found in all but one of the C. cuneatus var. cuneatus isolates that group with C. roderickii in Clade A (Fig. 3; PP 0.88). Four additional motifs (J, K, L, and M; Fig. 3) are found only in taxa other than C. cuneatus var. cuneatus and C. roderickii. Two of these are found in more than one taxon (J and K; Fig. 3), [Vol. 58 and two are unique to a particular isolate (L and M; Fig. 3). Soil Analyses At a large geographic scale (Fig. 2), consider- ing all 52 soil samples collected within popula- tions of C. cuneatus (all five varieties) and C. roderickii (Tables | and 2), the soils of C. roderickii have, on average, lower pH, lower electrical conductivity, and lower concentrations of nitrate, K, P, S, Na, Fe, Zn, and Mn (Table 3, C. roderickii vs. C. cuneatus all samples). Concentrations of Mg, Ca, and Cu, on the other hand, are on average higher in the soils of C. roderickii than in those of C. cuneatus (Table 3). Differences are significant in the case of K, P, S, Fe, and Zn (Student’s paired t-tests, P < 0.03). Principal component analysis summarizes these results for the 13 soil chemistry variables. In PCA the first two principal components account for 39% of total variance, with 21% on the first principal component and 19% on the second. The first principal component is strongly positively correlated with Mg (vector loading = 0.48) and electrical conductivity (vector loading = 0.34), and strongly negatively correlated with P (vector loading = 0.46) and K (vector loading = 0.41). These results are summarized in a biplot of the first two principal components (Fig. 5A). Stu- dent’s paired t-tests allow for rejection of the null hypothesis of no difference between the mean PCA scores for C. roderickii and C. cuneatus on the second principal component (P < 0.001; Fig. 5B) but not the first (P = 0.052). At a smaller geographic scale (Fig. 1), consid- ering only those 28 soil samples collected in El Dorado Co., California (Tables 1 and 2), there are differences in chemistry between soils of C. roderickii and C. cuneatus var. cuneatus that are partitioned with respect to both taxon and geological parent material (Table 3). Within this geographic region C. cuneatus var. cuneatus grows on soils derived from a variety of geological parent materials, including gabbro (Tables 1 and 2). In comparison to C. roderickii, soils of C. cuneatus var. cuneatus that are derived from non-gabbro parent material (including serpentinite; Table 2; see below) have, on aver- age, higher pH and higher concentrations of nitrate, Mg, K, P, Fe, Zn, and Mn (Table 3, C. roderickii vs. C. cuneatus El Dorado non-gabbro). Electrical conductivity and concentrations of Ca, S, Na, and Cu, on the other hand, are lower in soils of C. cuneatus var. cuneatus derived from non-gabbro parent material than in the soils of C. roderickii (Table 3). Differences are significant in the case of Fe, Zn, and Cu (Student’s paired t-tests, P < 0.04). Comparing the exclusively gabbro-derived soils of C. roderickii to the soils of C. cuneatus var. cuneatus that are also derived 2011] / {p-0-Og : (1460-03) ce = linac ’ Clade B wee §=150 km (inset) BURGE AND MANOS: EDAPHIC ECOLOGY OF CEANOTHUS RODERICKII ere een--=, ee Wa eeceeen (1117a-c2)}———— yeas ne, @ 1008a-c1 { 982a-c5 ‘ Gigmec2s = | QO —-, pee ey, ; Ne = Mo77ac13 VY G 093a-c2 3 D alesiaiaiates e U) pe Ce id 2 @ OQ} O @ eon omen 1078a-c4 } eee td eureeeeree-= © (1075a-c1} : ® : Oo cane ay (7083a-c44 / { 1083a-c1} i ites? | tetera i" NRO ’ Sewen gene e? : ; 1151a-c1 4132a-c13 # DP cisclages (ome) (1117a-c1} ee ee. (1023a-c4 3 i Cl SY“ ai ices wine f 1150a-c1 (ee ‘ i 1149a-c1 9840-06 }—C) aa OOO} 1136a-c1 ! '~, \ 1134a-c2_ nee ay, Cen : 16736-c5 pened mee " { 1132a-c2} ai f7eerct> | j- -|~ “ses#esese : 1161a-c2 OF 0-0-0 T4833) sok [LE Seemeeeoore 959a-c3 pn @ pen Ope One Onn Onn Onn © 1138a-c1 ) oe OR 9940-03 887a-c6 994a-c2 834a-c1 907a-c1 16736-c1 916a-c2 Figure key oN CO) 895b-c3 O Missing sequence a) 10032-c2 A ag ¥* C. roderickii @ 904a-c4 Cc) C. cuneatus var. cun. © 1003a-c3 [___] Other Cerastes 907a-c11 rrr 908a-c1 ( ) 913a-c9 Fic. 4. Gene genealogy of NIA isolates generated under the statistical parsimony criterion in the program TCS (Clement et al. 2000). Open circles represent un-sampled (missing) sequences, as inferred by TCS. Some branch lengths not proportional to number of substitutions. Ceanothus roderickii (solid black stars): A, 1087a-c3 & 824b- cl; B, 1080a-c2; C, 1087a-c2; D, 1111-cl. “‘Clade A”: group discussed in text. Sequences color-coded according to geography (see inset map). 14 MADRONO [Vol. 58 Second PC Scores (Second PC) First PC Second PC Scores (First PC) -0.4 -0.2 0.0 0.2 0.4 First PC 4% C. roderickii © C. cuneatus gabbro @ C. cuneatus serpentinite O C. cuneatus other Fic. 5. Plots from principal component analysis (PCA) of soil chemistry data. A, biplot for first two principal components of PCA on soil chemistry for 52 assayed soil samples; arrows represent direction and magnitude of loading on the principal component axes; bottom and left axes apply to loading; top and right axes apply to PCA scores. B, boxplot of PCA scores from the second principal component of PCA on all 52 assayed soil samples, partitioned by species. C, biplot for first two principal components of PCA on soil chemistry for all soils collected in El Dorado County, CA; axes as in A. D, boxplot of PCA scores from the first principal component of PCA on El 2011] from gabbro parent material, the gabbro-derived soils of C. cuneatus var. cuneatus have, on average, higher pH, higher electrical conductivity, and higher concentrations of nitrate, Ca, K, P, S, Na, Fe, Zn, Mn, and Cu (Table 3). Mg is the only element that is present in lower concentrations in soils of C. roderickii than in gabbro-derived soils of C. cuneatus var. cuneatus. Differences are significant for K, Ca, F, S, Fe, Mn, and Cu (Student’s paired t-tests, P < 0.04). Principal component analysis summarizes these results for the 13 soil chemistry variables. In PCA the first two principal components account for 45% of total variance, with 27% on the first principal component and 18% on the second. The first principal component is strongly positively corre- lated with Mg (vector loading = 0.35) and pH (vector loading = 0.24), and strongly negatively correlated with S (vector loading = 0.49) and K (vector loading = 0.36). These results are summarized in a biplot of the first two principal components (Fig. 5C). Differences among the exclusively gabbro-derived soils of C. roderickii, the gabbro-derived soils of C. cuneatus var. cuneatus, and the non-gabbro derived soils of C. cuneatus var. cuneatus were tested using ANOVA on PCA scores (Fig. 5C). ANOVA allowed for rejection of the null hypothesis of no difference among the three group means on the basis of the first principal component (F = 10.96; P < 0.001; Fig. SD) as well as the second (F = 6.34; P = 0.006). Tukey’s HSD test allowed for rejection of the null hypothesis of no difference between the gabbro-derived and non-gabbro derived soils of C. cuneatus var. cuneatus (P = 0.002), as well as between the gabbro-derived soils of C. cuneatus var. cuneatus and those of C. roderickii (P< 0.001). The mean PCA scores for the non-gabbro derived soils of C. cuneatus var. cuneatus was not significantly different from those of C. roderickii (P = 0.611). Comparing the gabbro-derived soils of C. roderickii and C. cuneatus var. cuneatus to serpentinite-derived soils of C. cuneatus (includ- ing C. cuneatus var. cuneatus and C. cuneatus Nutt. var. ramulosus Greene), there are strong differences among groups. Average Ca:Mg for serpentinite-derived soils of C. cuneatus (n = 8; Table 2) was 0.6 (standard deviation = 0.3), the average for soils of C. roderickii (n = 10) was 2.9 (+0.6), the average for the gabbro-derived soils of C. cuneatus var. cuneatus (n = 6) was 5.5 (+1.5), and the average for all ‘‘other’” (non- gabbro and non-serpentinite derived) soils occu- <— BURGE AND MANOS: EDAPHIC ECOLOGY OF CEANOTHUS RODERICKII 15 pied by C. cuneatus (n = 27) was 7.2 (+4.1). The difference in Ca:Mg is significant for all three contrasts among 1) the exclusively gabbro- derived soils of C. roderickii, 2) the gabbro- derived soils of C. cuneatus var. cuneatus and 3) the serpentinite-derived soils of C. cuneatus (Student’s paired t-tests, P < 0.01). Overall differences in soil chemistry among these groups are summarized in a biplot of the first two principal components from the PCA described above (Fig. 5A). The differences in soil chemistry among the three groups listed above, as well as the “other” group (non-gabbro and non-serpen- tinite derived soils occupied by C. cuneatus) were tested using ANOVA in terms of scores on the second principal component of the PCA (Fig. SA). ANOVA allowed for rejection of the null hypothesis of no difference among group means (F = 5.01; P = 0.004). Furthermore, Tukey’s HSD test allowed for rejection of the null hypothesis of no difference between means for two contrasts among the four groups listed above: a) gabbro-derived soils of C. cuneatus var. cuneatus versus “other” (non-gabbro & non- serpentinite derived) soils of C. cuneatus (P = 0.014), and b) gabbro-derived soils of C. cuneatus var. cuneatus versus those of C. roderickii (P = 0.002). The remaining three contrasts among the four groups were not significant. DISCUSSION Phylogenetic Relationships Our results indicate a very close relationship between the gabbro-endemic C. roderickii and the less soil-specialized C. cuneatus var. cuneatus. Nevertheless, relationships among the 87 NIA isolates included in this study are poorly resolved, with few nodes receiving high levels of support (Fig. 3). This result is consistent with past genetic work on Cerastes as a whole, in which nuclear and chloroplast DNA sequence data failed to resolve species-level relationships (Hardig et al. 2000; 2002). Nevertheless, a lack of phylogenetic signal is consistent with the hypothesis that Cerastes diversified recently, perhaps as late as 5 mya (Ackerly et al. 2006; Burge et al. in press). If the diversification of Cerastes took place during so short a time interval, then a lack of phylogenetic resolution is not unexpected. In addition, genetic divergence among taxa might be further eroded by hybridization, which is com- mon among Cerastes taxa and has long been Dorado County soil samples, partitioned by species-soil group (see Results). ““C. cuneatus gabbro” corresponds to soil samples obtained from C. cuneatus populations growing on soils derived from gabbro parent material; “*C. cuneatus serpentinite” corresponds to serpentinite parent material, and “C. cuneatus other” to non-gabbro and non-serpentinite parent materials. Symbols: Con = electrical conductivity; NO3 = nitrate. 16 MADRONO thought to play an important role in Cerastes evolution (McMinn 1942; Nobs 1963). In spite of the low phylogenetic resolution achieved in the present study using NIA, comparison of phylogenetic results with the gene genealogy and information from highly-variable region motifs (Figs. 3, 4) allows for interpretation of the relationship of C. roderickii to remaining Cerastes. All of the NIA isolates obtained from C. roderickii, representing four populations, are nested within a small clade made up of NIA isolates from C. cuneatus var. cuneatus popula- tions sampled in the Sierra Nevada and Cascade mountains of California (Fig. 3, Clade A). This group is also present in the gene genealogy for NIA, in which only two potential connections were reconstructed between this group and remaining isolates (Fig. 4 ““Clade A’’). The close relationship between C. roderickii and C. cunea- tus var. cuneatus is further emphasized by the nested position of Clade A within Clade B, which is made up almost entirely of isolates from cuneatus var. cuneatus (Fig. 3). However, it is important to note that members of Clade B are more strongly connected to the remaining NIA isolates in the gene genealogy than are members of Clade A (Fig. 4). Finally, all NIA isolates from C. roderickii contained an identical highly-vari- able region motif that has proven unique to C. roderickii (Fig. 3, type N). The type N motif was present in all 20 isolates obtained from C. roderickii. Four C. roderickii individuals repre- senting additional populations were also found to share the type N highly-variable region motif (unpublished data). The presence of a unique highly-variable region motif in all sampled C. roderickii individuals indicates that C. roderickii populations are genetically cohesive, in spite of the fact that they do not form a clade in the phylogeny reconstructed using complete NIA sequences (Fig. 3). Thus, the type N_ highly- variable region motif may be taken as a genetic autapomorphy of C. roderickii. Genetic evidence for the cohesiveness of C. roderickii with respect to C. cuneatus is supported by the morphology of C. roderickii, which differs from that of C. cuneatus in several significant ways. First, the habit of C. roderickii is always prostrate to decumbent, with shrubs rarely attaining more than a meter in height (Knight 1968; James 1996), whereas C. cuneatus 1s invariably erect and ascending, frequently reach- ing more than three meters is height (Fross and Wilken 2006). However, some populations of C. cuneatus in the Sierra Nevada and Klamath- Siskiyou region of California and Oregon are much lower-growing (Fross and Wilken 2006). Ceanothus roderickii also differs from C. cuneatus with respect to mode of reproduction. Individuals of C. roderickii spread laterally via arching or creeping branches that root adventitiously when [Vol. 58 they contact soil. This mode of reproduction allows C. roderickii to reproduce clonally during fire-free intervals when seedling recruitment is limited (James 1996; Boyd 2007). Clonal repro- duction is not known in C. cuneatus. Finally, the leaves of C. roderickii are usually strongly ascending (Knight 1968), such that the leaf surface is typically held perpendicular to the soil surface. Few other Cerastes species are known to possess this trait (Knight 1968), which may represent an adaptation to the very high light levels that are typical of the open habitats favored by C. roderickii (James 1996). Overall, phylogenetic findings of the present study agree with previous systematic work on C. roderickii. Citing general similarities in habit, ecology, and geographic distribution, Knight (1968) argued that C. roderickii is probably most closely related to C. cuneatus, although he did not rule-out the possibility of a relationship with several other Cerastes species from the Sierra Nevada. The results of the present study also agree with those of Hardig et al. (2000), in which an individual of C. roderickii grouped with an individual of C. cuneatus var. cuneatus in phylog- enies based on sequences from ITS and matK. In addition to the relationship between C. roderickii and C. cuneatus, results presented here bear on relationships among other Cerastes included in the taxonomically diverse clade that is the focus of the present study (Table 1). The Bayesian consensus tree contains a moderately supported clade comprising 38 out of the 52 NIA isolates from C. cuneatus var. cuneatus, all of the isolates for C. roderickii, and a single isolate from Ceanothus cuneatus Nutt. var. fascicularis (McMinn) Hoover (Fig. 3, Clade B, PP 0.79). Clade B is made up almost entirely of NIA isolates from C. cuneatus var. cuneatus individu- als collected in the mountains of Baja California, Mexico, southern California, eastern California, and eastern Oregon, which includes the Sierra Nevada, Cascade Ranges, Peninsular Ranges, and Transverse Ranges (Fig. 4). Although the relationship is less obvious than in the Bayesian consensus tree (Fig. 3), Clade B and its unusual geography is recognizable in the gene genealogy, which contains few connections between mem- bers of Clade B and remaining isolates (Fig. 4). The relationship between Clade B and remaining NIA isolates is not resolved in the Bayesian consensus tree (Fig. 3); several small clades of isolates, as well as some individual isolates, form a large polytomy with Clade B (Fig. 3, Group C). In an expanded analysis of Ceanothus phylogeny (Burge et al. in press) the root of our tree (Fig. 3) falls within this polytomy, indicating that a lack of resolution here is not an artifact of sampling. All but 10 of the plants represented by the Group C isolates were collected in the Klamath- Siskiyou and Coast Ranges of California, the 2011] exceptions being seven isolates from individuals of C. cuneatus var. cuneatus collected in the Sierra Nevada, Peninsular Ranges, and Transverse Ranges (1149a-c3, 1095a-c3, 1084a-cl, 1134a-cl, 1070a-cl, 1030a-2, and 982a-c2), one isolate of C. cuneatus var. cuneatus collected in the Sutter Buttes (1093a-cl), and one isolate each from individuals of C. fresnensis (1138a-cl) and C. prostratus (952a-cl) collected in the Sierra Ne- vada (Fig. 4). The genetic break between the Klamath- Siskiyou/Coast Ranges and the remaining CFP mountains (Sierra Nevada, Peninsular Ranges, and Transverse Ranges; Fig. 4) appears to represent a biogeographic split between Cerastes inhabiting these regions, although the presence of isolates from individuals collected in the Kla- math-Siskiyou/Coast Ranges within Clade B, and the presence of individuals collected in other mountain ranges of the CFP in Group C, suggests that opportunities for migration and/or gene-flow between the regions have been avail- able (Figs. 3, 4). In addition, the frequent lack of monophyly between NIA isolates from the same individual (Fig. 3), including 6 cases in which isolates from a single individual are found in both Clade B and Group C (1093a, 1149a, 1134a, 1030a, 982a, 1095a), suggests the operation of gene-flow or hybridization. Hybridization and gene flow are thought to be common in Cerastes (McMinn 1942; Nobs 1963; Fross and Wilken 2006), and so it is not unexpected to find evidence consistent with these phenomena. Edaphic Ecology At a large geographic scale, considering all sampled populations of C. cuneatus and C. roderickii (Fig. 2), results of our study show that edaphic conditions experienced by the narrowly distributed gabbro-endemic C. roderickii repre- sent a small, highly cohesive subset of the range of conditions experienced by the widespread soil- generalist C. cuneatus (Fig. 5A, B). Soils of C. roderickii are characterized by low concentrations of available K, P, S, Fe, and Zn, all of which are necessary plant nutrients (Brady and Weil 2002). For many plants, low availability of these elements results in disorders affecting growth and reproduction (Brady and Weil 2002). At the scale of the Pine Hill intrusive complex in western El Dorado Co., California (Fig. 1), our study shows that C. roderickii is specialized to nutrient-deficient forms of gabbro-derived soil (Fig. 5C, D). On the Pine Hill intrusive complex C. cuneatus var. cuneatus and C. roderickii both occur on soils that are considered gabbro-derived (Fig. 1). However, the gabbro-derived soils of C. roderickii sampled in our study, which are Classified as ‘“‘Rescue extremely stony sandy loam” (Rogers 1974), contain significantly lower BURGE AND MANOS: EDAPHIC ECOLOGY OF CEANOTHUS RODERICKII i7 levels of K, Ca, P, S, Fe, Mn, and Cu than gabbro-derived soils of C. cuneatus var. cuneatus (P < 0.04; Table 3), which are classified as ‘““Rescue sandy loam” or “‘Rescue very stony sandy loam” (Fig. 1; Rogers 1974). Although these elements are necessary plant nutrients, high levels of some, such as Mn, Fe, and Cu, are known to induce growth and reproductive disorders in plants (Brady and Weil 2002). Our work is the first to report this strong soil- chemistry divergence between C. cuneatus var. cuneatus and C. roderickii on the Pine Hill intrusive complex. The relatively higher fertility of gabbro-derived soils occupied by C. cuneatus var. cuneatus compared to those occupied by C. roderickii may result from the greater development of the former, which are typically found in swales and at the bases of steep slopes, where they receive run- off from the Rescue extremely stony sandy loams that are found on the steeper slopes, hills, and ridge crests of the Pine Hill intrusive complex (Rogers 1974; D.O. Burge, personal observation). While our study is the first to report significantly divergent chemistry between groups of gabbro- derived soils on the Pine Hill intrusive complex, similar phenomena are known from other soils; on some serpentinite outcrops, soils at the base of steep slopes have strongly divergent chemistry from the soils closer to the top of the slope, despite their common geological parent material (Rajakaruna and Bohm 1999), Endemism on gabbro-derived soils of the Pine Hill intrusive complex, as well as the presence on these soils of many taxa normally restricted to serpentinite-derived substrates, have been attrib- uted to similar properties in gabbro-derived as compared to serpentinite-derived soils (Wilson 1986). Soils derived from serpentinite contain little Ca relative to Mg, and are rich in heavy metals such as Ni, Cr, and Co (Kruckeberg 2002). Gabbro rock itself is usually rich in heavy metals and tends to contain little Ca relative to Mg, although these parameters are not as extreme in gabbro as in serpentinite (Alexander 1993, unpublished). Research on the Pine Hill intrusive complex, however, found that the gabbro-derived soils from this area do not contain unusually low levels of Ca relative to Mg, or elevated heavy metals (Hunter and Horenstein 1991), results that are corroborated by regional geochemical studies (Goldhaber et al. 2009; Morrison et al. 2009). A later study focused on the gabbro-endemic plants of the Pine Hill intrusive complex asked whether soils from locations harboring endemics had low Ca to Mg ratios, or differences in a suite of other chemical and physical parameters, compared to areas without these plants (Alexander, unpub- lished). This study did not detect significant differences in Ca to Mg ratio between sites harboring rare plants versus those without, and 18 MADRONO failed to identify other parameters that might explain the differences in plant distribution. However, it is possible that the results of this study were confounded by plant demography. This may be especially true of C. roderickii, which depends on fire for recruitment (Boyd 2007). Although the present study did not focus on the contrast between serpentinite and gabbro, our results show that Ca to Mg ratios in serpentinite- derived soils of C. cuneatus (average 0.6 + 0.3) are closest to those in the exclusively gabbro- derived soils of C. roderickii (average 2.9 + 0.6). Values become successively higher in gabbro- derived soils of C. cuneatus var. cuneatus (5.5 + 1.5), and “‘other’’ (non-gabbro and non-serpenti- nite derived) soils of C. cuneatus (7.2 + 4.1). Although soils of C. roderickii have Ca to Mg ratios that are closest to those in serpentinite- derived soils, ratios in serpentinite-derived soils are still significantly lower (Student’s paired t-test, P < 0.001). Nevertheless, serpentinite- derived soils associate closely with the exclusively gabbro-derived soils of C. roderickii in PCA (Fig. SA, C). Furthermore, the two groups are not significantly different in terms of their scores on these axes (Tukey’s HSD test, P = 0.489), indicating that the serpentinite-derived soils are similarly nutrient deficient. Overall, nutrient deficiency and low Ca to Mg ratios may provide an explanation for the evolution of endemics on some gabbro-derived soils of the Pine Hill intrusive complex, and the presence on these soils of plants that are usually restricted to serpentinite-derived substrates (Wilson 1986). Evolution of Edaphic Ecology Evolution of the gabbro-endemic C. roderickii appears to have been associated with specializa- tion to strongly nutrient-deficient forms of gabbro-derived soil. The closest relative of C. roderickii, C. cuneatus var. cuneatus, has a very wide distribution in the California Floristic Province (Fig. 2), and is a common component of chaparral habitats in the Sierra Nevada. On the Pine Hill intrusive complex of western El Dorado Co., California, C. cuneatus var. cuneatus occupies nutrient-rich forms of gabbro-derived soils in close geographic proximity to the poorer forms favored by C. roderickii, sometimes no more than 100 m distant from the latter species (Fig. 1). Although there is not a well-supported “‘pro- genitor-derivative” relationship (Gottlieb 2003; Baldwin 2005) between C. cuneatus var. cuneatus and C. roderickii, the nested position of C. roderickii within a large group of C. cuneatus var. cuneatus individuals collected predominantly in the Sierra Nevada, Transverse Ranges, and Peninsular Ranges is suggestive of this pattern (Figs. 3, 4). Rocks of the Pine Hill intrusive [Vol. 58 complex have probably been exposed since Eocene time (J. Wakabayashi, personal commu- nication). Thus, it is possible that during the diversification of Ceanothus in western North America, which began approximately 5 mya (Ackerly et al. 2006; Burge et al. in press), C. cuneatus var. cuneatus colonized the Pine Hill region and gave rise to C. roderickii through specialization to the nutrient-poor forms of gabbro-derived soil. Because intrinsic (pre-zygotic) barriers to gene flow are not known in Cerastes (Nobs 1963), it is expected that hybridization will occur when different species come into contact with one another (Fross and Wilken 2006), potentially leading to gene flow and introgression. However, C. roderickii persists as a relatively genetically isolated, morphologically divergent entity in spite of its close proximity to C. cuneatus var. cuneatus on the Pine Hill intrusive complex (Fig. 1). One possible explanation for the lack of introgression 1s the action of environmental isolating factors. The fact that soil chemistry associations of C. cuneatus var. cuneatus and C. roderickii are most divergent where the taxa come into close contact on gabbro outcrops, with C. cuneatus var. cuneatus occupying comparatively nutrient-rich forms of gabbro- derived soil, is suggestive of character displace- ment and possibly reinforcement based on soil- chemistry (Levin 1970). Overall, edaphically- based barriers to gene-flow might provide an explanation for the initial divergence and contin- ued persistence of C. roderickii, as well as other edaphic-endemic Cerastes taxa. ACKNOWLEDGMENTS The authors thank Lauren Fety, Albert Franklin, Graciela Hinshaw, Sandra Namoff, and Dieter Wilken for providing constructive criticism of drafts. The authors also would like to thank Bonnie McGill, Kaila Davis, Sang-Hun Oh, and Sandy Bowles for assistance with lab work and development of methods. Assistance with field logistics was provided by Lauren Fety, Graciela Hinshaw, Sandra Namoff, and Dieter Wilken. Funding was provided by the American Society of Plant Taxonomists, The Hunt Institute for Botanical Docu- mentation, Duke University, and a National Science Foundation grant to DOB and PSM (DEB 0808427). LITERATURE CITED ACKERLY, D. D., D. W. SCHWILK, AND C. O. WEBB. 2006. Niche evolution and adaptive radiation: testing the order of trait divergence. Ecology 87:S50-61. ALEXANDER, E. B. 1993. Gabbro and its soils. Fremontia 21:8—10. BALDWIN, B. G. 2005. Origin of the serpentine-endemic Layia discoidea from the widespread L. glandulosa (Compositae). Evolution 59:2473—2479. Boyb, R. S. 2007. Response to fire of Ceanothus roderickii (Rhamnaceae), a federally endangered California endemic shrub. Madrono 54:13-21. 2011] BRADY, N. C. AND R. R. WEIL. 2002. The nature and properties of soils. Thirteenth Ed. Pearson Educa- tion, Upper Saddle River, NJ. BURGE, D. O., D. M. ERWIN, M. B. ISLAM, J. KELLERMANN, S. S. KEMBEL, P. S. MANOS, AND D. H. WILKEN. In Press. Diversification of the genus Ceanothus (Rhamnaceae) in the California Floristic Province. International Journal of Plant Sciences. CLEMENT, M., D. POSADA, AND K. A. CRANDALL. 2000. TCS: a computer program to estimate gene genealogies. Molecular Ecology 9:1657—1659. EDGAR, R. C. 2004. MUSCLE: multiple sequence alignment with high accuracy and high throughput. Nucleic Acids Research 32:1792—1797. FISHER, R. A. 1918. The correlation between relatives on the supposition of Mendelian inheritance. Philosophical Transactions of the Royal Society of Edinburgh 52:399_433. Fross, D. AND D. WILKEN. 2006. Ceanothus. Timber Press, Portland, OR. GANKIN, R. AND J. MAJor. 1964. Arctostaphylos myrtifolia, its biology and relationship to the problem of endemism. Ecology 45:792-808. GOLDHABER, M. B., J. M. MORRISON, J. M. HOLLO- WAY, R. B. WANTy, D. R. HELSEL, AND D. B. SMITH. 2009. A regional soil and sediment geo- chemical study in northern California. Applied Geochemistry 24:1482-1499. GOSLEE, S. C. AND D. L. URBAN. 2007. The ecodist package for dissimilarity-based analysis of ecolog- ical data. Journal of Statistical Software 22:1—19. GOTTLIEB, L. D. 2004. Rethinking classic examples of recent speciation in plants. New Phytologist 161:71-82. HARDIG, T. M., P. S. SOLTIS, AND D. E. SOLTIS. 2000. Diversification of the North American shrub genus Ceanothus (Rhamnaceae): conflicting phylogenies from nuclear ribosomal DNA and chloroplast DNA. American Journal of Botany 87:108—123. ; , , AND R. B. HUDSON. 2002. Morphological and molecular analysis of putative hybrid speciation in Ceanothus (Rhamnaceae). Systematic Botany 27:734-746. HICKMAN, J. C. (ed.). 1993. The Jepson Manual: higher plants of California. University of California Press, Berkeley, CA. HINSHAW, G. 2008. Pine Hill Preserve management plan. U.S. Department of the Interior. Bureau of Land Management, Mother Lode Field Office, El Dorado Hills CA. Website: http://www.blm.gov/ pgdata/etc/medialib/blm/ca/pdf/folsom/plans.Par. 44114. File.dat/FINAL_PHPmgmtPlan0708. pdf [accessed 10 April 2011]. HOWARTH, D. G. AND D. A. BAUM. 2002. Phylogenetic utility of a nuclear intron from nitrate reductase for the study of closely related plant species. Molecular Phylogenetics and Evolution 23:525—528. HUNTER, J. C. AND J. E. HORENSTEIN. 1991. The vegetation of the Pine Hill area (California) and its relation to substratum. Pp. 197-206 in A. J. M. Baker, J. Proctor, and R. D. Reeves, (eds.), The vegetation of ultramafic (serpentine) soils. Inter- cept, Andover, U.K. JAMES, S. C. 1996. A demographic study of Ceanothus roderickii (the Pine Hill ceanothus) El Dorado County, California. M.S. Thesis. California State University, Sacramento, CA. BURGE AND MANOS: EDAPHIC ECOLOGY OF CEANOTHUS RODERICKII pe) JENNINGS, C. W. 1977. Geologic map of California: California Division of Mines and Geology Geo- logic Data Map No. 2, scale 1:750,000. Website http://www.conservation.ca.gov/cgs/cgs_history/ Pages/geologicmaps.aspx [accessed 11 March 201 1]. KEENEY, D. R. AND D. W. NELSON. 1982. Nitrogen - inorganic forms. Pp. 643-687 in A. L. Page, T. H. Miller, and D. R. Keeney, (eds.), Methods of soil analysis, part 2 (second ed.). American Society of Agronomy, Madison, WI. KNIGHT, W. 1968. A new California Ceanothus. The Four Seasons 2:23—24. KRUCKEBERG, A. R. 1986. An essay: the stimulus of unusual geologies for plant speciation. New Phy- tologist 11:455—463. . 2002. Geology and plant life. University of Washington Press, Seattle, WA. LEVIN, D. A. 1970. Reinforcement of reproductive isolation: plants versus animals. The American Naturalist 940:571—581. LINDSAY, W. L. AND W. A. NORVELL. 1978. Develop- ment of a DTPA soil test for zinc, iron, manganese, and copper. Soil Sciences Society of America Journal 42:421—428. MASON, H. L. 1946. The edaphic factor in narrow endemism. I. The nature of environmental influ- ences. Madrono 8:209-240. McCMINN, H. E. 1942. A systematic study of the genus Ceanothus. Pp. 131-279 in Ceanothus Part II. Santa Barbara Botanical Gardens, Santa Barbara, CA. MEHLICH, A. 1978. New extractant for soil test evaluation of phosphorus, potassium, magnesium, calcium, sodium, manganese and zinc. Communi- cations in Soil Science and Plant Analysis 9:477-492. . 1984. Mehlich-3 soil test extractant: a modifi- cation of Mehlich 2 extractant. Communications in Soil Science and Plant Analysis 15:1409-1416. MorRISON, J. M., M. B. GOLDHABER, L. LEE, J. M. HOLLOWAY, R. B. WANTY, R. E. WOLF, AND J. F. RANVILLE. 2009. A regional-scale study of chro- mium and nickel in soils of northern California, USA. Applied Geochemistry 24:1500—1511. MYERS, N., R. A. MITTERMEIER, C. G. MITTERMEIER, G. A. B. DA FONSECA, AND J. KENT. 2000. Biodiversity hotspots for conservation priorities. Nature 403:853—858. Noss, M. A. 1963. Experimental studies on species relationships in Ceanothus. Carnegie Institute of Washington, Washington, D.C. NYLANDER, J. A. A. 2004. MrModeltest, v2. Program distributed by the author, Evolutionary Biology Centre, Uppsala University, Sweden. PEARSON, K. 1901. On lines and planes of closest fit to systems of points in space. Philosophical Magazine 2:559-572. R DEVELOPMENT CORE TEAM. 2009. R: a language and environment for statistical computing. R Founda- tion for Statistical Computing, Vienna, Austria. Website http://cran.r-project.org [accessed 11] March 2011]. RAJAKARUNA, N. 2004. The edaphic factor in the origin of plant species. International Geology Review 46:471—-478. AND B. A. BOHM. 1999. The edaphic factor and patterns of variation in Lasthenia_ californica (Asteraceae). American Journal of Botany 86:1576—-1596. 20 MADRONO RAVEN, P. H. AND D. I. AXELROD. 1978. Origin and Relationships of the California Flora. University of California Press, Berkeley, CA. RHOADES, J. D. 1982. Soluble salts. Pp. 167-178 in A. L. Page, T. H. Miller, and D. R. Keeney, (eds.), Methods of soil analysis, part 2 (second ed.). American Society of Agronomy, Madison, WI. RoGERS, J. H. 1974. Soil survey of El Dorado area, California. U.S. Government Printing Office, Washington, D.C. RONQUIST, F. AND J. P. HUELSENBECK. 2003. MrBayes 3: Bayesian phylogenetic inference under mixed models. Bioinformatics 19:1572—1574. SCHOFIELD, R. K. AND A. W. TAYLOR. 1955. The measurement of soil pH. Soil Science Society of America Proceeding 19:164—-167. SPRINGER, R. K. 1971. Geology of the Pine Hill intrusive complex, El Dorado County, California. Ph.D. dissertation. University of California at Davis, Davis, CA. 1980. Geology of the Pine Hill intrusive complex, a layered gabbroic body in the western Sierra Nevada foothills, California: summary. Geological Society of America Bulletin, Part I 91:381-385. STEBBINS, G. L. 1942. The genetic approach to problems of rare and endemic taxa. Madrono 6:241—-258. STUDENT. 1908. The probable error of a mean. Biome- trika 6:1—25. TEMPLETON, A. R., K. A. CRANDALL, AND C. F. SING. 1992. A cladistic analysis of phenotypic associa- tions with haplotypes inferred from restriction endonuclease mapping and DNA sequence data. Ill. Cladogram estimation. Genetics 132:619—633. UNITED STATES DEPARTMENT OF AGRICULTURE, NATURAL RESOURCE CONSERVATION SERVICE. 2007. Soil Survey Geographic (SSURGO) database for El Dorado Area, California. U.S. Department of Agriculture, Natural Resources Conservation Service, Fort Worth, TX. UNITED STATES FISH AND WILDLIFE SERVICE (USFWS). 1996. Endangered and threatened wild- life and plants; determination of endangered status for four plants and threatened status for one plant from the central Sierran foothills of California. Federal Register 61:54346—-54358. WILSON, J. L. 1986. A study of plant species diversity and vegetation patterns associated with the Pine Hill gabbro formation and adjacent substrata, El Dorado County, California. M.S. thesis. California State University, Sacramento, Sacramento, CA. , D. R. AYRES, S. STEINMAUS, AND M. BAAD. 2009. Vegetation and flora of a_ biodiversity hotspot: Pine Hill, El Dorado County, California, USA. Madrono 56:246—-278. ZAR, J. H. 1999. Biostatistical analysis. Prentice Hall, Upper Saddle River, NJ. APPENDIX | SAMPLED CEANOTHUS POPULATIONS GenBank accession numbers for the first and second NIA sequence (where available) are in brackets []. See Table 2 for additional population information. [Vol. 58 Ceanothus cuneatus Nutt. var. cuneatus—USA. CAL- IFORNIA. Amador Co.: Grass Valley Creek watershed, NE of Mount Zion, D.O. Burge 1150a (DUKE) [NIA: HM 240330; HM240329]. Butte Co.: Feather Falls, D.O. Burge 1109a (DUKE); Doe Mill Ridge, D.O. Burge S15a (DUKE); Magalia Reservoir, D.O. Burge 1078a (DUKE) [NIA: HM240306; HM240307]. Calaveras Co.: North Fork Calaveras River watershed, NE of Golden Gate Hill (VABM 2064), D.O. Burge 1149a (DUKE) [NIA: HM240327; HM240328]. Contra Costa Co.: Mount Diablo State Park, roadside on South Gate Rd, D.O. Burge 9l6a (DUKE) [NIA: HM240341; HM 240342]. El Dorado Co.: Roadside on Wentworth Springs Rd, 1.9 road mi (3.0 km) from intersection with SR 193, D.O. Burge 101 1a (DUKE); Green Valley Rd, D.O. Burge 1024a (DUKE); Folsom Lake Watershed, W side of North Fork American River arm, S slope of Kelly Ravine, D.O. Burge 1074a (DUKE); Martinez Creek watershed, roadside on Pleasant Valley Rd, D.O. Burge 1174a (DUKE); Weber Creek watershed, road- side on Lotus Rd, D.O. Burge 1076a (DUKE); South Fork American River watershed, D.O. Burge 1088a (DUKE); City of Cameron Park, D.O. Burge 1089a (DUKE); S side of U.S. Hwy 50, between Durock Rd and U.S. Hwy 50, D.O. Burge 110/a (DUKE); Pine Hill, eastern slope, D.O. Burge 1116a (DUKE); South Fork American River watershed, Dave Moore Nature Area, D.O. Burge 1175a (DUKE); S side of U.S. Hwy 50, between Durock Rd and Hwy 50, D.O. Burge 1023a (DUKE) [NIA: HM240297; HM240296]; Tennessee Creek watershed, roadside on Shingle Springs Rd, D.O. Burge 1075a (DUKE) [NIA: HM240303; HM240302]; Shingle Creek watershed, S of the city of Cameron Park, D.O. Burge 1095a (DUKE) [NIA: HM240314; HM7240315]; S shore of Bass Lake, D.O. Burge 1110a (DUKE) [NIA: HM240316]; South Fork American River watershed, Icehouse Rd, D.O. Burge I1117a (DUKE) [NIA: HM240318; HM240317]. Fresno Co.: Dalton Mountain, south-eastern slope, head of Tret- ten Canyon, D.O. Burge 1136a (DUKE) [NIA: HM 240323]. Kern Co.: Clear Creek watershed, S of Ball Mountain and SE of Hooper Hill, D.O. Burge 1132a (DUKE) [NIA: HM240319; HM240320]. Lake Co.: Mayacmas Mountains, Cow Mountain Recreation Area, Fourmile Glade, D.O. Burge 1008a (DUKE) [NIA: HM240295]. Los Angeles Co.: Sierra Pelona Mountians, Ruby Canyon, D.O. Burge 1071a (DUKE) [NIA: HM240301]. Mariposa Co.: Chowchilla River watershed, East Fork, N of Miami Mountain and E of Paloni Mountain, D.O. Burge 1140a (DUKE) [NIA: HM 240324]. Monterey Co.: Nacimiento-Fergusson Rd, D.O. Burge 858a (DUKE) [NIA: HM240338]. Napa Co.: Vaca Mountains, on east-west trending ridge S of East Mitchel Canyon, D.O. Burge 899a (DUKE) [NIA: HM 240339; HM240340]. Nevada Co.: Community of Hills Flat, near the City of Grass Valley, D.O. Burge 1084a (DUKE) [NIA: HM240310]. Placer Co.: North Fork American River Watershed, Forest Hill Divide, D.O. Burge 1077a (DUKE) [NIA: HM240305; HM240304]. Riverside Co.: San Jacinto Mountains, at intersection of Chimney Flats Rd and USFS Rd 5S13, D.O. Burge 803a (DUKE); Tucalota Creek watershed, roadside on Sage Rd (County Rd R3), D.O. Burge 982a (DUKE) [NIA: HM240344; HM240345]. Sacramento Co.: American River watershed, near outlet of Willow Creek into Lake Natoma, D.O. Burge 1094a (DUKE) [NIA: HM240313]. San Bernardino Co.: Rialto Munic- ipal Airport (Miro Field), D.O. Burge 1070a (DUKE) 2011] [NIA: HM240300]. San Diego Co.: Morena Valley, roadside on Buckman Springs Rd, D.O. Burge 984a (DUKE) [NIA: HM240346]. San Luis Obispo Co.: Santa Lucia Mountains, Arroyo Grande Creek water- shed, NW of Arroyo Grande, D.O. Burge 959a (DUKE) [NIA: HM240343]. Shasta Co.: Crystal Creek watershed, N of Crystal Creek Rd, D.O. Burge 11]5la (DUKE) [NIA: HM240331; HM240332]. Sierra Co.: Goodyears Bar, near confluence of Goodyears Creek and North Yuba River, D.O. Burge 1083a (DUKE) [NIA: HM240308; HM240309]. Sutter Co.: Sutter Buttes, Peace Valley, D.O. Burge 1093a (DUKE) [NIA: HM240312; HM240311]. Tehama Co.: Paynes Creek watershed, immediately W of Palmer Gulch, D.O. Burge 1168a (DUKE) [NIA: HM240336]. Tulare Co.: Middle Fork Tule River, roadside on SR 190, D.O. Burge 1134a (DUKE) [NIA: HM240322; HM240321]. Tuolumne Co.: Red Hills, SW of Taylor Hill, D.O. Burge 1145a (DUKE) [NIA: HM240326; HM240325]. OREGON. Douglas Co.: South Umpqua River water- shed, roadside on Dole Drive, D.O. Burge 116la (DUKE) [NIA: HM240333]. Jackson Co.: Cottonwood Creek watershed, D.O. Burge 1164a (DUKE) [NIA: HM 240334; HM240335]. MEXICO. Baja CA: Sierra San Pedro Martir, Los Llanitos, D.O. Burge 1030a (DUKE) [NIA: HM240298; HM240299]; Sierra San Pedro Martir, 40.4 road mi (64.6 km) E of Mexico Hwy 1, D.O. Burge 783a (DUKE) [NIA: HM240337]. Ceanothus cuneatus Nutt. var. dubius J.T. Howell— USA. CALIFORNIA. Santa Cruz Co.: Henry Cowell Redwoods State Park, D.O. Burge 918a (DUKE) [NIA: HM 240347]. Ceanothus cuneatus Nutt. var. fascicularis (McMinn) Hoover—USA. CALIFORNIA. Santa Barbara Co.: Vandenberg Village, D.O. Burge 87la (DUKE) [NIA: HM 240348]. Ceanothus cuneatus Nutt. var. ramulosus Greene— USA. CALIFORNIA. San Luis Obispo Co.: Prefumo Canyon, D.O. Burge 847b (DUKE) [NIA: HM240349]. Ceanothus cuneatus Nutt. var. rigidus (Nutt.) Ho- over—USA. CALIFORNIA. Monterey Co.: Fort Ord Military Reservation, on hillside W of South Boundary Rd, D.O. Burge 89/b (DUKE) [NIA: HM240351:; HM 240350]. Ceanothus divergens Parry subsp. confusus (J.T. Howell) Abrams—USA. CALIFORNIA. Sonoma Co.: Mayacmas Mountains, western slope of Mount Hood, D. O. Burge 1003a (DUKE) [NIA: HM240352; HM240353]. Ceanothus divergens Parry subsp. occidentalis (McMinn) Abrams—USA. CALIFORNIA. Lake Co.: Boggs Mountain Demonstration State Forest, D.O. Burge 943a (DUKE) [NIA: HM240354]. Ceanothus ferrisiae McMinn—USA. CALIFORNIA. Santa Clara Co.: Pigeon Point, D.O. Burge 834a (DUKE) [NIA: HM240356; HM240355]. Ceanothus fresnensis Abrams—USA. CALIFOR- NIA. Fresno Co.: Big Creek Watershed, E flank of north-south trending ridge W of Ely Mountain, D.O. Burge 1138a (DUKE) [NIA: HM240357]. Ceanothus gloriosus J.T. Howell var. exaltatus J.T. Howell—USA. CALIFORNIA. Mendocino Co.: Oil- well Hill, near the N end of Little Lake Valley, D.O. Burge 994a (DUKE) [NIA: HM240358; HM240359]. BURGE AND MANOS: EDAPHIC ECOLOGY OF CEANOTHUS RODERICKII pa Ceanothus gloriosus J.T. Howell var. gloriosus— USA. CALIFORNIA. Marin Co.: Point Reyes Nation- al Seashore, D.O. Burge 908a (DUKE) [NIA: HM 240361; HM240360]. Ceanothus gloriosus J.T. Howell var. porrectus J.T. Howell—USA. CALIFORNIA. Marin Co.: Point Reyes National Seashore, Inverness Ridge, D.O. Burge 907a (DUKE) [NIA: HM240362; HM240363]. Ceanothus jepsonii Greene var. albiflorus J.T. Ho- well—USA. CALIFORNIA. Colusa Co.: Rathburn- Petray Mine, D.O. Burge 997a (DUKE) [NIA: HM 240364; HM240365]. Ceanothus jepsonii Greene var. jepsonii—USA. CAL- IFORNIA. Marin Co.: Alpine Lake, D.O. Burge 914a (DUKE) [NIA: HM240366]. Ceanothus maritimus Hoover—USA. CALIFOR- NIA. San Luis Obispo Co.: Roadside on Hwy 1, 0.5 road mi (0.8 km) N of bridge over Arroyo de los Chinos, D.O, Burge 887a (DUKE) [NIA: HM240367]. Ceanothus masontii McMinn—USA. CALIFORNIA. Marin Co.: Golden Gate National Recreation Area, Bolinas Ridge, D.O. Burge 91/3a (DUKE) [NIA: HM240368]. Ceanothus pinetorum Coville—USA. CALIFORNIA. Trinity Co.: Un-named rd along ridge, Trinity-Shasta County line, ca. 2.2 linear km SSE of Hoadley Peaks, D.H. Wilken 16736 (DUKE) [NIA: HM240369; HM240370]. Ceanothus prostratus Benth.—USA. CALIFORNIA. El Dorado Co.: El Dorado National Forest, roadside on Wentworth Road, D.O. Burge 952a (DUKE) [NIA: HM 240371]. Ceanothus pumilus Greene—USA. CALIFORNIA. Del Norte Co.: Smith River watershed, near the confluence of Middle Fork Smith River and North Fork Smith River, D.O. Burge 1156a (DUKE) [NIA: HM 240372; HM240373]. Ceanothus purpureus Jeps.—USA. CALIFORNIA. Napa Co.: Wooden Grade, NE of Mount George, D.O. Burge 904a (DUKE) [NIA: HM240374; HM240375]. Ceanothus roderickii W. Knight—USA. CALIFOR- NIA. El Dorado Co.: Pine Hill, just E of summit, D.O. Burge 1080a (DUKE) [NIA: HM240376]; South Fork American river canyon, near confluence with Weber Creek, D.O. Burge 1087a (DUKE) [NIA: HM240377; HM 240378]; City of Cameron Park, N side of U.S. Hwy 50, D.O. Burge 1090a (DUKE); City of Cameron Park, E of Cameron Airpark, D.O. Burge 1096a (DUKE); City of Cameron Park, D.O. Burge 1100a (DUKE); S side of U.S. Hwy 50, between Durock Rd and U.S. Hwy 50, D.O. Burge 1102a (DUKE); South Fork American River watershed, NW of Mormon Hill, D.O. Burge 1104a (DUKE); South Fork American River watershed, NW of Mormon Hill, D.O. Burge 110S5a (DUKE); Kelley Creek watershed, roadside on Sierrama Rd, D.O. Burge J111 (DUKE) [NIA: HM240379]; City of Cameron Park, N side of U‘S. Hwy 50, Bureau of Land Management Pine Hill Preserve, D.O. Burge 117la (DUKE): Cameron Park, D.O. Burge 824b (DUKE) [NIA: HM240380]. Ceanothus sonomensis J.T. Howell—USA. CALI- FORNIA. Sonoma Co.: Mayacmas Mountains, head of Hooker Canyon, D.O. Burge 895b (DUKE) [NIA: HM 240381]. MADRONO, Vol. 58, No. 1, pp. 22-31, 2011 POLLINATION BIOLOGY OF DARLINGTONIA CALIFORNICA (SARRACENIACEAB), THE CALIFORNIA PITCHER PLANT GEORGE A. MEINDL' AND MICHAEL R. MESLER Department of Biological Sciences, Humboldt State University, Arcata, CA 95521 gam35@pitt.edu ABSTRACT The pollination ecology of Darlingtonia californica Torr., including especially the identity of its pollinators, has remained enigmatic for more than a century. The flowers of this well-known charismatic species are unusual in form and color, and have been the subject of much speculation. Accordingly, in this study we sought to identify D. californica’s floral visitors and determine their potential effectiveness as pollinators, in the context of D. californica’s unusual floral morphology. We also used hand-pollinations and emasculations to determine whether plants were pollen-limited at five study sites in northwest California, and to evaluate the potential for self-pollination in natural populations of D. californica. A generalist solitary bee, Andrena nigrihirta, visited and pollinated D. californica flowers at five sites in northern CA. Despite very low visitation rates, individual flowers at all study sites were predicted to receive at least one visit by A. nigrihirta. Other regular floral visitors included thrips and several species of spiders. Plants at all five study populations were found to be pollen-limited with respect to the number of seeds produced per capsule. Fruit and seed production by emasculated flowers indicated a large degree of cross-pollination. However, emasculated flowers did not produce as many fruits and seeds as unmanipulated flowers, suggesting that self-pollination contributes to D. californica reproductive success as well. Observations of A. nigrihirta on flowers revealed that the shape and orientation of D. californica’s ovary and petals promote stigma contact both when pollinators enter and exit a flower, contrary to previous thought. Our findings provide evidence that D. californica is melittophilous, and suggest a resolution of the long-standing mystery surrounding the pollination of this rare species. Key Words: Andrena nigrihirta, autogamy, Darlingtonia californica, pollination, xenogamy. The only thing we are lacking is a pollinator. (Schnell 1976) The study of the interactions between plants and their pollinators can provide adaptive explanations for floral traits (Harder and John- son 2009). Careful observation of flower patches —the essential first step in the process—usually generates a list of flower visitors, at least some of which are pollinators. Once the pollinators are known, adaptive hypotheses can be proposed based on an understanding of the biology of the animals as well as the ecological and the phylogenetic context. In spite of the crucial importance of “knowing the pollinators”, the pollinator assemblages of a surprising number of plant species remain poorly or entirely unknown. The California pitcher plant, Darlingtonia cali- fornica Torr., 1s a case in point. The flowers of this well-known charismatic species are unusual in form and color, and have been the subject of much speculation (Debuhr 1973; Schnell 1976). For example, some have theorized that the bell- shaped ovary serves to limit self-pollination by directing pollinators away from the stigmas as they exit the flowers (Schnell 1976), but very few 'Present address: Department of Biological Sciences, University of Pittsburgh, Pittsburgh, PA 15260-3929. reports of pollinator visits exist despite serious interest from several workers (Austin 1875-1877 in Juniper et al. 1989; Elder 1997; Nyoka and Ferguson 1999; Nyoka 2000; Rice 2006). Pub- lished observations of flower handling by polli- nators are lacking. In fact, the pollination ecology of this plant, including especially the identity of its pollinators, has remained enigmatic for more than a century. The paucity of pollinator sightings is perplex- ing because fruit set in natural populations of D. californica is relatively high, and flowers do not self-pollinate autonomously (Elder 1997; Nyoka 2000). Based on appearances, the flowers of D. californica seem adapted for pollination by bees. They are large, showy, sweetly fragrant, and produce abundant pollen (Debuhr 1973; Nyoka and Ferguson 1999)—all features commonly associated with melittophily (Waser 2006). In — addition, D. californica’s sister taxa, Sarracenia and Heliamphora spp., are pollinated predomi- nantly by bumble bees (Thomas and Cameron | 1986; Renner 1989; Ne’eman et al. 2006), | suggesting that bee pollination may be primitive for Sarraceniaceae. Nevertheless, bees have sel- dom been observed as visitors to D. californica | flowers (Austin 1875-1877 in Juniper et al. 1989; | Elder 1997; Nyoka and Ferguson 1999; Nyoka © 2000; Rice 2006). Spiders, in contrast, commonly — 2011] use the flowers as hunting grounds (Austin 1875— 1877 in Juniper et al. 1989; Elder 1997; Nyoka 2000). Although arachnids have generally not been given serious consideration as pollinators, Nyoka (2000) noted that they frequently con- structed webs and stalked prey inside D. califor- nica flowers and carried pollen on their bodies. By experimentally introducing spiders to bagged flowers, she showed that they can cause autog- amy. She also found spiders carrying D. califor- nica pollen outside of tlowers, and detected fluorescent dye particles on spider draglines indicating their potential as cross-pollinators. The hypothesis that the flowers of D. califor- nica are pollinated by spiders is appealing, partly because it would explain why previous workers have seldom seen flying pollinators. However, spider pollination is problematic for at least two reasons. First, the effectiveness of spiders as pollinators may not be sufficient to account for observed levels of fruit and seed set. At her study site in southern Oregon, Nyoka (2000) found that fruit set of open-pollinated flowers approached 100%, and, on average, capsules produced more than 900 seeds. In contrast, fruit set of flowers bagged with spiders was less than 50%, and capsules produced 95% fewer seeds than open- pollinated flowers. This discrepancy implies that other visitors, perhaps bees, play a more impor- tant role as pollinators. Second, the morpholog- ical fit between spider and flower seems weak at best, making it difficult to conjure plausible adaptive explanations. In fact, the traits that create suitable conditions for spiders, such as the unusually long duration of anthesis (up to 48 days) and the protective tent-like corolla, may be adaptations for bee pollination. Occupa- tion and occasional pollination of D. californica flowers by spiders may be incidental and secondary. In spite of the long history of interest in the pollination of D. californica, to date there has been only one published account of a thorough, systematic survey for flower visitors (Nyoka 2000), making it premature to conclude that bees play little cr no role in the pollination of the species. Here we report the results of extensive pollinator surveys at five sites in northwestern California. We combine these observations with results of experimental pollination treatments designed to estimate the degree of pollination limitation at these sites as well as the relative importance of self- vs. cross-pollination. The later should provide insight into the relative impor- tance of spiders and bees as pollen vectors, assuming that spiders mainly cause self-pollina- tion. We addressed four specific questions: (1) Who are the most important floral visitors, and are they capable of effecting pollination? (2) Is floral visitation by effective pollinators frequent in natural populations, i.e., is natural pollination MEINDL AND MESLER: DARLINGTONIA CALIFORNICA POLLINATION 23 sufficient, or are plants experiencing pollen- limitation? (3) Do cross-pollination and _ self- pollination each contribute to natural pollina- tion? (4) Are past interpretations of the function- al morphology of floral traits correct, 1.e., does the ovary shape limit self-pollination as has been suggested (Schnell 1976)? METHODS Study Species Darlingtonia californica is a carnivorous plant endemic to western Oregon and northern Cali- fornia. Its distribution across this range is patchy, being restricted to perennial wet seeps, generally on serpentine soils (Juniper et al. 1989; Schnell 1976; Whittaker 1954). A long-lived perennial, D. californica produces rosettes of leaves from a creeping rhizome every year. Plants often occur in dense patches, which likely result from clonal spread by rhizomes and stolons (Schnell 1976). The solitary flowers begin as upright buds, but become pendant when mature (Debuhr 1973). Unlike some Sarracenia (Ne’eman et al. 2006), the flowers of D. californica produce no nectar (Debuhr 1973). Abundant pollen is the only likely reward for pollinators, though a sugar-rich stigmatic exudate may also attract visitors (Nyoka 2000). Five lanceolate-ovate, yellow- green sepals hang loosely around five crimson petals. The five petals almost completely enclose the reproductive whorls, except for windows formed by notches in adjacent petals, which allow access to the flower’s interior. The windows are level with the five stigmatic lobes, a feature that has been predicted to promote the deposition of outcrossed pollen as pollinators initially enter a flower (Schnell 1976). Twelve to fifteen stamens are located at the base of the ovary. The bell- shaped ovary is flared towards the stigmas, which has been postulated to function to guide pollina- tors away from the stigmas as they exit a flower and thus limit self-pollination (Schnell 1976). Flowers mature into upright capsules capable of producing around 2000 seeds (Debuhr 1973). The flowers of D. californica are self-compatible, but are not autonomously autogamous (Elder 1997; Nyoka 2000). Study Sites Five seeps, located near Scott Mountain and Mt. Eddy, CA were used in this study (Table 1). The five study sites will hereafter be referred to as SM1, SM2, CL, N17, and DF. Distance between sites ranged from ~0.1 to 14.5 km. Near the border of Trinity and Siskiyou counties, this portion of the Klamath Bioregion represents the center of D. californica’s range (Debuhr 1973). Flowering occurred at all study populations 24 MADRONO TABLE 1. ELEVATION AND GEOGRAPHIC COORDI- NATES OF FIVE STUDY SITES. Site Elevation Spatial coordinates SM1 1635 m 41°16'25.00"N; 122°41'58.21”W SM2 1630 m 41°16'38.57"N; 122°41'57.54"W CL 1693 m 41°18'01.49"N; 122°40'59.90"W N17 1945 m 41°20'08.05"N; 122°31'41.53"W DF 2001 m 41°20'09.13"N; 122°31'11.39"W between June 12, 2008 and June 22, 2008, except for CL where flowering started earlier (June 6, 2008). A total of 51 angiosperm species, all with blooming periods that at least partially over- lapped that of D. californica, were present at the study sites (Meindl 2009). Within the study populations, common associates included white rushlily (Hastingsia alba S. Watson), California bog asphodel (Narthecium californicumBaker), Sierra shootingstar (Dodecatheon jeffreyi Van Houtte), marsh marigold (Caltha leptosepala DC var. biflora (DC) G. Lawson), and Bigelow’s sneezeweed (Helenium bigelovii A. Gray in Torr.). Flower Visitation and Pollinator Identification Three observation points were established in each seep in order to monitor pollinator activity. At these points a series of 15-minute surveys were conducted, focusing on 13-17 flowers at one time. Ten surveys (2.5 hours total) were conduct- ed during each day of observation at a field site. Each site was visited three to five times between June 6, 2008 and July 3, 2008 to conduct surveys. Most surveys were made between 10:00 a.m. and 6:00 p.m. In total, 57.5 hr of observations were conducted. Mean flower visitation rates (visits/ flower/hour) and the estimated number of visits individual flowers would receive over their lifetimes were calculated for each study site (Meindl 2009). The expected number of visits a flower received over its lifetime was estimated by multiplying the flowering period (in days) by the number of hours in a day pollinators were active (six hr) by the visits/hour calculated for each site. Darlingtonia californica pollinators were consid- ered to be active for six hours a day because all visits occurred between 10:30 a.m. and 4:30 p.m. Flower lifespan was determined by monitoring the development of 30 tagged buds at each study site (Meindl 2009). Following each 15-min census period, five flowers were carefully examined by spreading apart the sepals and petals to check for pollina- tors already present within the flowers. A total of 1125 flowers were inspected in this way for spiders, spider webs (either inside or outside the flower), fungus gnats, and thrips. Insects were captured by aerial netting or by hand, and identified. For bees collected within the genus Andrena, individuals were identified using keys [Vol. 58 and descriptions from Laberge and Ribble (1975) and compared against previously identified ref- erence specimens in the HSU invertebrate collec- tion. Vouchers of all collected pollinators have been deposited at HSU for future reference. Pollinator Behavior To determine if floral visitors carried D. californica pollen, each insect collected during surveys (n = 88) was systematically dabbed with a small cube of glycerin jelly containing basic fuchsin stain (Kearns and Inouye 1993). Follow- ing pollen removal, the jelly was placed on a microscope slide, melted and covered with a cover slip for analysis. Pollen grains were identified by comparing them to a reference collection prepared from flowers at each site. Darlingtonia californica pollen was readily distin- guishable from other pollen observed due to its unique morphology, which includes five elongate apertures extending from the grain walls. Polli- nators were collected from the flowers of D. californica, as well as other coflowering species, to determine which members of the pollinator community carried D. californica pollen. We could not determine the effectiveness of flower visitors directly, but instead recorded how often visitors gathered pollen and contacted stigmas, and how long they spent in flowers. A subset of observed floral visits was filmed with a digital camera. Along with other observed visits, the videos were analyzed to determine if pollina- tors handled the flowers in a manner that would result in pollination. These observations were also used to indicate whether or not the shape of D. californica’s ovary really serves to limit the occurrence of self-pollination. Pollination Sufficiency and Estimates of Cross-Pollination Hand-pollinations were performed to estimate pollination sufficiency. At each of the five study sites, 30 flowers were marked as controls and an additional 30 flowers were hand-pollinated. Supplemental pollen was applied twice (separated by one week) to flowers in the hand-pollinated treatment group by rubbing two-three mature anthers directly against stigmatic surfaces, when the appearance of stigmatic exudates indicated receptivity. Pollen used for hand-pollinations was collected from flowers at least five meters away in the same population. Fruit and seed set resulting from unmanipulated control flowers were com- pared against that of hand-pollinated flowers. If there is no difference in fruit set between these two treatment groups then we can conclude that natural pollination is sufficient, 1.e., plants were not pollen-limited. 2011] TABLE 2. FLORAL VISITATION RATES. The mean number of visits a flower was expected to receive per hour and over its lifetime is presented for each study site. Standard error values for mean visits/hour are given in parentheses for each study site. MEINDL AND MESLER: DARLINGTONIA CALIFORNICA POLLINATION 25 points were zeroes and thus the data set could not be adjusted to meet the assumption of normality. Log linear analysis was used to compare the fruit set of the three experimental treatment groups, with treatment, site and the interaction term included in the model. A two-way ANOVA was Estimated Site Visits/hour visits/lifetime used to compare seed set across all sites, with CL 0.016 (SE = 0.029) 1.60 treatment and site as the independent variables. SM1 0.041 (SE = 0.042) 71 Due to a significant interaction term from the SM2 0.073 (SE = 0.025) 4.84 two-way ANOVA (P = 0.041), separate one-way N17 0.077 (SE = 0.025) 5.08 ANOVAs were run for each site independently DF 0.067 (SE = 0.025) 4.42 using treatment as the independent variable. To gauge relative levels of cross-pollination vs. self-pollination, 30 flowers 1n each study popula- tion were emasculated prior to maturity. Fruits and seeds produced by flowers in the emasculated treatment group were interpreted to be the result of cross-pollination, whereas fruit and seed set by unmanipulated control flowers resulted from both cross-pollination and self-pollination. Thus the contribution of self-pollination to total pollination can be estimated by comparing the fruit and seed set of the emasculated flowers with the fruit and seed set of unmanipulated flowers. A total of 450 flowers were used for fruit and seed set experiments, with 150 flowers in each of the three treatments: hand-pollinated, emasculat- ed, and unmanipulated. These treatments were spread equally across the five study sites (1.e., 90 flowers at each site in 3 treatment groups of 30). Once fruit maturation began, all treatment flowers were bagged with Reemay® (Fiberweb, TN), a polyester fabric, to ensure seeds were not lost when capsules began to dehisce. Fruit set was determined for each site, as well as the number of seeds produced by each flower that matured a fruit. Statistical Analyses A Kruskal-Wallis one-way analysis of variance was used to compare average visitation rates across all sites. A non-parametric test was necessary to analyze visit rate data, as most data TABLE 3. Post-hoc Tukey-Kramer multiple-comparison tests were used to determine which group means were significantly different from one another. All statistical analyses were performed using NCSS (Hintze 2004). RESULTS Flower Visitation and Pollinator Identification In general, hymenopteran pollinators were abundant at our study sites, represented by eight genera of bees (Meindl 2009). However, D. californica received only 38 visits by flying pollinators in 57.5 hr of observations, and nearly all (37) were by a solitary bee, Andrena nigrihirta. One visit by a European honeybee (Apis melli- fera) was also observed. Estimated visit rates varied widely (Table 2), but were not significantly different across the five sites (Kruskal-Wallis xy? = 5.72, P = 0.22). Based on the average visit rate (pooled data across sites), flowers received 3.9 visits during their entire blooming period. Visits by A. nigrihirta were observed throughout the flowering season (6/13/09 through 6/22/09), and multiple visits were observed at each site (Meindl 2009). Spiders, particularly members of the families Clubionidae, Salticidae, and Theridiidae, were common on flowers at all five study sites, and were active at all hours of the day (Table 3). Whereas a minority of examined flowers con- tained a spider, the majority showed evidence of spider occupancy (webbing and/or spider present) THE PERCENTAGE OF EXAMINED FLOWERS AT EACH STUDY SITE THAT CONTAINED ONE OR MORE OF THE FOLLOWING: THRIPS, SPIDERS, AND SPIDER WEBS (EITHER INSIDE OR OUTSIDE THE FLOWER). A total of 1125 flowers were individually examined (150 at CL, 225 at SM1, 250 at SM2, 250 at N17, and 250 at DF). “Evidence of Spider” column represents the percentage of examined flowers at each site that had a spider and/or webbing present. Only 3/1125 (0.27%) flowers contained one or more fungus gnats. Site Web outside flw. Web inside flw. Spider present Evidence of spider Thrips CL Sheil 13.3 16.7 48.7 2253 SM1 48.9 39.6 20 61.8 40.9 SM2 47.2 30.4 24.8 S72 31.6 N17 69.6 20 34 74 TS0 DF 68.8 30.8 oil.2 75.6 58.4 TOTAL 56.2 Ziad 26.2 64.8 48.4 26 MADRONO (Table 3). Thrips were also present in large numbers at all five sites: nearly half of all examined flowers contained thrips actively for- aging for pollen (Table 3). Fungus gnats, while frequently encountered in the seeps, were only observed within D. californica flowers three times. Pollinator Behavior Individual bees spent up to several minutes within D. californica flowers and were found to carry D. californica pollen following visits. On average, A. nigrihirta foraged on a single D. californica flower for approximately two minutes and eight seconds (128 sec + 12 seconds; n = 14). Eight individuals were collected immediately following visits, and all carried D. californica pollen in their scopae. Of these, six carried D. californica pollen exclusively while two carried heterospecific pollen as well (Asteraceae). One individual of A. nigrihirta was collected in flight (i.e., not on a flower) that carried both D. californica and Asteraceae pollen. Andrena nigri- hirta was the only floral visitor collected that carried the poilen of D. californica (Meindl 2009). Detailed observations of visits by A. nigrihirta revealed that the ovary shape of D. californica promotes stigma contact by bees both when they enter and exit flowers (Fig. 1). Immediately above the windows (towards the morphological base of the pendant flower), the flower’s petals overlap and the underlying petal is appressed to the flared portion of the ovary, which limits the ability of a pollinator the size of A. nigrihirta to enter a window and crawl directly up onto the ovary on its way to collect pollen. In between the windows, however, the petals bulge outward (Fig. 2), and it is this space that allows the bee to ascend up to the stamens. This convex portion of each of the five petals is located directly opposite each of the five windows, such that a pollinator enters a window and walks in a straight line across the stigmas and then onto the ovary (directed by the convex portion of the petal). After ascending the ovary, bees were observed to systematically gather pollen before descending down the ovary towards the stigmas. The shape of D. californica’s ovary has previously been thought to guide an insect pollinator away from the receptive stigmatic surfaces as it exits the flower, thus preventing self-pollination. However, in exiting the flower, bees were observed to leave in the same fashion as they entered (guided by petal convexities across the stigmas and out one of the windows, thus likely effecting autogamy). This behavioral sequence was exhibited by multiple (n = 27) individuals and was consistent at all sites. These observations, plus evidence that A. nigrihirta carried the pollen of D. californica, strongly suggest these bees are acting as pollinators. [Vol. 58 Pollination Sufficiency and Estimates of Cross-Pollination Seed production, but not fruit set was pollen- limited. Fruit production by unmanipulated flowers (76%) was not significantly lower than that of hand-pollinated flowers (96%) (y* = 3.50, P = 0.06). However, hand-pollinated flowers produced more than twice as many seeds per capsule than unmanipulated flowers at each of the five study sites (Fig. 3). Self- and cross-pollination both contribute to D. californica reproductive success. Emasculated flowers produced fruit and seed at all five sites, indicating that cross-pollination occurred. How- ever, overall fruit set of emasculated flowers (39%) was significantly lower than that of unmanipulated flowers (y? = 17.79, P < 0.001), highlighting the importance of autogamous pollen transfer for fruit production. Unmanipu- lated flowers produced significantly more seeds, on average, than emasculated flowers at SM1 and SM2, but there was no significant difference found between these two treatment groups at the remaining three sites (Fig. 3). Average seed production by unmanipulated flowers was always higher than that of emasculated flowers, regard- less of statistical significance, suggesting that cross-pollination cannot account for all of the seeds that were produced. Therefore, fruit and seed production of naturally pollinated flowers were likely the result of both autogamous and xenogamous pollen transfer. DISCUSSION Near the summits of Scott Mountain and Mount Eddy in northwestern California, popu- lations of D. californica are pollinated by the solitary bee Andrena nigrihirta, with additional pollination likely provided by spiders. This conclusion is based on direct observations of floral visits, analysis of bee pollen loads, and the results of our pollination treatments. In particu- lar, even though visit rates were very low, we observed bee visits at all of our sites and estimate that flowers received an average of 4 visits over their extended blooming periods. We could not demonstrate directly that the bees deposited pollen on stigmas, but foragers consistently contacted stigmas when they visited flowers. Moreover, all captured individuals of A. nigri- hirta carried D. californica pollen. Perhaps most importantly, nearly 40% of emasculated flowers produced fruits with seed sets equivaient to controls at three of the five sites. This result indicates substantial cross-pollination, and strongly implicates bees as pollen vectors. How- ever, autogamy must have dominated at our sites because, with one exception (CL), fruit set of unmanipulated flowers was at least twice as high 2011) MEINDL AND MESLER: DARLINGTONIA CALIFORNICA POLLINATION ea Fic. 1. Step by step foraging behavior of A. nigrihirta on a D. californica flower. The bee initially lands on the petals below the windows (a—c) and then enters a window and walks across stigmatic surfaces (d, e). The bee then utilizes the convex portion of one of the five petals to walk onto the ovary and up to collect pollen (f-i). Following pollen collection, the bee uses a petal convexity as before to walk down the ovary, across the stigmas again, then out one of the windows (j—n), before leaving the flower (0, p). The flower is shown in d—m with the front petal removed and half of the two lateral petals removed. Panels f and k show the bee using the convex portion of the petal, which allows the bee to access the stamens. 28 MADRONO Fic. 2. Interior view of a D. californica flower with bottom portion of petals removed. Arrows highlight the distance between the petals and the ovary both immediately above a window (shorter arrow) and in between two adjacent windows (larger arrow). More space is provided for A. nigrihirta in between the windows than above them, which encourages the bee to enter a window and then walk across the stigmatic surfaces. The bee then utilizes the convex portion of the petal opposite the window it entered en route to the flower’s stamens. as fruit set of emasculated flowers (Fig. 3). Bees probably accounted for much of this self- pollination because they contacted stigmatic surfaces when they exited flowers after collecting large pollen loads. Spiders, which were abundant on flowers and known to be capable of effecting limited autogamy in D. californica (Nyoka 2000), likely further contributed to fruit and seed production via self-pollination. They often con- structed webs inside flowers linking anthers and stigmas, and in several instances these webs were completely dusted with pollen. In contrast, spider draglines connecting flowers were very rare at our sites, making it unlikely that spiders contributed significantly to cross-pollination. Pollen-eating thrips were also present in large numbers within flowers, but were rarely seen on stigmas and thus likely played a limited role as pollen vectors. Nyoka and Ferguson (1999) collected fungus gnats carrying D. californica pollen in southwest- ern Oregon, where they may have contributed to seed set. However, although fungus gnats were abundant at our sites, we rarely discovered them inside flowers. Our findings suggest a resolution of the long- standing mystery surrounding the pollination of D. californica. Like us, previous workers (Elder 1997; Nyoka 2000) reported high levels of fruit and seed production at their study sites in southwestern Oregon and the northern Sierra [Vol. 58 Nevada, but rarely or never observed flying pollinators — a discrepancy that led to the provocative hypothesis that omnipresent spiders are the most important pollinators. However, the high levels of pollen limitation observed in this study make it unlikely that spiders are the predominant pollen vectors for D. californica, given their abundance on flowers. Although spiders almost certainly contribute to pollen transfer in some degree, we propose instead that D. californica is melittophilous, as predicted by Schnell (1976), and specifically that A. nigrihirta is responsible for the majority of pollination across its range. Consistent with this view, we now know that A. nigrihirta pollinates D. californica in northwestern California as well as the northern Sierra Nevada (this study; Rice 2006). The same may be true for populations in southwestern Oregon, where Nyoka (2000) col- lected a pair of unidentified dark-bodied Andrena inside a flower. However, visit rates appear to be very low at all sites, which may partly explain why even observers who spent long periods in populations seldom observed visits. In addition, foragers tend to remain inside flowers for protracted periods (after quickly entering), and usually leave a population after visiting only one or two flowers (G. Meindl, unpublished). The difficulty of detecting these elusive bees is highlighted by the fact that although we spent well over 100 hours at our study sites setting up and monitoring experiments, we observed visits only during our focused census watches (10% of 230 watches). The alternative explanation for the limited number of previously reported visits is that A. nigrihirta was either absent or extremely rare at the sites studied by Austin, Elder, Nyoka, and Rice. Although spatial and temporal varia- tion in the local abundance of bee species is well documented (Williams et al. 2001), this explana- tion begs the question of how to account for the high levels of fruit and seed production docu- mented at these sites. A more parsimonious explanation may be infrequent but effective visits by A. nigrihirta coupled with the long period of anthesis of individual D. californica flowers. Clearly, additional timed surveys will be needed to document the relative abundance and impor- tance of A. nigrihirta as a pollinator across the range of D. californica. The relationship between A. nigrihirta and D. californica appears to be asymmetric, i.e., D. californica 1s specialized on A. nigrihirta, but A. nigrihirta is a generalist, at least on a broad scale. Across its range, which spans North America and | greatly exceeds that of D. californica, A. nigrihirta © is a generalist that has been observed to visit flowers from a diverse array of plants (Laberge and Ribble 1975), including members of Portu- | lacaceae (Motten et al. 1982), Fabaceae (Tepe- dino et al. 1995), and Ericaceae (Rice 2006), | 2011] O Emasculated 1 0.9 3 . rT) 7) — s h_ hoaben O Emasculated = = — ‘espe tees rT) c. 7) cs rt) rT) wn tego o th 0) 9) cS = rT) > <_< Fic. 3. O Unmanipulated 0.8 07 06 . Sut SM2 N17 CL DF O Unmanipulated MEINDL AND MESLER: DARLINGTONIA CALIFORNICA POLLINATION 29 0 Hand-Pollinated 0 Hand-Pollinated b : b Ps b b b : P| a 2g a a , : i i | 0 é g SM1 SM2 N17 CL DF Fruit and seed production by three treatment groups (emasculated, unmanipulated, and hand-pollinated flowers) at each study site. Top: fruit set (%) of three treatment groups at each field site. Emasculated flowers produced significantly fewer fruits than unmanipulated flowers. No significant difference was found between fruit production of unmanipulated vs. hand-pollinated flowers (y7 = 3.50, P = 0.06). Bottom: average number of seeds produced per capsule from three treatment groups at each field site. Different letters above bars indicate group means are significantly different (comparisons of group means are only made within sites). Differences in mean seed production between the three treatment groups were evident at all 5 sites (SM1: Fs63 = 65.35, P < 0.001; SM2: Fo 65 = 53.76, P < 0.001; CL: F259 = 52.49, P < 0.001; N17: Fo.sg = 46.31, P < 0.001; DF: F357 = 24.19, P < 0.001). along with D. californica. Three individuals of A. nigrihirta were collected during this study that carried both D. californica and Asteraceae pollen, indicating that A. nigrihirta is utilizing floral resources from multiple species of flowering plants. While asymmetrical species interactions are known to be common in ecological networks (Vazquez et al. 2007), it is unclear why D. californica relies so heavily on A. nigrihirta for pollination, considering the abundance of other bee species at our study sites. Despite visits by A. nigrihirta being rare, the morphometric fit between bee and flower appears strong. While bumblebees were among the most abundant pollinators active at our study sites, their large body size prevented them from utilizing D. californica as a floral resource (G. Meindl, unpublished). Likewise, honeybees also have difficulty entering and handling the flowers (Rice 2006). Andrena nigrihirta was able to enter the small windows of D. californica flowers quickly and efficiently, and proved to be of an 30 MADRONO ideal size to contact stigmas, climb onto the ovary beneath the petal convexities and gather pollen from the flower’s anthers. Paradoxically, several other bee species collected at our field sites were of similar size to A. nigrihirta (e.g., other Andrena spp., Osmia spp., and Lasioglos- sum spp.; G. Meindl, unpublished), yet only A. nigrihirta was observed to forage on D. califor- nica flowers. Further studies are needed to characterize the relationship between D. califor- nica and its bee pollinators, and to determine why visits are made predominantly by A. nigrihirta and not by other similarly sized bee species. However, the preference shown to the flowers of Darlingtonia at our field sites in northern California, along with the morpholog- ical match between bee and flower, suggest that A. nigrihirta and D. californica have an established relationship. The detailed accounts of floral visitation in this study, combined with the results of pollination treatments, provide sound evidence that D. californica produces melittophilous flowers that are effectively, though rarely, pollinated by the solitary bee A. nigrihirta. There are several interesting ecological ques- tions that have yet to be considered regarding D. californica pollination. For instance, why are visits by bees so infrequent? How do spiders occupying D. californica flowers interact with bees? Does the presence of spiders within flowers deter visitation by bees, or do bees frequently fall victim to lurking spiders, and what bearing does this have on D. californica reproductive success? Over the course of floral observations conducted in this study, A. nigrihirta was seen “buzzing” flowers, 1.e. approaching flowers but not entering them, more frequently than entering flowers (37 flowers visited, 50 flowers buzzed). While this “buzzing” behavior could be males searching flowers for females, other explanations are also possible. For example, this behavior could be the result of floral marking by bees, which may be done to alert future visitors of resource avail- ability (Schmitt and Bertsch 1990; Goulson et al. 2001), or may also be the result of altered foraging behavior caused by the presence of flower-occupying spiders (Bruce et al. 2005; Goncalves-Souza 2008). It is also unclear how floral form influences pollination by bees vs. spiders, i.e., do the same floral traits that promote pollen deposition on stigmatic surfaces by bees (shape of ovary, etc.) also promote pollen deposition by spiders, or should we expect divergence of floral morphology in D. californica populations that occur in areas where A. nigrihirta 1s absent over time? As we seek to explain the adaptive significance of D. californi- ca’s floral traits, we need to understand, in greater detail, the effects of these multi-species interactions on trait selection. [Vol. 58 ACKNOWLEDGMENTS The authors thank E. S. Jules, T. W. Henkel, J. O. Rice, and T. L. Ashman for their comments on previous versions of this manuscript. We also thank T. Buonaccorsi and S. Hariri-Moghadam for their work as research assistants. S. Hariri-Moghadam illustrated Figs. 1 and 2. LITERATURE CITED BRUCE, M. J.. A. M. HEILING, AND M. E. HERBER- STEIN. 2005. Spider signals: are web decorations visible to birds and bees? Biology Letters 1:299-302. DEBUHR, L. E. 1973. Distribution and reproductive biology of Darlingtonia californica. M.S. thesis. Claremont Graduate School, Claremont, CA. ELDER, C. L. 1997. Reproductive biology of Darlingto- nia californica. M.A. thesis. Humboldt State University, Arcata, CA. GONCALVES-SOUZA, T., P. M. OMENA, J. C. SOUZA, AND G. Q. ROMERO. 2008. Trait-mediated effects on flowers: artificial spiders deceive pollinators and decrease plant fitness. Ecology 89:2407—2413. GOULSON, D., J. W. CHAPMAN, AND W. O. H. HUGHES. 2001. Discrimination of unrewarding flowers by bees; direct detection of rewards and use of repellent scent marks. Journal of Insect Behavior 14:669-678. HARDER, L. D. AND S. D. JOHNSON. 2009. Darwin’s beautiful contrivances: evolutionary and functional evidence for floral adaptation. New Phytologist 183:530-545. HINTZE, J. 2004. NCSS and PASS. Number Cruncher Statistical Systems, Kaysville, UT. JUNIPER, B. E., R. J. ROBBINS, AND D. M. JOEL. 1989. The carnivorous plants. Academic Press, London, U.K. KEARNS, C. A. AND D. W. INOUYE. 1993. Techniques for pollination biologists. University Press of Colorado, Niwot, CO. LABERGE, W. E. AND D. W. RIBBLE. 1975. A revision of the genus Andrena of the western hemisphere. Part VII. Subgenus Euandrena. Transactions of the American Entomological Society 101:371-446. MEINDL, G. A. 2009. Pollination biology of Dar- lingtonia californica, the California pitcher plant. M.A. thesis. Humboldt State University, Arcata, CA. MOoTTEN, A. F., D. R. CAMPBELL, D. E. ALEXANDER, AND H. L. MILLER. 1981. Pollination effectiveness of specialist and generalist visitors to a North Carolina population of Claytonia virginica. Ecolo- gy 62:1278—1287. NE’EMAN, C., G. NE’EMAN, AND A. M. ELLISON. 2006. | Limits to reproductive success of Sarracenia | purpurea (Sarraceniaceae). American Journal of Botany 93:1660—1666. | NYOKA, S. E. 2000. The spider and the fly: a proposed | pollination scenario for Darlingtonia californica, the California pitcher plant. Bulletin of the Native | Plant Society of Oregon 33:68—69. | AND C. FERGUSON. 1999. Pollinators of Darlingtonia californica Torr., the California pitch- er plant. Natural Areas Journal 19:386—391. RENNER, S. S. 1989. Floral biological observations on Heliamphora tatei (Sarraceniaceae) and other 2011] plants from Cerro de la Neblina in Venezuela. Plant Systematics and Evolution 163:21—29. Rice, B. 2006. Field trip report: Darlingtonia pollina- tion studies in 2006. Website http://www.sarracenia. com/faq/faq7623.html [accessed 10 April 2011]. SCHMITT, U. AND A. BERTSCH. 1990. Do foraging bumblebees scent-mark food sources and does it matter? Oecologia 82:137—144. SCHNELL, D. E. 1976. Carnivorous plants of the U.S. and Canada. John F. Blair, Winston-Salem, NC. THOMAS, K. AND D. M. CAMERON. 1986. Pollination and fertilization in the pitcher plant Sarracenia purpurea. American Journal of Botany 73:678. TEPEDINO, V. J., T. L. GRISWOLD, AND W. R. BOWLIN. 1995. Pollinator sharing by three sympatric milk- veiches including the endangered species Astragalus montii. Great Basin Naturalist 55:19—25. MEINDL AND MESLER: DARLINGTONIA CALIFORNICA POLLINATION 31 VAZQUEZ, D. P., C. J. MELIAN, N. M. WILLIAMS, N. BLUTHGEN, B. R. KRASNOV, AND R. POULIN. 2007. Species abundance and asymmetric interaction strength in ecological networks. Oikos 116:1120—1127. WASER, N. M. 2006. Specialization and generalization in plant-pollinator interactions: a historical per- spective. Pp. 3-17 in N. M. Waser and J. Ollerton (eds.), Plant-pollinator interactions: from general- ization to specialization. The University of Chicago Press, Chicago, IL. WHITTAKER, R. H. 1954. The ecology of serpentine soils IV. The vegetational response to serpentine soils. Ecology 35:275—288. WILLIAMS, N. M., R. L. MINCKLEY, AND F. A. SILVEIRA. 2001. Variation in native bee faunas and its implications for detecting community changes. Conservation Ecology 5:7. MADRONO, Vol. 58, No. 1, pp. 32-49, 2011 ALASKANUS AND ELYMUS VIOLACEUS (POACEAE): IMPLICATIONS FOR i) A MORPHOMETRIC ANALYSIS OF VARIATION BETWEEN ELYMUS RECOGNITION OF TAXA KRISTEN HARRISON! AND RICHARD J. HEBDA'!?* 'Department of Biology, University of Victoria, P.O. Box 1700, Victoria, BC, Canada V8W 2Y2 kristenh@uvic.ca Natural History Section, Royal BC Museum, 675 Belleville Street, Victoria, BC, Canada V8W 9W2 *Schools of Environmental Studies and Earth Sciences, University of Victoria, P.O. Box 1700, Victoria, BC, Canada V8W 2Y2 ABSTRACT The aim of this study was to clarify the relationships between Elymus alaskanus and E. violaceus in northwest North America. We performed a morphological and biogeographic analyses of ca. 300 widely distributed herbarium specimens. Following a univariate analysis of morphological characters used in contemporary treatments, we found no clear character, or combination of characters, that differentiates unambiguously among the taxa at the specific level. However, glume and lemma trichome length reliably separated E. alaskanus subsp. hyperarcticus from other taxa. Specimens could not be differentiated at the specific level by habitat preferences or geographic distribution as described in the most current treatments. Further, principal components analysis and cluster analysis were unable to reliably segregate specimens into groups. Discriminant analysis reliably grouped EF. violaceus and E. alaskanus subsp. hyperarcticus, but not E. alaskanus specimens. In the development of a relevant treatment for E. alaskanus and E. violaceus, we recommend that (1) E. violaceus be treated as a subspecies of E. alaskanus and called E. alaskanus subsp. latiglumis, and (11) E. alaskanus subsp. alaskanus and E. alaskanus subsp. hyperarcticus continue to be recognized at the subspecific level. Key Words: British Columbia, E/ymus alaskanus, Elymus violaceus, taxonomy, Triticeae. Delineation of taxa within grass tribe 7viticeae (Poaceae) has been complicated and controversial (Dewey 1983a; Barkworth 1992; Zhang et al. 2000; Barkworth et al. 2007), with disagreement over taxonomic treatments at the generic and specific level (Hitchcock 1951; Tzvelev 1976; Love 1980a, b; Melderis 1980; Dewey 1983b, 1984; Barkworth 1992; Stewart and Barkworth 2001; Barkworth et al. 2007). The development of a stable nomenclature for the tribe has been inhibited by the morphological complexity of the group and lack of widely accepted criteria for the most appropriate taxonomic treatment (Bark- worth 1992). Elymus L., within the Triticeae, has the most species and widest distribution as interpreted by Dewey (1984), Love (1984) and Barkworth et al. (2007). It occurs worldwide in non-tropical regions and includes approximately 150 north- temperate perennial species (Dewey 1984; Zhang et al. 2000; Sun et al. 2006b; Barkworth et al. 2007). In the northwest North American province of British Columbia, Canada, there are twelve recognized species, of which Elymus alaskanus (Scribn. & Merr.) A. Love and E. violaceus (Hornem.) J. Feilberg are poorly resolved. Elymus species inhabit diverse ecological niches, including forests and forest edges, mountain slopes and valleys, semi-deserts and grasslands (Sun et al. 2006b). E/ymus morphology varies widely within and among species because of introgression, the ability of species to form intra- and interspecific fertile hybrids and the polyploid origin of the genus (Sun and Li 2005; Barkworth et al. 2007). Additionally, morphological vari- ability among species is partially under environ- mental control (Sun and Li 2005; Sun et al. 2006a; Barkworth et al. 2007). The high levels of variability observed in morphological traits are consistent with the genetic variability observed in molecular studies (Diaz et al. 1999; Zhang et al. 2000, 2002; Sun and Salomon 2003). Alaskan wheatgrass, Elymus alaskanus and Arctic wheatgrass, Elymus violaceus are perenni- al, allotetraploid species (StStHH, 2n = 4x = 28) that illustrate the taxonomic difficulty of E/ymus (Zhang et al. 2000; Sun and Salomon 2003; Barkworth et al. 2007). Previously, this species complex has been placed in several different taxa (cf. Hitchcock 1951; Welsh 1974; Love 1984; Baum et al. 1991; Cody 1996; Barkworth et al. 2007) (Table 1). Morphological similarity be- tween Elymus alaskanus and Elymus violaceus has lead to contradictory taxonomic conclusions, and taxonomists are not in agreement on whether or not the two are separate species (Zhang et al. 2011] 2000; Stewart and Barkworth 2001; Sun et al. 2006a; Barkworth et al. 2007). The issue of distinguishing the two taxa morphologically is illustrated in the two comprehensive treatments covering British Columbia: The Flora of North America (FNA) Volume 24 (Barkworth et. al. 2007) and The Illustrated Flora of British Columbia Volume 7 (Stewart and Barkworth 2001). Stewart and Barkworth (2001), recognize only one member at the specific level, E. dlaskanus (Scribn. & Merr.) Pex. Love subsp. latiglumis (Scribn. & J.G. Sm.) A. Love (=E£. violaceus), whereas Barkworth et al. (2007), recognize two species, Elymus adalaskanus and Elymus violaceus. The treatment in the FNA (Barkworth et al. 2007), in accordance with Hultén (1968), asserts that FE. alaskanus is differentiated from E. violaceus in having rela- tively shorter glumes than E. violaceus (Bark- worth et al. 2007). Those of E. alaskanus are said to be 3 to % as long as the adjacent lemmas, and those of E. violaceus v4 to equal to the lemma length (Barkworth et al. 2007). Following Love (1984) and Cody (1996), Barkworth et al. (2007) further divide E. alaskanus into subspecies, naming plants with relatively glabrous glumes and lemmas as E. alaskanus subsp. alaskanus, and those with glumes and lemmas covered densely by trichomes as E. alaskanus subsp. hyperarcticus (Polunin) A. Love & D. Love. Both taxa are mostly arctic or alpine (sometimes subalpine) species with a northern circumpolar distribution. However, the more restricted range of E. alaskanus is thought to distinguish it from E. violaceus (Barkworth et al. 2007). Elymus alaska- nus grows across the high arctic of North America to eastern Russia, through Siberia, Alaska, northern USA and Greenland (Zhang et al. 2000; Sun and Salomon 2003), but according to the FNA distribution maps is almost absent from British Columbia (Barkworth et al. 2007: 326). The distribution of E. violaceus extends from Alaska across arctic Canada to Greenland and south in the Rocky Mountains to southern New Mexico (Barkworth et al. 2007). In western North America E. dlaskanus is often associated with valleys and flat sites in low- competition habitats such as limestone outcrops, scree, moraines and dry meadows (Zhang et al. 2000; Barkworth et al. 2007), whereas E. violaceus favours calcareous or dolomitic rock in arctic, subalpine and alpine habitats. In general, E. alaskanus is thought to be found at lower elevations than E. violaceus (Barkworth et al. 2007). The aim of this study is to clarify the relationships between E. alaskanus and E. viola- ceus by performing morphological and biogeo- graphic analyses of herbarium specimens col- lected from a broad geographic range in northwest North America, and to answer two HARRISON AND HEBDA: MORPHOMETRIC ANALYSIS OF VARIATION IN ELYMUS 33 questions. 1) Can E. alaskanus and E. violaceus be regarded as separate species in British Columbia and adjacent regions? And if so, 2) what morphological, geographical and habitat charac- ters can be used to discriminate between the species? Our overall objective is to contribute to the development of a single taxonomic treatment for E. alaskanus and E. violaceus in northwest North America and advance our understanding of these taxa over their broader ranges. Increased knowledge of the relationship among entities will be especially useful in British Columbia because of the widespread geographic overlap of the two species and current disagreement over their treatment within the province (e.g., Stewart and Barkworth 2001; Barkworth et al. 2007). METHODS Nomenclatural Considerations Two sets of infraspecific taxa can be considered in Table 1, those in the ‘“‘boreale/alaskanus” complex and those in the “‘/atiglumis/violaceus/ hyperarcticus’ complex. When considering the infraspecific taxa from the boreale/alaskanus column (Table 1), we regard FE. alaskanus and E. alaskanus subsp. borealis (Turcz.) A. Love & D. Love as constituting the same taxon because in general taxonomists agree that differences between the potential subspecies do not warrant recognition (Stewart and Barkworth 2001; Bark- worth et al. 2007). Hultén (1968) and Welsh (1974) recognized three subspecies within Agro- pyron boreale Drobow, as did Love (1984) and Cody (1996), but they placed the subspecies in Elymus. Taxonomists placing the members of this nomenclatural set in Elymus had to change the specific epithet used from ‘“‘horeale”’ to “‘alaska- nus’ in order to conform with the rules of the International Code of Botanical Nomenclature (McNeill et al. 2006). We followed Barkworth et al. (2007) who differed from pre-existing treat- ments in combining these two infraspecific taxa into a single taxon, which, according to the rules of priority, were called E/ymus alaskanus subsp. alaskanus. The fundamental question concerning the treatment of “‘J/atiglimis” and “violaceus” concerns the appropriate names to be applied. Scribner and Smith (1897) originally named these plants Agropyron violaceum (Hornem.) Lange var. /atiglume Scribn. & J. G. Sm. Their description provided a brief description of the new variety, but did not state how the entity differed from var. violaceum. Generally, taxono- mists agree that “‘/atiglumis” and “violaceus” refer to the same taxon (Stewart and Barkworth 2001; Soreng et al. 2003; Barkworth et al. 2007), with the exception of Léve (1984) who applied separate names, but this compendium of taxo- nomic groups within the 7riticeae was based on ~ MADRONO [Vol. 58 SNIIVDIOIA ‘dsqns syjnpodyopay sna SNAIDIOIA ‘dsqns snynpodAyopal snd uinadDjola UosAdOAB py wunadnjola uosAdoss py WINAIDIOIA ‘dsqns winaspjoia uosddois py WINAIDIOIA ‘dsqns winaosnjoia uosddoss py wunadnjola UosAAdOss py SUN] B1]D] DIAIUBION SUN] 811] ‘dsqns snupyspjp sniuAyy AUN) BID] “TRA wnndIAYyIDA] UOAAdOAB auun] 8110] “IBA WiNUIUDI UOsAAdOAS ‘QUIN]s1ID] “ed sn{put ‘dsqns winuiupn) uosddoisp ‘awn siy] “ed sn{piu ‘dsqns wnuups uosddois py SMUN]BIJD] DIAAUBIOY aun s1jD] UdAAdOAS py aun s1jp] UosAdOAS py JUIN] BID] "IeA UNIIDIOI UOLAdOAB snoyosDsaday ‘dsqns snupyspjp sniuAyq DIN IADAIGAY “IRA S1]VALOG “Y 10 SIDIAOG DIAIUBIOY snoyaspsadady ‘dsqns snupyspjp snuajyq WNIIJIADAIAUAY “IBA LUNAIDIONA UOAAdOIS snIyIADAAdUAY ‘dsqns sisuaupips snuidjq UNIT IADAIUAY "IBA ajpasog uosddoss py WNILIADAIAAY ‘dsqns ajpai0g uosddoisp pIIAWAIdAY ‘dsqns sijpai0g p1lddusIOy WNI1IADAIGAY ‘IBA UMIIDIOIA UOAAdOAS SNUDYSDID ‘dsqns snupyspjp snudpy SNUDYSDID ‘dsqns snupyspjp snuApy UINUDYSDID “IBA a/Das10g UosAdO.18 py WNUDYSDID ‘dsqns ajva.i0g uosddois py uNnUDYSDID UuoAAdOIS sypasoq ‘dsqns snupyspjp snuuajq SIJDIAOG VIAAUBION S1]DAAOg ‘dsqns snupyspjp snudayy Sjpas10q ‘dsqns snupyspjp sniuAyy a/DALOG ‘IBA ajpasog uosddois py a[DALOG ‘dsqns ajpai0g uostdoisp ajpasog uoAddoss py SIJDIAOG DIAQUBIOM ajpasog uoAddoss fy SIVDAAOG SNUALA I[DIAOG WUNOIT]IA (9661) APOD (1661) ‘Te 19 wineg (P861) 2AQT (€861) SSO (0861) IONS pure o10q (0861) ApoD pur pjisiog (8/61) Uess09g (9L61) AS[PAZL (9L61) SAQT 2 9AQT (p61) USI9M (€L61) IsInbuosD pure yoooysuH (6961) YOoyNH (8961) UsIINH (6S61) wruntog (9961) 9AQT 2 9AQT (TS61) 9f1994 (TS61) YOooystH (Or61) ULUNTOg (PE61) PISAON (9161) MOQOIG (O161) [TLS pue TSUqLID§ (6061) S19qpAy (0061) 19UqLIDS (L681) yUuug pure JIoUuUqlISS (9S8][) MOUTURZOIN TL, _ SNIIVIONA, SUN] 8117, _ SNINIADAIAAY, , Aynuy SNUDYSVYIV,, DIAG, , ‘ JOUdIIIJIY (L661 HLYOMAUVG WOU GAISIGOP) VOIMPANY HLYON NI SAXATKINOD SAFIVTIOIA “FY ANV SANFYSVTF SQAWATY AHL AO AYNLVIONAWON TVOIAOLSIH ‘T ATavV 3b 2011] HARRISON AND HEBDA: MORPHOMETRIC ANALYSIS OF VARIATION IN ELYMUS Continued. TABLE |. Entity “violaceus” “boreale”’ “alaskanus”’ “hyperarcticus”’ “latiglumis”’ Reference Elymus alaskanus subsp. Stewart and Barkworth latighimis (2001) Barkworth et al. (2007) Elymus violaceus Elymus alaskanus subsp. Elymus alaskanus subsp. hyperarcticus Elymus alaskanus subsp. alaskanus Elymus alaskanus subsp. Elymus alaskanus subsp. Harrison and Hebda latiglumis hyperarcticus alaskanus (this study) eS) SN names, not the plants themselves. The name Agropyron violcaeum var. latighime, as it appears on the holotype for this entity, was called E/ymus violaceus by Barkworth et al. (2007) in the Flora of North America not to reflect a new entity but to include E. dlaskanus subsp. latighimis [= Ag- ropyron latiglume Rydb.]. Here we regard E. violaceus and E. dlaskanus subsp. latiglumis as synonyms following the work of contemporary taxonomists (Stewart and Barkworth et al. 2001; Soreng et al. 2003; Barkworth et al. 2007). Sampling and Measurements Herbarium specimens from the Royal BC Museum (V), the University of British Columbia (UBC), the Canadian Museum of Nature (CAN) and the United States National Herbarium (US) were used as the basis for this study (Appendix 1). All specimens included in the analysis evidently belonged in the taxa of interest, thus none were disqualified. Potential hybrid speci- mens (i.e., intermediate morphologies) were not excluded from the analysis because doing so could potentially create artificial groupings. Specimens retaining historical nomenclature had current names applied to them following the Flora of North America (FNA) (Barkworth et al. 2007) and were divided into three categories (1) E. alaskanus sensu stricto (includes specimens named FE. alaskanus and E. alaskanus subsp. alaskanus) (2) E. alaskanus subsp. hyperarcticus and (3) E. violaceus. A preliminary analysis of specimens revealed that identifiers correctly applied the name E. a. subsp. /hyperarcticus to specimens with hairier glumes and lemmas as described in the FNA (Barkworth et al. 2007). Hence, we are confident that our analysis of the broader taxonomic group E. alaskanus did not include specimens of E. a. subsp. /Ayperarcticus. From herein we will refer to specimens of E. alaskanus and E. dlaskanus subsp. alaskanus collectively as E. dlaskanus sensu stricto (s.s.) and specimens including all three taxa as E. dlaskanus sensu lato (s./.). In total, 109 E. alaskanus s.s., 18 E. alaskanus subsp. hyperarcti- cus and 169 E. violaceus specimens were included in the analysis. Plants originated from the northwest continental United States, Alaska and Canada (Table 2). Type specimens from CAN and US were examined separately and included (1) Agropyron alaskanum Scribn. and Mierr... (Contrib. -‘U_S., Natl. Herb: 13: 83; -1910. Type: United States: Alaska. Circle City. 18 Aug. 1899. W.H. Osgood s.n. [holotype: US]); (2) Agropyron violaceum var. latighime Scribn. and J.G. Sm. (U.S. Dept. Agric. Div. Agrost. Bull. 4: 30. 1897. Type: United States: Montana. Gallatin Co., Lone Mountain, Tweedy 10/1 [holotype: US]); (3) Agropyron violaceum var. hyperarcticum Polunin (Bull. Natl. Mus. Canada 92 (Biol. Ser. 36 MADRONO TABLE 2. SUBSP. HYPERARCTICUS (N = [Vol. 58 GEOGRAPHIC ORIGIN AND NUMBER OF ELYMUS ALASKANUS SENSU STRICTO (N = 110), E. ALASKANUS 18) AND E. VIOLACEUS (N = 169) SPECIMENS EXAMINED FOR MORPHOLOGICAL ANALYSIS IN THIS STUDY. AK = Alaska, AB = Alberta, BC = British Columbia, MT = Montana, NU = Nunavut, NWT = Northwest Territories, ON = Ontario, QC = Quebec, UT = Utah, WA = Washington, YT = Yukon Territory. AK AB BC MT E. alaskanus 9 2 38 a E. a. subsp. hyperarcticus 7 — = E. violaceus 3 4 134 l 24): 95. 1940. Type: Canada: Nunavut, Baffin Is., Arctic Bay, 9 Sept. 1936. N. Polunin 2531 [isotype: CAN]). We used 22 morphological characters for analyses (Table 3). All measurements of glume and lemma characteristics were made under 10 magnification to the nearest 0.1 mm using an ocular micrometer. Blade length and width, spikelet, culm, and inflorescence length were measured with a line ruler to the nearest 1mm. Spikelets were selected from the middle of the inflorescence and the glume and lemma were chosen from the same spikelet. All lemmas, regardless of their stage of development, were counted. Ratios between lower glume and spike- let length, the lower glume and lemma length, and between glume margin width at widest point to total glume length were calculated. Measure- ments of both glumes and lemmas did not include the awns which were considered separately. Habitat, elevation and geographical informa- tion were recorded from herbarium sheets. All specimens from Alaska, Alberta, British Colum- bia, Northwest Territories, Nunavut and Yukon with sufficient geographic information on her- barium labels were mapped using ArcView 9.3 (2008). Morphological Analysis Univariate analysis. We used univariate analy- ses to examine the effectiveness of using glume to lemma ratio as the key diagnostic character separating E. alaskanus s.l. and E. violaceus (as currently done in the Flora of North America volume 24, Barkworth et al. 2007). We also considered the effectiveness of using lemma and glume trichome length to identify E. alaskanus subsp. Ayperarcticus. Data did not meet assump- tions for normality (Shapiro-Wilk test statistic) and homogeneity of variance (plot of residuals versus fits), thus a Kruskall-Wallis test of the equality of medians was performed as a non- parametric alternative to analysis of variance (ANOVA). Boxplots were used for visual com- parison of these traits. Additionally, we took as a subset of specimens, those identified by M. Barkworth (Intermountain Herbarium, Utah State University), to analyze differences in glume NU NWT ON QC UT WA YT = aT = I = SS a3 I 4 ee 6 =2 8 I _ I 2 15 to lemma ratio among taxa while reducing the variation in the interpretation of the diagnostic criteria. This subset of data met assumptions of normality and equal variance; thus ANOVA was performed and boxplots were created to investi- gate differences among groups. All univariate analyses were computed with Minitab (2007). Null hypotheses were rejected at P < 0.05. Lower glume to lower lemma measurements and ratios of type specimens from CAN and US were examined separately. Multivariate analysis. Multivariate analyses tests included principal components analysis, discriminant analysis and cluster analysis. Corre- lation matrices were constructed to investigate linear relationships between morphological vari- ables using Pearson’s product moment correla- tion. Lower glume length, lower lemma length and spikelet length were excluded from multivar- late analyses because they were components of computed ratios and elevation was excluded because a preliminary analysis indicated it varied with latitude. Because tests require that all observations are present for all cases, we excluded anther length which had a high proportion of missing values. In total, 286 specimens were used. Morphological characters included in these anal- yses are reported in Table 3. We used principal components analysis (PCA) to identify morphological characters that con- tributed most to the variation among specimens and to characterize the pattern of trait relation- ships between E. alaskanus s.s., E. alaskanus subsp. Ayperarcticus and E. violaceus. Eighteen variables were included in the anlaysis. PCA was performed using a correlation matrix and six principal components were computed. Factor scores were used in subsequent ANOVAs to test the significance of factors among the taxa. To assess how well trait measures could be used to correctly classify plants into taxonomic groups, we used discriminant analysis. For this analysis a quadratic discriminant function with fits was applied. To determine if our observations could be segregated into groups that were not defined in advance we used cluster analysis. A dendrogram was produced using single linkage and Euclidean distance, with variables standard- 2011] HARRISON AND HEBDA: MORPHOMETRIC ANALYSIS OF VARIATION IN ELYMUS oi) TABLE 3. CHARACTERS MEASURED OR RECORDED FOR ANALYSIS. *Characters used in Principal Components Analysis (PCA), discriminant analysis and cluster analysis. ‘Margin to glume length ratio excluded from discriminant analysis because it was highly correlated with other predictors in E. alaskanus subsp. hyperarcticus. Character Culm Culm length* Blade Blade length* Blade width* Inflorescence Inflorescence length* Inflorescence width* Spikelet Spikelet length Spikelet width* Glume Lower glume length Lower glume width * Glume margin width * Glume trichome length* Glume veins* Glume awn length* Lemma Lower lemma length Lower lemma width* Lemma awn length* Lemma trichome length* Anther length Floret number* Ratios Margin/glume length*+ Description Length (cm) from below the inflorescence to culm base Length (cm) of longest blade Width (cm) of widest point of longest blade Length (cm) of longest inflorescence; without awns Width (mm) of widest point of longest inflorescence Length (mm); awnless; spikelet from mid-inflorescence Width (mm) at widest point; spikelet from mid-inflorescence Length (mm) of lower glume; awnless Width (mm) at widest point of lower glume Width (mm) of glume margin Length (mm) of glume trichomes Number of glume veins Length (mm) of glume awn Length (mm) of lower lemma ; awnless Width (mm) of lower lemma at widest point Length (mm) of awn length of lower lemma Length (mm) of lemma trichomes Length (mm) of anthers Total number of florets within spikelet; all stages of development Width of glume margin at widest point to total glume length Glume/spikelet* Lower glume length to spikelet length Glume/lemma* Lower glume length to lower lemma length Other Habitat From herbarium sheet Location From herbarium sheet Elevation From herbarium sheet ized. All multivariate analyses were computed RESULTS with Minitab (2007). Morphological Analysis Biogeographic analysis. To determine if differ- ences in elevation exist among E. alaskanus s.s., E. alaskanus subsp. hyperarcticus and E. viola- ceus, Specimens were placed in latitude categories: (1) all latitudes (2) =60°N (true arctic) (3) 55°— 60°N (transition-boreal) (4) <55°N (southern alpine). Data in the first three groups did not meet assumptions of normality or homogeneity of variance, thus a Kruskall-Wallis test was performed to test for differences in elevation among taxa. Data in group 4 met parametric assumptions and ANOVA was performed. For the habitat analysis, all specimens with adequate information on herbarium labels (Appendix 1) were classified into two categories (1) rocky habitats or (2) valleys/flat areas and a chi-square test was performed to look at associations between habitat type and taxa. Univariate analysis. All morphological charac- ters generally had overlapping ranges (Table 4). Taxa differed in glume to lemma ratio (Kruskall- Wallis, df = 2, P < 0.001 adjusted for ties; Fig. 1). A subset of specimens, those identified by Barkworth, also differed in glume to lemma ratio among taxa (ANOVA, Foo.113) = 43.15, P< 0.001; R* = 0.423; Fig. 2). Following ANOVA, pairwise comparisons among taxa (Tukey 95% simultaneous confidence intervals) showed no significant differences between FE. alaskanus s.s. and E. alaskanus subsp. hyperarcticus, but did find that E. alaskanus subsp. hyperarcticus 1s significantly different from EF. violaceus, and E. alaskanus s.s. is different from EF. violaceus. Highly significant differences among taxa were detected for both lemma trichome length (Krus- 38 MADRONO [Vol. 58 TABLE 4. MEAN, STANDARD DEVIATION AND RANGE (IN PARENTHESIS) FOR 22 TAXONOMIC TRAITS OF ELYMUS ALASKANUS SENSU STRICTO, E. ALASKANUS SUBSP. HYPERARCTICUS AND E. VIOLACEUS. Variable E. alaskanus E. a. subsp. hyperarcticus E. violaceus Culm length (cm) 33.1 + 1.3 (10.0-69.0) 24.26 + 1.98 (12.0-45.1) 31.20 + 1.17 (10.0~77.5) Blade width (cm) 0.3 + 0.02 (0.1-0.9) 0.3 + 0.03 (0.1—0.5) 0.3 + 0.01 (0.1—1.9) Blade length (cm) 8.8 + 0.4 (2.1—20.0) 7.4 + 0.6 (4.2—14.0) 7.8 + 0.4 (1.9-41.0) Inflorescence length (cm) 7.5 + 0.2 (3.2-15.0) 6.4 + 0.5 (3.5—10.0) 6.9 + 0.2 (3.1—-15.0) Inflorescence width (cm) 0.5 + 0.02 (0.3—2.0) 0.6 + 0.4 (0.4-1.1) 0.6 + 0.01 (0.31.8) Spikelet length (mm) 12.6 + 0.2 (7.7—20.7) 11.4 + 0.4 (8.8—-14) 12.3 + 0.1 (7.8—20.7) Spikelet width (mm) 2.1 + 0.05 (0.43.9) 2.2 + 0.09 (1.8-3.1) 2.2 + 0.03 (1.1—3.5) Lower glume length (mm) 6.9 + 0.2 (2.4-13.5) 5.9 + 0.2 (5.0-—7.4) 8.2 + 0.1 (4.9-13.0) Lower glume width (mm) 1.5 + 0.03 (0.6—2.8) 1.4 + 0.08 (0.7—-1.9) 1.7 + 0.02 (1.0-2.5) Lower glume awn length (mm) 0.8 + 0.08 (0.0—-6.5) 0.7 + 1.3 (0.2—2.6) 0.8 + 0.04 (0.0-4.0) Number of glume veins 2-5 2-3 2-5 Width of lower glume margin at widest point (mm) 0.4 + 0.1 (0.0—0.8) 0.4 + 0.02 (0.2-0.6) 0.5 + 0.01 (0.1-1.0) Lower lemma length (mm) 9.1 + 0.1 (5.8—14.0) Sef eS Cea 8.8 + 0.08 (6.4—-12.0) Lower lemma width (mm) 1.7 + 0.03 (0.6—2.5) 1.8 + 0.06 (1.2—2.2) 1.7 + 0.02 (1.0-2.5) Lower lemma awn length (mm) 2.1 + 0.2 (0.0—7.5) 3.3 + 0.4 (1.0- a 2) 1.0 + 0.08 (0.09.9) Number of florets 2-6 14 1-6 + + Lower glume trichome length (mm) 0.04 + 0.008 (0.0—0.3) 0.2 Lower lemma trichome length 0.3 (0.0—0.6) 0.007 + 0.002 (0.0—0.3) (mm) 0.2 + 0.01 (0.0—0.6) 0.4 + 0.02 (0.2—0.6) 0.2 + 0.01 (0.0—1.0) Anther length (mm) 12 = 0.02 (0.7-2.)) 1.2 + 0.05 (1.0-1.7) 1.1 + 0.02 (0.5—-1.8) Lower glume length/ spikelet length 0.6 + 0.01 (0.2—0.9) 0.5 + 0.02 (0.4-0.7) 0.7 + 0.008 (0.3—1.03) Lower glume length/lower lemma length 0.8 + 0.01 (0.4—1.2) 0.7 + 0.02 (0.5—0.8) 0.9 + 0.009 (0.7—1.5) Glume margin width at widest point/ lower glume length 0.7 + 0.1 (0.41.4) 0.7 + 0.04 (0.3—1.0) 0.7 + 0.01 (0.28—1.0) kall-Wallis, df = 2, P < 0.001 adjusted for ties; var. /atiglume) had a glume to lemma ratio of Fig. 3) and glume trichome length (Kruskall- 0.91, Elymus alaskanus subsp. alaskanus (= Agro- Wallis, df = 2, P < 0.001 adjusted for ties; pyron alaskanum) had a ratio of 0.59, and Elymus Fig. 4). Type specimen measurements indicate alaskanus subsp. hyperarcticus (=Agropyron vio- that Elymus violaceus (=Agropyron violaceum laceum var. hyperarcticum) a ratio of 0.76. 1.50 * S 1.25 % 6 ® £ = 1.00 O — O° eae = 0.75 =) © 0.50 * E. asaskanus E. a. subsp. hyperarcticus _ E. violaceus Fic. 1. Glume to lemma ratio for Elymus alaskanus sensu stricto (n = 110), E. alaskanus subsp. hyperarcticus (n = 18) and E. violaceus (n = 169). Glume to lemma ratio for type specimens Elymus violaceus (= Agropyron violaceum var. latiglume), Elymus alaskanus subsp. alaskanus (=Agropyron alaskanum), and Elymus alaskanus subsp. hyperarcticus (=Agropyron violaceum var. hyperarcticum) indicated by @ symbol. 2011] HARRISON AND HEBDA: MORPHOMETRIC ANALYSIS OF VARIATION IN ELYMUS 39 1.50 ¥% s 1.25 1. © fad © = E 1.00 — 8 = 07 = © 0.50 E. alaskanus E. a. subsp. hyperarcticus E. violaceus Fic. 2. Glume to lemma ratio for specimens of Elymus alaskanus sensu stricto (n = 32), E. alaskanus subsp. hyperarcticus (n = 3) and E. violaceus (n = 81) identified by Barkworth. Multivariate analysis. Correlations among morphological characters used in the multivariate analysis ranged from 0.021 to 0.8, thus none were excluded from the analysis. Five principle com- ponents (PC) had eigenvalues >1 and the first three components accounted for 47% of the variation in the data set (Table 5; Fig. 5). The first principle component (PC1) accounted for 20% of the total variance, with the lower glume width and glume length to lemma length ratio and lower lemma width having the highest coefficients, and all loading positively on PC1. In contrast, blade length, lemma awn length and 0.6 0.5 0.4 0.3 0.2 0.1 Lemma trichome length (mm) 0.0 E. alaskanus Fic. 3. and E. violaceus (n = 169). E. a. subsp. hyperarcticus culm length loaded negatively on PCl. PC2 accounted for 15.7% of the total variance and reflected increased inflorescence length, blade length and culm length, but decreased trichome lengths of both glumes and lemmas. Spikelet width, lower lemma width and glume trichome length loaded negatively on PC3 and glume to spikelet length ratio and glume trichome length loading positively. An ANOVA using PCI scores confirmed differences among taxa (ANOVA, F(2283) = 28.65, P < 0.001; R* = 0.168), with E. violaceus having significantly larger PC1 scores than either E. vioalceus Lemma trichome for E/ymus alaskanus sensu stricto (n = 110), E. alaskanus subsp. hyperarcticus (n = 18) AO MADRONO Glume trichome length E. alaskanus E. a. subsp. hyperarcticus [Vol. 58 E. violaceus Fic. 4. Glume trichome length for Elymus alaskanus sensu stricto (n = 110), E. alaskanus subsp. hyperarcticus (n = 18) and E. violaceus (n = 169). E. alaskanus or E. alaskanus subsp. hyperarcticus (Table 6). Pairwise comparisons among taxa of PCA factor 1 (Tukey 95% simultaneous confi- dence intervals) showed no significant differences between E. alaskanus s.s. and E. alaskanus subsp. hyperarcticus. However, E. alaskanus subsp. hyperarcticus was different from E. violaceus, and E. alaskanus s.s. was different from E&. violaceus. ANOVA of PC2 scores showed highly significant differences among taxa (ANOVA, F(2.283) = 28.65, P= 0.001; R? = 0.136), with E. alaskanus subsp. hyperarcticus different from both E. violaceus and E. alaskanus s.s. ANOVA of PC3 scores also confirmed highly significant differences among taxa (ANOVA, F2283) = 26.45, P < 0.001; R* = 0.151). Pairwise compari- sons among taxa of PC3 indicate significant differences among all taxa. Discriminant analysis of morphological char- acters (Table 3) indicated that EF. alaskanus s.s., E. alaskanus subsp. hyperarcticus and E. violaceus were assigned to their true group 72.1%, 100% and 93.9% of the time, respectively. When using a subset of the total morphological characters, those characters used in the FNA (Barkworth et al. 2007) including glume to lemma ratio, glume trichome length and lemma trichome length, E. alaskanus s.s., E. alaskanus subsp. hyperarcticus and E. violaceus were assigned to their true group 39.4%, 94.4% and 86.6% of the time, respectively. Cluster analysis results indicate that our obser- vations could not be segregated into three discrete groups. All specimens fell within a single cluster. Biogeographic analysis. Elevation differed among taxa when specimens were combined from all latitudes (Kruskall-Wallis, df = 2, P < 0.001 adjusted for ties; Fig. 6). However, significant differences for elevation between E. alaskanus s.1. and E. violaceus were not detected when speci- mens were grouped by latitude (1) below 55°N (Kruskall-Wallis, df = 1, P < 0.090 (adjusted for ties) (2) 55° N—60°N (Kruskall-Wallis, df = 1, P< 0.0191 (adjusted for ties) (3) above 60°N (AN- OVA, Foa4a1 = 0.09, P < 0.916; R* < 0.01 adjusted). Note that there are no herbarium specimens of FE. alaskanus subsp. hyperarcticus south of 60°N. Further, no evidence exists for association between taxa and habitat type (Fig. 7; . Chi-square test P < 0.528). With the inclusion of recently collected specimens the distribution of the two species overlaps broadly, particularly in British Columbia (Fig. 8). This pattern differs markedly from data of Barkworth et al. (2007) where E. alaskanus s.l. was restricted to extreme northern BC and northward. DISCUSSION The close morphological association among taxa makes it difficult to differentiate among entities. We found, as Barkworth et al. (2007) did, that the glume to lemma ratio of E. alaskanus s.s. is significantly less than that of E. violaceus. Our average ratios indicate that the glumes of E. alaskanus s.s. and E. alaskanus subsp. hyperarcti- cus are on average 3 to % as long as the adjacent lemmas, and those of E. violaceus are *%s to equal the lower lemma length (Fig. 1). Though the mean values for glume to lemma ratio concur with Barkworth et al. (2007), boxplots (Fig. 1) demonstrate that the range of overlap is too large for discrimination between the proposed species based on this character alone. Moreover, a subset 2011] TABLE 5. HARRISON AND HEBDA: MORPHOMETRIC ANALYSIS OF VARIATION IN ELYMUS 4] COEFFICIENTS AND EIGENVALUES FOR THE FIRST THREE COMPONENTS OF ELYMUS ALASKANUS SENSU STRICTO, E. ALASKANUS SUBSP. HYPERARCTICUS AND E. VIOLACEUS INDIVIDUALS. * Percent of the total variability accounted for by each principle component. Variable Culm length (cm) Blade width (cm) Blade length (cm) Inflorescence length (cm) Inflorescence width (cm) Spikelet width (cm) Lower glume width (cm) Lower glume awn length (mm) Number of glume veins Width of widest point of glume margin (mm) Lower lemma width (mm) Lower lemma awn length (mm) Number of florets Lower glume trichome length (mm) Lower lemma trichome length (mm) Lower glume to spikelet length ratio Lower glume length to lower lemma length ratio Width of widest point of glume margin to lower lemma length ratio of specimens identified by Barkworth (Fig. 2) suggests that even when the distinguishing criteria are strictly applied, there is a continuum of values rather than discrete ranges for glume to lemma ratio that might indicate distinct entities. E/ymus alaskanus subsp. hyperarcticus clearly has longer glume and lemma trichomes than the other taxa. Elymus alaskanus s.s. and E. violaceus trichome lengths are very similar (Figs. 3 and 4). These observations demonstrate that FE. alaskanus subsp. Ayperarcticus is easily distinguishable from other taxa as has been noted by others (Polunin 1940; Love and Love 1956; Hultén 1968; Welsh 1974; Tzvelev 1976; Love 1984; Baum et al. 1991; Cody 1996; Barkworth 1997; Barkworth et al. 2007). Type specimens of the taxa were distin- guishable based on lower glume to lower lemma ratio and followed the criteria outlined in the FNA (Barkworth et al. 2007). We expected the type specimens to fit the criteria outlined in the FNA (Barkworth et al. 2007) because they were named differently based on morphological dif- ferences of the particular specimens collected. However, it must be recognized that the useful- ness of a type specimens for clarifying taxonomic issues may be limited because it represents only one population. Type specimens of E. alaskanus subsp. dlaskanus (=Agropyron alaskanum), E. alaskanus subsp. hyperarcticus (=Agropyron vio- laceum var. hyperarcticum) and Elymus violaceus (= Agropyron violaceum var. latiglume) originated from Alaska, Nunavut and Montana, respective- ly and thus may be discrete compared to geographically intermediate material from British Columbia. Using multivariate techniques we were unable to find a combination of characters that permit PCI (20%)* PO? (15.1%) PCS CUa1%)* =0 212 0.379 0.032 —0.094 0.308 =—O2719 —0.236 0.344 =o 158 —O17S 0.424 —0.106 0.237 0.101 —Ui273 0.263 0.072 0.378 0.421 0.089 0.014 0.099 0.085 =0,129 0.084 0.278 —0.014 0.296 0.041 —0.047 0.339 0.013 = 325 —0.224 O.l21 —0.247 0.085 0.248 =).354 —0.115 —0.331 —0,335 0.207 —0.289 =—212 0.318 0.139 0.336 0.346 0.210 0.284 0.046 ==(), 147 —0.146 an unambiguous determination of groups at the specific level. Scatterplots of PCA factors 1-3 (Fig. 5) reveal a great deal of overlap among taxa, and the most defined group appears to be E. alaskanus subsp. hyperarcticus. Correlations between PCA scores and original traits are relatively low in magnitude, thus indicating that the morphological characters represent a small proportion of the overall variability. Discrimi- nant analysis indicated that E. alaskanus subsp. hyperarcticus and E. violaceus could be assigned to their predefined taxonomic groups most of the time, but that E. alaskanus s.s. was a less reliable grouping. Further, we did a second discriminant analysis using a subset of data (glume to lemma ratio, glume trichome length and lemma trichome length) and found that E. alaskanus s.s. was correctly classified only 39.4% of the time. This may indicate that people making identifications have an easier time classifying E. violaceus and E. alaskanus subsp. hyperarcticus specimens than they do E. alaskanus s.s. specimens, however why this might be remains unknown. We used cluster analysis to determine if specimens could be put into groups that were not defined in advance but the results indicate that the observations were not divisible into groups. According to Barkworth et al. (2007) E. alaskanus s.l. is thought to inhabit lower eleva- tions than E. violaceus. Our analysis indicates a trend for E. violaceus to be at higher elevations below 60°N, but these differences were not significant (Fig. 6). Above 60°N no differences were detected among taxa. Environmental con- ditions to which plants are exposed at similar elevations are not constant across latitudes (Pojar and MacKinnon 1994), and this may explain our 42 MADRONO [Vol. 58 (a) Increasing inflorescence length, culm length, blade length @ &. alaskanus xX E. a. subsp. hyperarcticus © E. violaceus FACTOR 2 Decreasing glume and lemma trichomes -5.0 -2.5 0.0 25 5.0 FACTOR 1 pecreasing blade Tengen, Increasing glume and lemma width serine ay nee and glume to lemma length ratio culm length S et (b) @ &. alaskanus xX E. a. subsp. hyperarcticus © E. violaceus Increasing glume to spikelet length ratio and glume to lemma length ratio FACTOR 3 Decreasing spikelet width, lower lemma fee glume trichome -5.0 -2.5 0.0 25 5.0 en FACTOR 1 Decreasing blade length, Increasing glume and lemma width lemma awn length, and glume to lemma length ratio culm length Fic. 5. Scatter graphs of principal components scores in pairwise relationships: a) factor 1 vs. factor 2; b) factor | vs. factor 3; c) factor 2 vs. factor 3. See Table 5 for the morphological characters included in the analysis. 2011] (Cc) Increasing glume to spikelet length ratio and glume to lemma length ratio FACTOR 3 Decreasing spikelet width, lower lemma width, glume trichome HARRISON AND HEBDA: MORPHOMETRIC ANALYSIS OF VARIATION IN ELYMUS 43 @ E. alaskanus xX E. a. subsp. hyperarcticus © E. violaceus l h ore 50 -25 O00 25 50 75 FACTOR 2 Decreasing glume and Increasing inflorescence lemma trichomes length, culm length, blade length Fic. 5. Continued. results. As a general rule, species occur at lower s./. and E. violaceus demonstrates that the elevations as one moves north. At lower latitudes, plants inhabiting higher elevations are exposed to similar environmental conditions (e.g., extremes in daily temperature, shorter growing season, limited water supply, exposure to wind and colder temperatures) as plants at lower elevations but higher latitudes (Forbes 1997; Sohlberg and Bliss 1984). When latitude is not considered E. violaceus does appear to be found at higher elevations than E. alaskanus s.1. taxa which may explain the current perception that E. violaceus is found at higher elevation. Contrary to Barkworth et al. (2007) who contend FE. alaskanus s./. is often associated with valleys/ flat areas and E. violaceus restricted to rocky habitats, we found that both E. alaskanus s.l. and E. violaceus were approximately equally likely to occur in either habitat type (Fig. 8). Based on our analysis, habitat cannot be used to differentiate among taxa. Habitat data recorded on herbarium sheets may be too general in order to make inferences about micro-habitat prefer- ences. In order to analyze primary habitat difference future research should include a detailed and standardized procedure for scoring such habitat characteristics. In the past, specimens of E. alaskanus s.1. have not been widely reported throughout British Columbia nor as far south as in our study (Barkworth et al. 2007). With the inclusion of new collections our map (Fig. 8) of E. alaskanus distributions of the two taxa overlap broadly in range, particularly in British Columbia south of 60°N, except on the coast where no E. alaskanus sl occurs. E. alaskanus subsp. hyperarcticus only occurs north of 60°N. Biogeographically, the distributions of E. alaskanus s.l. and E. violaceus are of interest because it is surprising that such closely related species should both have spread and colonized similar and relatively isolated geographical areas, such as Greenland for example, since the last ice-age. Nomenclatural Considerations Deciding how different a taxon must be to warrant consideration as a separate entity has guided this study. In order to validate differen- tiating between species it is necessary to have a character or combination of characters that can discriminate unequivocally between them (Bark- worth 1992). According to Barkworth et al. (2007) infraspecific taxa that show clear morpho- logical and ecological distinctions are treated as subspecies. Despite a large sample size, wide geographic breadth and inclusion of morpholog- ical characters currently used to discriminate between E. alaskanus s.l. and E. violaceus in the Flora of North America (Barkworth et al. 2007), no clear difference morphologically, geographi- cally or in habitat could be established in our study. According to taxonomic ranking rules 44 MADRONO [Vol. 58 TABLE 6. ANOVA RESULTS OF PC1—3 VERSUS TAXON (ELYMUS ALASKANUS SENSU STRICTO, E. ALASKANUS SUBSP. HYPERARCTICUS AND E. VIOLACEUS). PC1 (R? = 0.1625); PC2 (R? = 0.1359); PC3 (R? = 0.1516). Source DF SS MS F P PCl TAXON 2 172.35 86.17 28.65 <0.001 Error 283 851.22 3.01 Total 285 1023.57 PC2 TAXON 2 114.21 57.10 23.41 <0.001 Error 283 690.32 2.44 Total 285 804.53 PC3 TAXON 2 89.34 44.67 26.45 <0.001 Error 283 477.87 1.69 Total 285 567.20 following the International Code of Botanical Nomenclature a subspecies should be more similar to its parent species than different species are to one another (McNeill et al. 2006). Yet, the most distinct entity in the group studied here was E. alaskanus subsp. hyperarcticus. In fact Bark- worth (1997), after examining specimens of E. alaskanus subsp. hyperarcticus, suggests that the entity 1s so distinct that it should not be included in the same species as E. alaskanus subsp. alaskanus and recommended it be group within E. sajanensis (Nevski) Tzvelev as Tzvelev (1976) had done (Fig. 1). If morphological differences between E. alaskanus s.s. and E. alaskanus subsp. hyperarcticus warrant subspecies designation than how could less variation between E. alaskanus s.s. and E. violaceus warrant species designation? Preparing morphological identification keys when the characters holding a group together are non-morphological is not practical. Based on this study, there is no meaningful method to separate North American E. alaskanus s.s. and E. violaceus either morphologically or geographical- ly. Thus, we propose a nomenclatural reconsid- eration of the E. alaskanus s.s. and E. violaceus complex based on the specimens used in this study and suggest that Elymus alaskanus is most correctly applied to all specimens that we examined following the International Code of Botanical Nomenclature (McNeill et al. 2006). Concurrent with the treatments of Love (1984), Cody (1996) and Barkworth et al. (2007), E. alaskanus subsp. hyperarcticus should continue to be treated as a subspecies of EF. alaskanus. Sub- specific recognition is warranted for E. alaskanus subsp. Ayperarcticus based on glume and lemma trichome length. With respect to this feature, Barkworth et al. (2007) consider the trichomes of E. alaskanus subsp. alaskanus up to 0.2mm long and E. alaskanus subsp. hyperarcticus trichomes 0.2-0.5mm long. We observed that some tri- chomes of FE. alaskanus subsp. alaskanus could reach 0.3mm rather than 0.2mm and some trichomes of E. alaskanus subsp. hyperarcticus could reach 0.6mm. Also, glume trichomes exceeded the glume margins in every specimen of E. alaskanus subsp. hyperarcticus. In the future, an analysis in which trichome density is quantitatively assessed may be useful. We recommend the name E. alaskanus subsp. alaskanus continue to be used for those specimens with glabrous glumes or glumes covered sparsely by trichomes following Barkworth et al. (2007). Unlike the treatment in the Flora of North America (Barkworth et al. 2007), we believe E. violaceus should not be regarded as a separate species from EF. alaskanus for those specimens with relatively long glumes. If recognized at all, it should be considered a subspecies of E. alaska- nus. At the sub-specific level, the epithet “‘latiglu- mis’ has priority following Article 11.4 of the International Code of Botanical Nomenclature (McNeill et al. 2006). The most appropriate name for those entities with relatively long glumes is E. alaskanus subsp. latiglumis rather than E. viola- ceus which would be the name that takes priority at the specific level. It would be practical to follow the treatment of Barkworth et al. (2007) and call specimens with glumes 1/3—2/3 as long as the adjacent lemmas E. alaskanus subsp. alaska- nus or E. alaskanus subsp. hyperarcticus (depend- ing on trichome length) and specimens with glumes 3/4 as long as, to slightly longer than the adjacent lemmas, E. alaskanus subsp. latiglu- mis. Based on our observations, there is no evidence for a third taxon in the complex, namely E. violaceus, within the region of our study. Having not compared E. violaceus specimens used in this study to Scandinavian and Green- landic specimens we cannot comment on whether or not they are similar entities to those found in British Columbia. For a thorough taxonomic revision of the complex, field and population studies over the whole circumboreal distribution must be made. Common garden experiments would be useful to examine specific morpholog- ical character differences as well. This study illustrates the challenges to taxon- omists of creating effective dichotomous keys that reflect biological reality. We attempted to differentiate between E. alaskanus and E. viola- 2011) HARRISON AND HEBDA: MORPHOMETRIC ANALYSIS OF VARIATION IN ELYMUS 45 2000 1800 on £ i600 Cc Oo 7 1400 > a” LLJ 1200 1000 TAXON A HV A HV A HV A HV Latitude All latitudes <55N 55-60N >60N Fic. 6. Mean elevation (m) of taxa for 4 categories of latitude: (1) all latitudes (A: n = 54, H: n = 7, V: n = 128); (2) <55°N (A: n = 7, V: n = 83); (3) 55°N—-60°N (A: n = 26, V: n = 29); (4) >60°N (A: n = 21, Hin = 7, Vin = 16). Bars are one standard error from the mean. E. alaskanus (A); E. alaskanus subsp. hyperarcticus (H); E. violaceus (V). ceus using published diagnostic features but were FE. alaskanus and E. violaceus was too great to unable to do so using morphological characters, discriminate between taxa. We also found that E. habitat preferences, or geographic distribution. violaceus and E. alaskanus inhabit similar habi- We determined that the range of overlap of tats and have overlapping geographic ranges and significant morphological characters examined of elevations. Our analysis indicates that E. alaska- 140 120 Valleys and Flat Areas 100 80 60 Rock 40 y Number of specimens 20 E. alaskanus E. a. subsp. hyperarcticus E. violaceus FIG. 7. A mosaic plot for habitat type and taxa. The stripped bars represent the number of specimens found in valleys and flat areas and the black bars represent the number of specimens found in rocky habitats. E alaskanus sensu Stricto n = 89; E. alaskanus subsp. hyperarcticus n = 15; E. violaceus n = 141. 46 MADRONO [Vol. 58 Taxon @ E. alaskanus + E. alaskanus subsp. hyperarcticus E. violaceus EE ee Kilometers FiG. 8. Geographic distribution of Elymus alaskanus sensu stricto, E. alaskanus subsp. hyperarcticus and E. violaceus specimens from Alaska, Alberta, British Columbia, Northwest Territories, and Yukon used in this study. 2011] nus and E. violaceus are potentially the same species with three infraspecific subspecies includ- ing E. alaskanus subsp. alaskanus, E. alaskanus subsp. Jatiglumis and E. alaskanus subsp. hyper- arcticus. New geographic distribution records of specimens, particularly in British Columbia, should be included in future maps of the species ranges. For future analysis we recommend a similar analysis of other closely related species such as E. scribneri (Vasey) M.E. Jones and E. trachycaulus (Link) Gould with which E. viola- ceus has been known to form intermediates and E. macrourus (Turcz.) Tzvelev of which large specimens of E. alaskanus resemble (Barkworth et al. 2007). Further morphological analysis in combination with genetic studies including the European and eastern North American part of range may help clarify relationships between taxa. Knowledge concerning genetic relationships among these taxa is still incomplete, but the accumulation of information suggests a close genetic relationship between E. alaskanus and E. violaceus, thus supporting our findings (Zhang et al. 2000, 2002; Sun and Salomon 2003; Sun et al. 2006). Using morphological types based on spike and vegetative characters, Zhang et al. (2000) investigated genetic variation and structure among Elymus alaskanus populations from a broad geographical area and found that allozyme patterns revealed clear similarities among types of “‘tall hyperarcticus’’, “‘hyperarcticus”, “‘latiglu- mis’, “‘virescens’, and ‘“‘violaceus’’. The taxon ‘violaceus’ was found to be more similar to ‘‘hyperarcticus” and “‘latiglumis” then to “‘vires- cens’’ (Zhang et al. 2000). Zhang et al (2002) and Sun and Salomon (2003) report that morpholog- ical types “violaceus” and “‘latiglumis’’ are genetically more similar to each other than to “‘hyperarcticus’, though later Sun et al. (2006) found a close genetic relationship between E. alaskanus subsp. hyperartcicus and E. violaceus. Future genetic studies should clarify how differ- entiation among morphological types was made, particularly between “‘violaceus”’ and “‘latiglumis”’ types given that these are currently regarded as synonyms (Stewart and Barkworth et al. 2001; Soreng et al. 2003; Barkworth et al. 2007). Studies which correlate morphology with genetic variability may help clarify the relationships between taxa. ACKNOWLEDGMENTS We would like to thank Geraldine Allen, Mary Barkworth, Adolf Ceska, Valerie Huff, Ken Marr and David Mazzucchi who offered many helpful suggestions to improve this paper. A special thank you to Jean Richardson and Zoé Lindo for editing the manuscript and help with the statistical analysis. Thank you to Brad Vidal for his assistance in mapping. We also thank the curators of the CAN, UBC, and US herbaria for the loan of specimens and John Pinder-Moss for his assistance in the herbarium at the Royal BC Museum (V). HARRISON AND HEBDA: MORPHOMETRIC ANALYSIS OF VARIATION IN ELYMUS 47 LITERATURE CITED ARCVIEW 9.3. © Copyright 2008. ESRI. 380 New York St. Redlands, CA. BARKWORTH, M. E. 1992. Taxonomy of the Triticeae: a historical perspective. Hereditas 116:1—14. 1997. Elymus alaskanus. Nomenclature set. Website http://herbarium.usu.edu/openherbarium/ elymalnm.html [accessed 11 April 2011]. , J. J. N. CAMPBELL, AND B. SALOMON. 2007. Elymus. Pp. 289-326 in Flora of North America Editorial Committee (eds.), Flora of North America North of Mexico, Vol. 24. New York, NY. BAUM, B. R., C. YEN, AND J. L. YANG. 1991. Roegneria: its generic limits and justification for its recognition. Canadian Journal of Botany 69:282-294. BEETLE, A. A. 1952. New names for Wyoming wheatgrasses. Rhodora 54:195—196. Copy, W. J. 1996. Flora of the Yukon Territory. NRC Research Press, Ottawa, ON. DEweEy, D. R. 1983a. Historical and current taxonomic perspectives of Agropyron, Elymus, and related genera. Crop Science 23:637—642. . 1983b. New nomenclatural combinations in the North American perennial Triticeae (Gramineae). Brittonia 35:30-33. . 1984. The genomic system of classification as a guide to intergeneric hybridization in the perennial Triticeae. Pp. 209-279 in J. P. Gustafson (ed.), Gene manipulation in plant improvement. Plenum Publishing Corporation, New York, NY. DiAz, O., B. SALOMON, AND R. VON BOTHMER. 1999. Genetic variation and differentiation in Nordic populations of Elymus alaskanus (Scrib. ex Merr.) Love (Poaceae). Theoretical and Applied Genetics 99:210-217. DorE, W. G. AND J. MCNEILL. 1980. Grasses of Ontario. Monograph 26, Research Branch. Agri- culture Canada. Ottawa, ON. Drosow, V. P. 1916. Trudy Moskovskogo Instituta Narodnogo Khozyaistva 16:84. FORBES, B. C. 1997. Tundra disturbance studies. IV. Species establishment on anthropogenic primary surfaces, Yamal Region, northwest Siberia, Russia. Polar Geography 21:79—100. HitrcHcock, A. S. 1951. Manual of the grasses of the United States. Miscellaneous Publications No. 200. U.S. Department of Agriculture, Washington, DC. 2nd ed. revised by A. Chase. 1971. Dover Publications, Inc., New York, NY. . 1969. Gramineae. Pp. 384-725 in C. L. Hitch- cock, A. Cronquist, M. Ownbey, and J. W. Thompson (eds.), Vascular plants of the Pacific Northwest, Part I. University of Washington Press, Seattle, WA. AND A. CRONQUIST. 1973. Flora of the Pacific Northwest. University of Washington Press, Seat- tle, WA. HULTEN, E. 1968. Flora of Alaska and neighboring territories. Stanford University Press, Palo Alto, CA. Love, A. 1980a. IOPB chromosome number reports LXVI. Poaceae-Triticeae-Americanae. Taxon 29: 166-169. . 1980b. Chromosome number reports LXVII. Poaceae-Triticeae. Taxon 29:350-351. A MADRONO 1984. Conspectus of the Triticeae. Feddes Repertorium 95:425—521. AND D. LOVE. 1956. Cytotaxonomical conspec- tus of the Icelandic flora. Acta Horti Gothobur- gensis 20:188. AND . 1976. Nomenclatural notes on Arctic plants. Botaniska Notiser 128:497—523. MCNEILL, J., F. R. BARRIE, H. M. BURDET, V. DEMOULIN, D. L. HAWKSWORTH, K. MARHOLD, D. H. NICOLSON, J. PRADO, P. C. SILVA, J. E. SKOG, J. H. WIERSEMA, AND N. J. TURLAND (eds. and compilers). 2006. International code of botan- ical nomenclature (Vienna Code) adopted by the Seventeenth International Botanical Congress, Vienna, Austria, July 2005. Website http://ibot. sav.sk/icbn/main.htm [accessed March 2009]. MELDERIS, A. 1980. Elymus. Pp. 192-198 in T. G. Tutin, V. H. Heywood, N. A. Burges, D. M. Moore, D. H. Valentine, S. M. Walters, and D. A. Webb (eds.), Flora Europaea, Vol. 5. Cambridge University Press, Cambridge, UK. MINITAB INC. 2007. Version 15.1.30.0. Minitab Inc., State College, PA. Moss, E. H. 1983. Flora of Alberta, 2nd ed., revised, J. G. Packer. University of Toronto Press, Toron- to, ON. NEvVSKI, S. A. 1934. Tribe Hordeae Benth. Pp. 264 in V.L. Komarov (ed.), Flora of the U.S.S.R., Vol. 2, Leningrad, USSR. POLUNIN, N. 1940. Botany of the Canadian eastern Arctic. Part | Pteridophyta and Spermatophyta. National Museum of Canada, Bulletin No. 92. Canada Department of Mines and Resources, Ottawa, ON. . 1959. Circumpolar Arctic flora. Clarendon Press, Oxford, UK. PORSILD, A. E. AND W. J. Copy. 1980. Vascular plants of continental NWT, Canada. National Museum of Natural Sciences, National Museums of Cana- da, Ottawa, ON. POJAR, J. AND A. MACKINNON. 1994. Plants of the Pacific Northwest coast revised: Washington, Oregon, British Columbia and Alaska. B.C. Ministry of Forests and Lone Pine Publishing, Vancouver, BC. RYDBERG, P. A. 1909. Distichlis dentata sp. nov. Bulletin of the Torrey Botanical Club 36:539. SCOGGAN, H. J. 1978. The Flora of Canada: Part 2. National Museum of Natural Sciences, Ottawa, ON. SCRIBNER, F. L. 1900. American grasses—III (illustrat- ed). Descriptions of the tribes and genera. Bulletin 20. United States Department of Agriculture, Division of Agrostology. Washington, DC. AND E. D. MERRILL. 1910. The grasses of Alaska. Contributions from the United States National Herbarium 13:47—92. AND J. G. SMITH. 1897. Native and introduced species of the genera Hordeum and Agropyron. United States Department of Agriculture, Division of Agrostology, Report of the Agrostologist 4:28. SOHLBERG, E. H. AND L. C. BLISS. 1984. Microscale pattern of vascular plant distribution in two high arctic plant communities. Canadian Journal of Botany 62:2033-—2042. SORENG, R. J., G. DAVIDSE, P. M. PETERSON, F. O. ZULOAGA, E. J. JUDZIEWICZ, T. S. FILGUEIRAS, AND O. MORRONE. 2003. Catalogue of new world [Vol. 58 grasses (Poaceae): IV. Subfamily Pooideae. Con- tributions from the United States National Her- barium 48:1—730. STEWART, H. AND M. BARKWORTH. 2001. Elymus. Pp. 132-135 in G. W. Douglas, D. V. Meidinger, and J. Pojar (eds.), Hlustrated flora of British Columbia, Vol. 7: Monocotyledons (Orchidaceae through Zosteraceae). B.C. Ministry of Sustainable Resource Management and B.C. Ministry of Forests, Victoria, BC. SUN, G. AND W. LI. 2005. Molecular diversity of species of the Elymus trachycaulus complex and their relationships to non-North American taxa. Plant Systematics & Evolution 256:179-191. AND B. SALOMON. 2003. Microsatellite vari- ability and heterozygote deficiency in the arctic- alpine Alaskan wheatgrass (Elymus alaskanus) complex. Genome 46:729—737. , J. SHEE, AND B. SALOMON. 2006a. Molecular diversity and relationships among Elymus trachy- caulus, E. subsecundus, E. virescens, E. violaceus, and E. hyperarcticus (Poaceae: Triticeae) as deter- mined by amplified fragment length polymorphism. Genome 49:1160—1169. , H. TANG, AND B. SALOMON. 2006b. Molecular diversity and relationships of North American Elymus trachycaulus and the Eurasian E. caninus species. Genetica 127:55—64. TURCZANINOW, N. S. 1856. Flora baicalensi-dahurica, Vol. 2. Moscou University Publication, Moscow, USSR. TZVELEV, N. N. 1976. Tribe 3. Triticeae Dumort. Pp. 147-181 in A. A. Fedorov (ed.), Poaceae URSS. Nauka Publishing House, Lenningrad, USSR. WELSH, S. L. 1974. Anderson’s flora of Alaska and adjacent parts of Canada. Brigham Young Uni- versity Press, Provo, UT. ZHANG, X. O., B. SALOMON, R. VON BOTHMER, AND O. DtAz. 2000. Patterns and levels of genetic differentiation in North American populations of the Alaskan wheatgrass complex. Hereditas 133:123—132. , , AND . 2002. Application of random amplified polymorphic DNA markers to evaluate intraspecific genetic variation in the Elymus alaskanus complex (Poaceae). Genetic Resources and Crop Evolution 49:397—407. APPENDIX | SPECIMENS EXAMINED * = Accessions included in mapping analysis. ° = Accessions included in habitat analysis. Herbarium abbreviations: V = Royal BC Museum; CAN = Canadian Museum of Nature; UBC = University of British Columbia; US = United States National Herbarium. : Elymus alaskanus (Scribn. & Merr.) A. L6ve— CANADA. ALBERTA. UBC: 62034*°, 82554*. BRIT- ISH COLUMBIA. V: 61973*, 76995*°; 105917*°, 106136*°, 125951*, 16671*, 17803*, 24489*, 79126*°, 91608*°, 194719*°, 195414*°, 195468*°, 196196*°, 196201*°, 196244*°, 196245*°, 196433*°, 198508*°, 198528*°, 198554*°, 198634*°, 198638*°, 198656*°, 198740*°, 198752*°, 198759*°, 198762*°, 198879*°, 2011] 198883*°, 198895*°, 198926*°, 198931*°, 198961*°, 199623*, 199783*°: UBC: 169655*°, 42328*°. NORTH- WEST TERRITORIES. CAN: 127440*°, 127441*°, 127442*°, 127443*°, 127444*, 200030*°, 203081*°, 203082*°, 203084*, 268362*°, 270867*, 279113*, 279114A*°, 279322*°, 39283*°, 39286*°, 39288*, 39329*, 527868*°, 529498*, 530883*°, 530891*°, 582469*°, 584015*°, 585091*°, 585093*°; UBC: 111282*°, 113135*, 113185*, 171348*°, 171489*°, 171504*°, 171572*°, 36871*, 37095*°, 90155*°: V25042*°. QUEBEC. V: 114219. YUKON. CAN: 276347*°, 276351*°, 276598*°, 303292*°, 306804*°, 318450*, 39772*°, 454931*°, 53085*°, 549414D*°; UBC: NOSIS" 181579; 27873" .99014*",99023"* 99/437". Ve S274 =. MIS228"—, 122789", 137591. A375°2*., 137610*°, 137611*°. USA. ALASKA. CAN: 211188*°, 211190*°, 211191*°, 248032*°, 274084*°, 211188*°, 211190*°, 211191*°, 248032*°, 274084*°, 276349*°, 367095*°, 514133°, 514134*°; US: 592341 holotype. | Elymus alaskanus subsp. hyperarcticus (Polunin) A. Léve & D. L6ve—CANADA. NORTHWEST TERRI- TORIES. CAN: 203083*, 203085*°, 225486°, 279114B*°. NUNAVUT. UBC: 184460*°; US: 203113 isotype. YUKON. CAN: 260928*°, 270276*°, 454932*°; UBC: 99024*, 115538*°, V: 198867*°. USA. ALASKA. CAN: 225257*°, 270277*°, 274083*°, 318764*°, 366745*°, 367096*°; V: 37905*. Elymus violaceus (Hornem.) J. Feilberg—CANADA. ALBERTA. CAN: 514030*; UBC: 21928*°, 77875*°; V: 25062*°. BRITISH COLUMBIA. UBC: 145869*°, ba5871*°,. 145872"... 155889*",. 155890"°,. 156195**., 17254", 17375", T7410"°, 17413", 17429*,. 21923**, HARRISON AND HEBDA: MORPHOMETRIC ANALYSIS OF VARIATION IN ELYMUS 49 21925*°, 220654*°, 45622*°, 58312*, 60491*°, 67864*", 86401*°, 86433*°, 98384*°, 988386*°; V: 123194, 104896*°, 106180*°, 106188*°, 107666*°, 112825*°, 11309*°, 115058*°, 118641*°, 118669*°, 118989*°, I195Z29°°. A 9606". “L19616"";. 1L97S8"". 119767". 120201". 120270**,.. 120310**, .127184*",;, 127185*; I27186%>, I27187", I3E360"". 132200"",. 137599", 13699*, 137663*°, 141176*°, 141179*°, 147702*°, 147703*, 147705*°, 148290*°, 160614*°, 160623*, 163871*, 16741*, 170331*°, 17763*, 184000*°, 188109*°, 18826*, 189980*°, 189981*°, 191286*°, 191307*°, 191896*°, 196248*°, 199824*°, 200057*°, 200534*°, 200900*°, 200910*°, 200979*°, 201806*°, DOTS? 2IO20* 5 27890" 4 2180)" 0232" 4 318337: 32552*°, 36900*°, 36919*°, 36929*°, 36943*°, 404*, 44524* 69404* 1695", 83139* 87482* 91279" 91878* HR08020*, °, 44565*°, “7 1451" ", 79578*°, ooh Le. °, 88408*' °, 91346*" °, 92000* 117436*°. 48251*°, 71457*°, 80869*", 83172*° °, 88434*° e DISI4 °, 92641*,93241*°, NORTHWEST TERRITO- 58714". 59089", o1972*, 75509*°, 83134*°, , 83780*", , 88444*", 91562", 76343*°, Sa135". 87478*", 91014*°, 91576*", 96089*", 10927 Bs 137 87478**, 91060*°, 91863", 96753." 2 RIES. CAN: 39289*; UBC: 182645*, 18398*, 83427*, 90154*°, 96157*°; V: 141141*°, 141142*°. ONTARIO. UBC: 17437. USA. ALASKA. CAN: 514025*°, 514027*°, 514028*°. MONTANA. V: 44690; US: 556692 holotype. UTAH. V: 141282°. WASHING- TON. V: 96357°, 137603°. YUKON. UBC: 99022*°, 99658*°; V: 137595*°, 137604*°, 137605*, 137607*°, 137608*°, 137609*°, 137612*°, 137613*°, 87601*°, O1097* 37) 38"" S185)", OSSole”, MADRONO, Vol. 58, No. 1, pp. 50-56, 2011 CHROMOSOME COUNTS AND TAXONOMY OF MENTZELIA THOMPSONI (LOASACEAE) JOSHUA M. BROKAW' School of Biological Sciences, P.O. Box 644236, Washington State University, Pullman, WA 99164-4236 josh.brokaw@acu.edu MICHAEL D. WINDHAM Department of Biology, Duke University, Durham, NC 27708-0338 LARRY HUFFORD School of Biological Sciences, P.O. Box 644236, Washington State University, Pullman, WA 99164-4236 ABSTRACT The species Mentzelia thompsonii Glad was published in 1976 based solely on herbarium material, and it is the only species in Mentzelia section Trachyphytum that has not been examined cytogenetically. Here we report that M. thompsonii is diploid (2n = 18), making it the easternmost diploid species in section Trachyphytum and the only one that does not occur in California. The diploid status and edaphic specialization of M. thompsonii suggest that it is a paleoendemic isolated from other diploids in Trachyphytum by specialization during Pleistocene climate change. Our investigations have also uncovered confusion in the literature and herbaria regarding the taxonomy of M. thompsonii and its overall place within Mentzelia. Mentzelia thompsonii has been synonymized with Mentzelia humilis (Urb. & Gilg) J. Darl. (a member of section Bartonia) in several prominent databases and herbaria. To the contrary, our studies reveal that M. thompsonii is distinct from M. humilis; furthermore, as a unique component of Colorado Plateau diversity, it is critical for inferences of biogeographic evolution in section Trachyphytum. Key Words: Acrolasia humilis, biogeography, diploid, Mancos Shale, Mentzelia humilis, Mentzelia thompsonii, Pleistocene climate change, polyploidy. Mentzelia thompsonii Glad (Fig. 1) is a small annual confined to the Colorado Plateau of the southwestern United States. Its range extends from the Four Corners region north along the Utah-Colorado border to the Uinta Basin (Fig. 2). The species is an edaphic endemic, usually occurring on barren, salty soils derived from the Mancos Shale Formation (Glad 1976; Holmgren et al. 2005; Brokaw 2009). Mentzelia thompsonii is one of the most recently described species in section Trachyphytum (Glad 1976), and its evolutionary significance within this group is only now becoming apparent (Brokaw and Hufford 2010a, b). Prior to cytogenetic surveys of the group, only eight North American species were recognized in section Trachyphytum (Loa- saceae) (Darlington 1934). However, the discov- ery of extensive polyploidy (Zavortink 1966) and coincident reproductive barriers subsequently led to the recognition of over 20 species in the southwestern United States alone. Mentzelia thompsonii was described ten years after Zavor- tink’s (1966) biosystematic revision of Trachy- phytum, and it is the only North American species 'Present address: Department of Biology, Abilene Christian University, Abilene, TX 79699-7868, U.S.A. currently lacking a chromosome count. Recent work has suggested that section Trachyphytum represents a monophyletic group exhibiting complicated polyploid evolution (Hufford et al. 2003; Brokaw and Hufford 2010a, b), and verification of the ploidal level of M. thompsonii is vital to the interpretation of molecular evolution and gene flow in the group. The goals of the study were: 1) to collect chromosome data for M. thompsonii and 2) to develop a set of chromosomally vouchered molecular and morphological samples for sys- tematic analyses. While these cytogenetic analy- ses have been essential to determine M. thompso- nii's role in the polyploid complexes of section Trachyphytym, our investigations have also led to new insights regarding the biogeography of Trachyphytum and resolved confusion in the literature and herbaria regarding the taxonomy and validity of M. thompsonii. MATERIALS AND METHODS Populations examined in this study were selected to represent the northwestern (Uinta Co., UT, Brokaw 234) and southeastern (San Juan Co., NM, Brokaw 345) limits of the FiG. 1. Habit of Mentzelia thompsonii Glad. Photo- graph courtesy of W. A. Weber (University of Colorado). distribution of M. thompsonii. Chromosome counts were made from field-collected microspo- rocytes fixed in Farmer’s solution (3 parts 95% ethanol: 1 part glacial acetic acid) at the time of voucher collection. Following the procedures outlined by Windham (2000), fixed materials were stored at —20°C and transferred to 70% ethanol immediately before making slides. Dissected anthers were macerated in a drop of 1% acetocarmine stain, which was mixed 1:1 with Hoyer’s solution prior to setting the cover slip and squashing. Slides were examined with an Olympus BH-2 phase contrast microscope, and representa- tive cells were photographed using Kodak Tech- nical Pan 2415 film. The voucher specimens, Brokaw 234 (WS375612) and Brokaw 345 (WS375773), have been deposited at the Marion Ownbey Herbarium (WS). Additional duplicate vouchers have been sent to ACU and COLO. RESULTS AND DISCUSSION Analyses of microsporocytes undergoing mei- osis revealed that the chromosome number of both the northwestern and southeastern popula- BROKAW ET AL.: CHROMOSOME COUNT OF M. THOMPSONII 51 tions of M. thompsonii is n = 9. The chromo- somes consistently formed nine bivalents during the reductional division (Fig. 3a) and _ these segregated normally during anaphase II to produce four daughter cells containing nine chromatids each (Fig. 3b). The base chromosome number of Mentzelia section Trachyphytum is x = 9, making M. thompsonii a diploid. It is the easternmost diploid in the section and the only one that does not occur in California (Fig. 2). Further, it is the only diploid species whose current distribution does not overlap with any other diploid in Trachyphytum. Evolutionary Ecology The discovery that Mentzelia thompsonii is diploid has important implications for our understanding of biogeography and evolution in section Trachyphytum. The lineage of M. thomp- sonii 1s nested within the section Trachyphytum clade (Brokaw and Hufford 2010a) leading to the most parsimonious hypothesis that M. thompso- nii represents a range extension far from the California origin of the Trachyphytum diploids. The section has its greatest species richness and representatives of all its major clades in southern California. The polyploid taxa generally have larger distributions, extending further north and east than those of diploids (Zavortink 1966). Prior to our investigation of M. thompsonii, only polyploid taxa were known to have gotten as far east as the Colorado Plateau. Given that polyploids are derived from diploids and thus more recently evolved, analyses of Trachyphytum lacking M. thompsonii would suggest that range expansions were associated novel trait combina- tions acquired during or following polyploidiza- tion. However, the geography of the diploid M. thompsonii represents a major exception to this generalization, suggesting that other factors must be considered. Patterns of edaphic specialization in Trachy- phytum may partly explain the disjunct range of M. thompsonii. Only two other species in Trachyphytum, the tetraploid M. mollis M. Peck and the octoploid M. packardiae Glad, occur entirely outside California; both are limited to unusual soils (Glad 1975, 1976). Although soils were not available to her for chemical analyses, Glad (1976) first noted that shales and grey clays of the Mancos Formation were commonly listed as substrates on specimen labels for M. thompso- nii. With a wider sampling of populations, it is now evident that M. thompsonii is limited almost exclusively to the Mancos Shale and other Cretaceous marine sediments of the Colorado Plateau (Holmgren et al. 2005; Brokaw 2009). Thus, all three Trachyphytum species absent from California appear to be associated with substrate specialization. 52 MADRONO 115° [Vol. 58 FiG. 2. Range of Mentzelia thompsonii in eastern Utah, western Colorado, and Four Corners region (white) and combined ranges of all other diploid species in Mentzelia section Trachyphytum in California, southern Oregon, southwestern Idaho, western Nevada, southwestern Arizona, and northwestern Mexico (black). These observations suggest that the simplest biogeographic hypothesis (a one-way expansion from California) may be insufficient to fully explain species distributions in Trachyphytum. The range of M. thompsonii shows that diploids have accomplished substantial range expansion outside of California. The current abundance of polyploids in intervening regions of the Great Basin and western Colorado Plateau suggest that a similar distribution of ancestral diploids is at least plausible. It is possible that diploid popu- lations formerly in this region have been dis- placed by competition. This line of reasoning, coupled with observed edaphic specialization, suggests that M. thompsonii may be a paleoen- demic, 1.e., a species isolated through extinctions of close relatives (Stebbins and Major 1965). Major migrations and extinctions of diploid populations could have been driven by dramatic shifts in vegetation during Pleistocene climate change (Dynesius and Jansson 2000; Thompson and Anderson 2000; Minnich 2007). During vegetational shifts, the ancestors of M. thompsonii may have persisted in northeastern portions of the diploid ranges by specializing for edaphically stressful habitats where most competing vegeta- tion was excluded. It is likely that shifting diploid ranges during the Pleistocene facilitated hybridization, leading to the extensive generation of allopolyploids in Trachyphytum documented by Brokaw and Hufford (2010b). Mentzelia thompsonii is one of the few diploids in Trachyphytum lacking allo- polyploid descendents (Brokaw and Hufford 2010b), which is not surprising considering its current isolation from other extant diploids. The 2011] FIG. 3. BROKAW ET AL.: CHROMOSOME COUNT OF M. THOMPSONII N Ww i « we a ' : ’ rae ee ae PY & Se ‘, ws ig : ~, 2 4 Chromosomes of Mentzelia thompsonii (Brokaw 234) at: A) Metaphase I cell showing nine pairs of chromosomes; B) Anaphase II cell showing four daughter nuclei each containing nine chromatids. Scale bars = 5 um. species almost certainly had fewer opportunities than most diploids for allopolyploid hybridiza- tions in Pleistocene ice age refugia. However, molecular data have implicated M. thompsonii in One introgressional event without polyploidiza- tion (Brokaw and Hufford 2010a, b). Brokaw and Hufford (2010a) showed evidence of recom- bination between nuclear genes of M. thompsonii and the Sonoran Desert diploid M. affinis Greene, suggesting gene flow between these species. Although M. thompsonii and M. affinis are not currently sympatric, populations of M. affinis approach the range of M. thompsonii more closely than those of any other diploid in Trachyphytum. Mentzelia affinis occurs in south- ern and western Arizona, separated from the nearest M. thompsonii populations (in the Four Corners region) by less than 400 km. The suitability of the intervening habitat during the Pleistocene is unclear, but it is interesting to note that a similar geographic pattern has been observed in the genus Boechera (Brassicaceae). In this instance, microsatellite data (Windham et al. unpublished) reveal that B. perennans (S. Watson) W. A. Weber (with a southern Arizona range similar to M. affinis) has hybridized with B. pallidifolia (Rollins) W. A. Weber, a Colorado Plateau endemic that reaches its southern limit near the Four Corners (like M. thompsonii). The striking similarity of these two cases suggests that genetic interactions between species occupying the warm deserts of southern Arizona and the cool deserts of the Four Corners region may be more common than previously supposed. Molecular data have revealed that unexpected interactions between species (like those inferred between M. thompsonii and M. affinis) are relatively common in Mentzelia section Trachy- phytum (Brokaw and Hufford 2010a, b). These provide intriguing evidence that Pleistocene migrations have contributed to complicated patterns of molecular evolution, ecological spe- cialization, and a burst of allopolyploid specia- tion (Brokaw and Hufford 2010a, b). The discovery that M. thompsonii is diploid adds another piece to the puzzle, allowing us to view our biogeographic hypotheses in a new light and critically examine the effects of geographic isolation on species evolution in Mentzelia section Trachyphytum. Taxonomic Status Mentzelia thompsonii is a poorly known taxon, rarely collected and confined to unusual sub- strates in a region of the United States that 1s only now receiving the botanical exploration it deserves (Heil et al. in press). This largely explains why it was overlooked by both previous monographic treatments of Trachyphytum (Dar- lington 1934; Zavortink 1966). Not only has this led to the delayed cytogenetic study of M. thompsonii, but it has also contributed to a complicated nomenclatural history and, ultimate- ly, to the synonymization of the taxon in prominent databases and herbaria. In fact, M. thompsonii is still listed as “‘not accepted” by the Integrated Taxonomic Information System (ITIS 2011) and the PLANTS Database (USDA- NRCS 2011). Through a series of nomenclatural and taxo- nomic errors, M. thompsonii (Fig. 4) has been incorrectly synonymized with M. humilis (Urb. & Gilg) J. Darl. (Fig. 5), a distantly related member of section Bartonia. The initial confusion in the taxonomy of M. thompsonii stems from a long- 54 MADRONO [Vol. 58 FOOL) DOIFAL AY FIG. 4. Holotype of Mentzelia thompsonii Glad. running disagreement between authors who recognize Trachyphytum as a section of the genus Menizelia (Torrey and Gray 1840; Urban and Gilg 1900; MacBride 1918; Darlington 1934; Thompson and Roberts 1974; Hufford et al. 2003) and authors who segregate the group as the genus Acrolasia (Presl 1831; Rydberg 1903; Davidson 1916; Weber and Wittman 2001). Following Rydberg’s (1903) recircumscription of Acrolasia, the entity now known as M. thompsonii was first described and published as Acrolasia humilis by Osterhout in 1922. The epithet humilis had been used previously in Mentzelia for a taxon treated by Urban & Gilg (1900) as a variety of M. pumila Torr. & A. Gray. Subsequently, Darling- wo 0} Judith B. Glad Yume, OREGON STATE UNIVERSITY PLANTS OF COLORADO Herbariom of the Usiversity af Colorsdo Houlter Mentgelis 8 MONTROSE CO.: EYP Ailis of Montrose, O000 ThrRs t+ Ma ton (1934) raised M. pumila var. humilis Urb. & Gilg of section Bartonia to the rank of species as M. humilis. Unfortunately, she also treated A. humilis and M. humilis as homotypic and incorrectly synonymized the name Acrolasia humilis under Mentzelia humilis. When Glad (1976) named M. thompsonii, she did not mention Acrolasia humilis, a name that, nevertheless, could not have been transferred to Mentzelia because of Darlington’s (1934) earlier elevation of M. pumila var. humilis to species status. Thus, M. thompsonii became the accepted name (in treatments recognizing Trachyphytum as a section of Mentzelia; e.g., Holmgren et al. 2005) for the species originally described as Acrolasia 2011] wi” 4 \ _ . & ve Cr Le » WD PSO. Poche, Cel Be. xs Fd M. Fehrs, A, _ FIG. 5. BROKAW ET AL.: CHROMOSOME COUNT OF M. THOMPSONII 55 wd OL WW WE NTE Lectotype of Mentzelia humilis (Urb. & Gilg) J. Darl. Specimen on right side of sheet is the lectotype (C. Wright 214, 1849) mounted with the non-type specimen (H. N. Patterson s.n., 1875) on the left. humilis. Subsequently, the nomenclatural combi- nation A. thompsonii (Glad) W. A. Weber, was proposed to accommodate use of Acrolasia in treatments of the Rocky Mountain flora (Weber 1984). However, rediscovery of the original description and holotype of A. humilis led to synonymization of A. thompsonii and a return to the use of A. humilis in later treatments (Weber and Wittman 1992). The final taxonomic error occurred when Kartesz (1999) repeated Darling- ton’s (1934) incorrect synonymy of A. humilis with M. humilis and went on to designate M. thompso- nii and A. thompsonii as synonyms of M. humilis. Our reconstruction of the taxonomic history of M. thompsonii indicates that a failure to consult type specimens has played a major role in the current taxonomic confusion surrounding this species. Even cursory examination of types of M. thompsonii and M. humilis (Figs. 4 and_ 5, respectively) reveals that these names do not represent the same taxon. Given the new evidence from cytology and molecular phylogenetics (Bro- kaw and Hufford 2010a) we hope that earlier taxonomic misinterpretations will be put to rest. Mentzelia thompsonii is a distinct species with a unique phylogenetic history and ecological niche, 56 MADRONO and it is an important piece of the evolutionary puzzle that is Mentzelia. ACKNOWLEDGMENTS We thank B. A. Prigge, J. J. Schenk, and J. L. Strother for helpful discussion of the taxonomy of Mentzelia and W. A. Weber for providing a photograph of Mentzelia thompsonii. The following herbaria pro- vided access to specimens used in this study: ACU, COLO, NY, RM, UNM, UT, and WS. Funding for this project was provided by the Betty W. Higinbotham Trust and the Hardman Native Plant Award in Botany. LITERATURE CITED BROKAW, J. M. 2009. Phylogeny of Mentzelia section Trachyphytum: origins and evolutionary ecology of polyploidy. Ph.D. dissertation, Washington State University, Pullman, WA. AND L. HUFFORD. 2010a. Phylogeny, intro- gression, and character evolution of diploid species in Mentzelia section Trachyphytum (Loasaceae). Systematic Botany 35:601—617. AND . 2010b. Origins and introgression of polyploid species in Mentzelia section Trachy- phytum (Loasaceae). American Journal of Botany 97:1457-1473. DARLINGTON, J. 1934. A monograph of the genus Mentzelia. Annals of the Missouri Botanical Garden 21:103—227. DAVIDSON, A. 1916. A revision of the western mentzelias. Southern California Academy of Sci- ences 5:13—18. DyYNEsIUS, M. AND R. JANSSON. 2000. Evolutionary consequences of changes in species’ geographical distributions driven by Milankovitch climate oscil- lations. Proceedings of the National Academy of Sciences, USA 97:9115-—9120. GLAD, J. B. 1975. Taxonomy and ecology of Mentzelia mollis Peck and related species. M.S. thesis, Oregon State University, Corvallis, OR. . 1976. Taxonomy of Mentzelia mollis and allied species. Madrono 23:283—292. HEIL, K...5. O’KANE, AND L. REEVES (eds.). In Press. Flora of the Four Corners. Missouri Botanical Garden Press, St. Louis, MO. HOLMGREN, N. H., P. K. HOLMGREN, AND A. CRONQUIST. 2005. Loasaceae, the Loasa family. Pp. 81-118 in N. H. Holmgren, P. K. Holmgren, and A. Cronquist (eds.), Intermountain flora 2, part B. The New York Botanical Garden Press, New York, NY. HUFFORD, L., M. M. MCMAHON, A. M. SHERWOOD, G. REEVES, AND M. W. CHASE. 2003. The major clades of Loasaceae: phylogenetic analysis using the plastid matK and trnL-trnF regions. American Journal of Botany 90:1215—1228. INTEGRATED TAXONOMIC INFORMATION SYSTEM (ITIS). 2011. Website: http://www. itis.gov [accessed 16 March 2011]. [Vol. 58 KARTESZ, J.T. 1999. A synonymized checklist and atlas with biological attributes for the vascular flora of the United States, Canada, and Greenland, Ist ed. In J. T. Kartesz and C. A. Meacham. Synthesis of the North American flora, version 1.0. North Carolina Botanical Garden, Chapel Hill, NC. MACBRIDE, J. F. 1918. A revision of Mentzelia, section Trachyphytum. Contributions from the Gray Her- barium 56:24—28. MINNICH, R. A. 2007. Climate, paleoclimate and paleovegetation. Pp. 43—70 in M. G. Barbour, T. Keeler-Wolf, and A. A. Schoenherr (eds.), Terres- trial vegetation of California, 3rd ed. University of California Press, Berkeley, CA. OSTERHOUT, G. E. 1922. Two new plants from western Colorado. Bulletin of the Torrey Botanical Club 49:183-184. PRESL, C. B. 1831. Reliquiae Haenkeanae, volume 2, part 1. JoG, Calve, Prague, CZ. RYDBERG, P. A. 1903. Some generic segregations. Bulletin of the Torrey Botanical Club 30:271—281. STEBBINS, G. L. AND J. MAJor. 1965. Endemism and speciation in the California flora. Ecological Monographs 35:1—35. THOMPSON, H. J. AND J. E. ROBERTS. 1974. Loasaceae. Pp. 549-559 in P. A. Munz (ed.), A flora of southern California. University of California Press, Berkeley, CA. THOMPSON, R. S. AND K. H. ANDERSON. 2000. Biomes of western North America at 18,000, 6000 and 0 '*C yr BP reconstructed from pollen and packrat midden data. Journal of Biogeography 27:555—584. TORREY, J. AND A. GRAY. 1840. A flora of North America, volume 1, part 3. Wiley and Putnam, New York, NY. URBAN, I. AND E. GILG. 1900. Monographia Loasa- cearum. Nova Acta Academiae Caesareae Leopol- dino-Carolinae Germanicae Naturae Curiosorum 76:1—370. USDA, NRCS. 2011. The PLANTS Database. Na- tional Plant Data Center, Baton Rouge, LA Website: http://plants.usda.gov [accessed 16 March 2011]. WEBER, W. A. 1984. New names and combinations, principally in the Rocky Mountain flora—IV. Phytologia 55:1—11. AND R. C. WITTMANN. 1992. Catalog of the Colorado flora: a biodiversity baseline. University Press of Colorado, Boulder, CO. AND . 2001. Colorado flora: eastern slope, 3rd ed. University Press of Colorado, Boul- der, CO. WINDHAM, M. D. 2000. Chromosome counts and taxonomic notes on Draba (Brassicaceae) of the Intermountain West. 1: Utah and vicinity. Madro- ho 47:21-28. ZAVORTINK, J. E. 1966. A revision of Mentzelia, section Trachyphytum (Loasaceae). Ph.D. dissertation, University of California, Los Angeles, CA. MADRONO, Vol. 58, No. 1, pp. 57-63, 2011 A NEW SPECIES OF MENTZELIA (LOASACEAE) FROM MONO COUNTY, CALIFORNIA JOSHUA M. BROKAW Department of Biology, Abilene Christian University, Abilene, TX 79699-7868 josh.brokaw@acu.edu LARRY HUFFORD School of Biological Sciences, P.O. Box 644236, Washington State University, Pullman, WA 99164-4236 ABSTRACT A new species Mentzelia monoensis Brokaw and Hufford, endemic to Mono County, California, is described. Mentzelia monoensis, a hexaploid, is most similar to two widespread species, M. montana (Davidson) Davidson (tetraploid) and M. albicaulis (Douglas ex Hook.) Douglas ex Torr. & A. Gray (octoploid). The three species can be distinguished based on differences in floral bracts, fruits, and seeds. Mentzelia monoensis is commonly found in volcanic soils derived from the eruptions of Mono Craters and may exhibit edaphic specializations that limit its distribution. Key Words: endemic, Mentzelia, Mono Craters, polyploidy, systematics. Mentzelia L. section Trachyphytum Torr. & A. Gray (Loasaceae) is a monophyletic group of at least 22 species of annual plants occurring primarily in the western United States with greatest taxon richness in California (Darlington 1934; Zavortink 1966; Hufford et al. 2003; Brokaw and Hufford 2010a, b). The group 1s taxonomically difficult because of overlapping morphological variation among polyploid species (Zavortink 1966; Brokaw and Hufford 2010b). In her revision of the section, Zavortink (1966) noted a single hexaploid population in Mono County, California, which she dubbed ‘M. monoensis” but never validly published. In addition to the rarity of specimens available for study, confusion regarding the status of “‘*M. monoensis”’ has been exacerbated by its similarity to other species in the section. However, recent molecular investigations (Brokaw and Hufford 2010b) have suggested a unique hybrid origin of ““M. monoensis” and stimulated a deeper inves- tigation of its form and distribution. That study indicated ““M. monoensis” is morphologically distinct from other species of section Trachyphy- tum and endemic to Mono Co., California, where it 1s distributed throughout the Mono Craters region. In this paper, we describe ‘““M. mono- ensis’’ as a new species of Mentzelia. MATERIALS AND METHODS During this study, we made field observations and inspected herbarium specimens of all North American species of Mentzelia section Trachy- phytum. Zavortink (1966) based her original recommendation for recognition of ‘M. mono- ensis’ on specimens from the collection Zavortink 2640 and chromosome counts from squashes of microsporocytes from the same population. We have located and included fifteen additional population of ‘M. monoensis’ in morphological comparisons with herbarium specimens, includ- ing an isotype (UC68802, Hall 6577), of M. montana (Davidson) Davidson. Morphological measurements were taken with digital calipers, using a dissecting microscope when necessary. Seed surface characters were assessed using scanning electron microscopy (SEM). Seeds of ““M. monoensis” obtained from the herbarium specimens WS376107 (Zavortink 2640) and WS375796 (Brokaw 368), and seeds of ™M. montana from the herbarium specimens LA100619 (Zavortink 2586) and WS367986 (Brokaw 72), were mounted on metal stubs and coated in gold prior to imaging. Seeds were examined at 12 kV using a Hitachi S-570 scanning electron microscope and micrographs were made at 70X, 80, and 750. Locality data for populations were gathered during fieldwork and from herbarium specimens. Latitude and longitude or UTM coordinates for new collec- tions were made in the field using GPS (WGS 84 map datum). TAXONOMY Mentzelia monoensis J. M. Brokaw & L. Hufford, sp. nov. (Fig. 1). —TYPE: USA, California, Mono Co., along State Hwy 120, 8 mi E of U.S. 395, 16 June 2007, J. M. Brokaw 367 (holotype: WS; isotypes: NY, UC, US). Herba annua, 10-30 cm alta, erecta. Inflor- escentia cymis; bractea integra viridia rarior basis albus. Calyx connatus, sepala 5, subulatus. 58 Fic. 1. MADRONO be. oo Stheieee nena =P 1 Bz AN ; ‘ ‘ Type specimen of Mentzelia monoensis. Scale bar 10 cm. [Vol. 58 MARION OWNBEY HERBARIUM [WS] WASHINGTON STATE UNIVERSITY Mentzelia monoensis J. M. Brokaw & L. Hufford UNITED STATES. CA. Mono Co. Highway 120 north of Mono Mills. In open Jeffrey pine forest. UTM zone: 11S 4196081mN 326371mE Elevation: 2257 meters Collector: Joshua M. Brokaw 367 Date collected: 16 Jun 2007 Loasaceae 2011] Apopetalus, petala 5, usque 4.1 mm longa, apex luteus et basis aurantiacus. Stamina 10—30, usque 3 mm longa. Ovarium inferum, placentae 3 parietalibus; stylus usque 3.3 mm longa. Fructus capsularis, Seminum numerosus, testa colliculate. Annual herbs, 10—30 cm tall; taprooted. Shoots densely pubescent throughout (except as noted) with both needle-like and glochidiate trichomes; needle-like trichomes with pointed apex and erect barbs arranged in many whorls along the needle shaft; glochidiate trichomes usually with apical and several other whorls of recurved barbs along the stalk. Stem erect; axillary branches ascending, straight to curved upward; stem epidermis light tan to salmon-colored, moderately pubescent. Leaves alternate, pubescent with greater density of trichomes on abaxial surface, trichomes needle-like and glochidiate; needle-like trichomes often with pustulose bases; basal rosette leaves 2.1-86 < 0.3—-10 mm, linear, all sessile or some appearing petiolate due to narrowed lamina at base, often not persisting to maturity, margins entire to 12-lobed on distal half, lobes nearly opposite, regular, 1.0—3.8 mm long with rounded apices; lower cauline leaves 6.2—55 x 2.0—7.9 mm, linear to lanceolate, sessile to appearing petiolate, margins entire to 12-lobed on distal half, lobes nearly opposite, regular, 0.5—3.5 mm long with rounded to acute apices; upper cauline leaves up to 34.2 < 8.5 mm, linear to ovate, sessile to clasping, margins usually entire. Inflorescence cymose, bract subtending inferior ovary |1.5—4.5 x 0.4-2.3 mm, entire, green (rarely with incon- spicuously white base). Flowers epigynous; bear- ing a hypanthium at the distal end of the ovary on which the calyx, corolla, and androecium are inserted. Calyx basally connate, five triangular lobes, lobes 1.9-3.1 X 0.8—1.1 mm, apices acute to attenuate, margins entire, pubescent; tri- chomes like those of leaves. Petals five, distinct, 2.1-4.1 * 1.9-3.1 mm, obovate, with yellow apex and orange base, glabrous, apex retuse to rounded with several trichomes at midvein; trichomes with erect barbs arranged in many whorls, apex needle-like, base not pustulose. Androecium yellow; stamens ca. 10-30, ca. 2— 3 mm, those of inner whorls shorter than outer whorls; filaments all filiform. glabrous; anther epidermis papillate. Gynoecium 3-carpellate; ovaries inferior, placentae 3, parietal; styles 2.2— 3.3 mm long, glabrous; stigmas three, lobes appressed, papillate. Fruit a capsule, 6-15 x 2- 3 mm, cylindrical to clavate and tapering near base, erect to curved less than 20°, opening apically, usually without prominent longitudinal ribs, pubescent; trichomes like those of leaves. Seeds ca. 1 X 1 mm, ca. 5—30 per capsule, in more than one row above mid-fruit; seeds above mid- fruit irregular-rounded; seeds below mid-fruit occasionally trigonal prisms with grooves along longitudinal edges; seed coat tan, colliculate BROKAW AND HUFFORD: NEW MENTZELIA FROM MONO COUNTY a9 under 10 magnification, domes less than 1/2 as tall as wide; seed coat cells 25-75 um wide (smaller near hilum), with straight anticlinal walls. Chromosome number: 1966). Phenology: Mentzelia monoensis begins flow- ering at lowest elevations in late May and continues through late July. Plants bearing both flowers and ripe capsules with mature seeds are most common in early to middle July. By early August, most plants have senesced and quickly disintegrate. Etymology: This species is named for the Mono Craters region and Mono Co., California, to which it appears to be endemic. We suggest the common name Mono Craters blazingstar. Distribution: Populations of M. monoensis occur primarily on course pumice soils and disturbed sites near the Mono Craters of Mono Co., California, at 2008-2482 m elevation. They are associated especially with antelope bitter- brush and Jeffrey pine communities and barren pumice slopes. Representative specimens. USA. CALIFOR- NIA. Mono Co.: Mono Craters, along State Hwy 120 about 8 mi. E of U.S. Hwy 395, Thompson 1696 (LA); along State 120, 8 mi. E of U.S. 395, J. Zavortink 2640 (RSA, WS); CA State Hwy 120 N of Mono Mills in barren pumice valley, UTM 11S 4197240mN 325071mE, Brokaw 368 (ACU, WS); CA State Hwy 120 NE of Crater Mountain, UTM 11S 4197256mN 32505IlmE, Brokaw 519 (WS); on North Crater S of Mono Lake, UTM 11S 4199798mN 319945mE, Brokaw 520 (WS); in the East Craters Sand Flat, S of Crater Mountain and E of Punch Bowl, 37.82022°N, 119.00142°W, Brokaw 547 (ACU, WS); along CA State Hwy 120 W of its junction with Sage Hen Meadows Rd, 37.89023°N, 118.86738°W, Brokaw 554 (ACU, WS); at the junction of U.S. Hwy 395 and CA State Hwy 120, 37.88666°N, 119.09028°W, Brokaw 558 (ACU, WS); E of U.S. Hwy 395 S of its junction with West Portal Rd, 37.86439°N, 119.08492°W, Brokaw 559 (ACU, WS); along CA State Hwy 120 SW of Granite Mountain, 37.89059°N, 118.78930°W, Brokaw 560 (ACU, WS); along CA State Hwy 120 E of Granite Mountain near its junction with Dobie Meadows Rd (Rd 3027), 37.92131°N, 118.70541°W, Brokaw 56/1 (ACU, WS); along CA State Hwy 120 E of Granite Mountain N of Indian Meadows, 37.91792°N, 118.71134°W, Brokaw 562 (ACU, WS); E of Lake Crowley along Benton Crossing Rd N of Round Moun- tain, 37.63795°N, 118.63873°W, Brokaw 566 (ACU, WS); along West Portal Rd E of USS. Hwy 395, 37.84638°N, 119.06704°W, Brokaw 571 (WS); along West Portal Rd E of U.S. Hwy 395, 37.84789°N, 119.06490°W, Brokaw 572 (ACU, WS); along CA State Hwy 167 at Wilson Creek, 2n = 54 (Zavortink 60 MADRONO [Vol. 58 PIG..2. M. montana. Scale bars = 100 um. 38.05379°N, 119.12708°W, Brokaw 573 (ACU, WS). DISCUSSION Recent phylogenetic analyses suggest that M. monoensis has a unique allopolyploid origin (Brokaw and Hufford 2010b). Mentzelia section Trachyphytum is composed of two major clades, ““Affines’” and “‘Trachyphyta”’ (Zavortink 1966; Brokaw and Hufford 2010a). Most polyploid species in Trachyphytum appear to be allopoly- ploids derived from hybridizations within the “Trachyphyta”’ clade (Zavortink 1966; Brokaw and Hufford 2010b). The only exceptions are M. dispersa S. Watson of the “‘Affines”’ clade, which may be an autopolyploid complex, and M. monoensis, the only species exhibiting substantial genetic signal from both major clades (Brokaw and Hufford 2010b). One hypothesis of origin consistent with the molecular data is that M™. monoensis was derived through allopolyploidiza- tion involving a diploid progenitor closely related to M. dispersa and a tetraploid progenitor closely related to M. montana (Brokaw and Hufford 2010b). Both M. dispersa and M. montana occur in the Mono Craters region. Mentzelia montana and M. congesta Torr. & A. Gray (diploid) are the only species in section Trachyphytum that have been found to co-occur with M. monoensis, though M. laevicaulis (Dou- glas ex Hook.) Torr. & A. Gray of section Bartonia Torr. & A. Gray has also been observed Scanning electron micrographs showing variation in seed coat cell shape in: A. Mentzelia monoensis and B. in proximity. Co-occurrence with M. montana is of particular concern because some populations of M. monoensis and M. montana can only be distinguished with difficulty. The bracts of M. monoensis are always entire and usually fully green, and those of M. montana are usually lobed with a white base. However, both species may exhibit entire floral bracts with whitish bases, though the white is more prominent in M. montana. In sympatric populations the grayish- green hue in leaves of M. monoensis may be also be apparent compared to the lighter green in leaves of M. montana. The two species can be most reliably distinguished when mature seeds are compared under 10-20 magnification. Mentzelia monoensis has a tan colored seed coat composed of cells that are rounded, appearing as shallow domes (Fig. 2A). In contrast, M. mon- tana has a mottled seed coat with cells that stand out as rough, pointed knobs along edges of the seed (Fig. 2B). Likewise, despite superficial similarity of the plants, the seeds of M. monoensis can be distinguished from those of M. albicaulis (Dou- glas ex Hook.) Douglas ex Torr. & A. Gray (octoploid), which also have a mottled seed coat but have cells that project from the surface to an even greater extent than those of M. montana, giving a distinctly rough appearance to the seeds. The bracts of M. albicaulis exhibit approximately the same range of form and color found in M. monoensis. However, M. albicaulis is distinct from both M. monoensis and M. montana in its BROKAW AND HUFFORD: NEW MENTZELIA FROM MONO COUNTY 61 FIG. 3. fruit shapes and distribution. Unlike the short, erect fruits of M. monoensis and M. montana, mature specimens of M. albicaulis usually have at least some long, recurved fruits greater than 15 mm and curved between 90° and 180°. In Mono Co. both M. monoensis and M. montana occur above 2000 m, and M. albicaulis occurs below 2000 m. Like M. monoensis, M. nitens Greene has entire, green bracts and shallowly domed seed coat cells. However, like M. albicaulis, it has long, curved fruits and only occurs below 2000 m. Mentzelia nitens is distinguished from other Habit of Mentzelia monoensis. East Craters Sand Flat, Mono Co., California. species from section Trachyphytum in Mono Co. by its large petals (= 8 mm). The only hexaploid species of Mentzelia in northeastern California other than M. monoensis is M. veatchiana Kellogg. Mentzelia veatchiana 1s more robust than M. monoensis with larger flowers (styles > 3.5 mm) and lobed bracts. Further, many populations of M. veatchiana exhibit orange petals with red bases, unlike M. monoensis and all other species in Trachyphytum that occur in eastern California. Despite the difficulty of identifying some populations of M. monoensis, this species is an 62 MADRONO important endemic component of the Mono Craters flora. Although often difficult to distin- guish morphologically, unrecognized polyploid species may result in substantial underestimates of diversity (Soltis et al. 2007). For example, recently published floras have lumped diploid, tetraploid, hexaploid, and octoploid cytotypes as the single taxon, M. albicaulis (Holmgren et al. 2005), and no morphological characters to distinguish diploid, tetraploid, and octoploid cytotypes of M. dispersa have been identified (Brokaw and Hufford 2010a). The occurrence M. monoensis in a region largely composed of volcanic substrates suggests [Vol. 58 that the establishment of populations after polyploidization may have been associated with new edaphic specializations. Mentzelia monoensis has been found in habitats ranging from pumice soils and gravels in open barrens, Purshia scrub, and pine forests (Fig. 3) to disturbed sites along roadsides. Similar to other species in Mentzelia (Prigge 1986), M. monoensis appears to avoid competition in productive communities through colonization of disturbed and/or stressful habi- tats. Further investigation of possible substrate specificity is needed in order to better understand the function of M. monoensis in the Mono Craters plant communities. Identification Key for Mentzelia Section Trachyphytum in Mono County, California 1. Seeds in one row above mid-fruit, all trigonal prisms with grooves along longitudinal edges . ...M. dispersa 1’ Seeds in more than one row above mid-fruit; seeds above mid-fruit irregular-rounded to -angular, seeds below mid-fruit occasionally trigonal prisms with grooves along longitudinal edges 2. Floral bracts usually entire, green only 3. Petals greater than or equalto8mm..... 3’ Petals less than 8 mm i Nt a Hees Gee a ee She Te ee ee M. nitens 4. Longest mature fruits usually greater than 15 mm, curved less than 180°; seeds tan and moderate- to dense-mottled brown to black in age; seed coat cells pointed or domed, in age greater than 1/2 tall as wide on seed surface edges; below 2000 m...... M. albicaulis (in part) 4’ Longest mature fruits less than 15 mm, curved less than 20°; seeds tan, not or sparse-mottled in age; seed coat cells flat-surfaced to domed, in age less than 1/2 tall as wide on seed surface edges: 2000-2500 Mm. . ea ews ees bas Sema ara nets aeeea th os minke Guineas M. monoensis (in part) 2’ Floral bracts toothed to lobed or floral bracts entire with white base and green margin 5. Floral bracts conspicuous, concealing fruits, mostly white with green fringe......... M. congesta 5’ Floral bracts not conspicuous, not concealing fruits, entirely green or mostly green with white below middle only 6. Petals with orange to yellow apex and red to orange base; styles greater than or equal to 3.5 mm; longest mature fruits usually greater than 15 mm.................. M. veatchiana 6’ Petals with yellow apex and orange base; styles less than 3.5 mm; longest mature fruits usually greater than 10 mm 7. Floral bracts 3-lobed to entire; longest mature fruits usually greater than 15 mm, usually curved less than 180°; below 2000 m Clete a eh cae Creede oer ors he M. albicaulis (in part) 7’ Floral bracts 7-lobed to entire; longest mature fruits usually less than 15 mm, curved less than 45°; above 2000 m 8. Floral bracts 7-lobed to entire, white base usually conspicuous; seeds tan and moderate- to dense-mottled brown to black in age; seed coat cells pointed, in age some greater than 1/2 tall as wide on seed surface edges; above 2000 m. . . M. montana 8’ Floral bracts entire, white base inconspicuous; seeds tan, not or sparse-mottled in age; seed coat cells flat-surfaced to domed, in age less than 1/2 tall as wide on seed surface edges; 2000-2500 m ... ACKNOWLEDGMENTS We thank B. Prigge for assistance and insights into the diversity of Mentzelia section Trachyphytum. Funding for this project was provided by the Betty W. Higinbotham Trust, the Hardman Native Plant Award in Botany, and the California Native Plant Society. We thank the following herbaria for access to specimens used in this study: ACU, LA, NY, RSA, SD, UC, UCR, US, and WS. Scanning electron micro- graphs were imaged at the Franceschi Electron Micros- copy Center at Washington State University. LITERATURE CITED BROKAW, J. M. AND L. HUFFORD. 2010a. Phylogeny, introgression, and character evolution of diploid species in Mentzelia section Trachyphytum (Loasa- ceae). Systematic Botany 35:601—617. ee ee ee ea ee M. monoensis (in part) AND . 2010b. Origins and introgression of polyploid species in Mentzelia section Trachy- phytum (Losaceae). American Journal of Botany 97:1457-1473. DARLINGTON, J. 1934. A monograph of the genus Mentzelia. Annals of the Missouri Botanical Garden 21:103—227. HOLMGREN, N. H., P. K. HOLMGREN, AND A. CRONQUIST. 2005. Loasaceae, the Loasa family. Pp. 81-118 in N. H. Holmgren, P. K. Holmgren, and A. Cronquist, (eds.), Intermountain flora 2, part B. New York Botanical Garden, New York, NY. HUFFORD, L., M. M. MCMAHON, A. M. SHERWOOD, G. REEVES, AND M. W. CHASE. 2003. The major clades of Loasaceae: phylogenetic analysis using the plastid matK and ¢trnL-trnF regions. American Journal of Botany 90:1215—1228. 2011] BROKAW AND HUFFORD: NEW MENTZELIA FROM MONO COUNTY 63 PRIGGE, B. A. 1986. New species of Mentzelia have we grossly underestimated the number of (Loasaceae) from Grand County, Utah. Great species. Taxon 56:13—30. Basin Naturalist 46:361—36S. ZAVORTINK, J. E. 1966. A _ revision of Mentzelia, SoLTIS, D. E., P. S. SOLTIS, D. W. SCHEMSKE, J. F. section Trachyphytum (Loasaceae). Ph.D. dis- HANCOCK, J. N. THOMPSON, B. C. HUSBAND, AND sertation, University of California, Los Angeles, W.S. JUDD. 2007. Autopolyploidy in angiosperms: CA. MADRONO, Vol. 58, No. 1, pp. 64-65, 2011 NOTEWORTHY COLLECTION ARIZONA ARTEMISIA PYGMAEA A. Gray (ASTERACEAE).— Mohave Co., flatland, E of road to Antelope Valley, 5000 ft, T40ON R4W S36, 10 Sept 1977, S. Clark 3700 (BLM Arizona Strip District Herbarium, St. George, UT); ca 10 mi SW of Fredonia and 4 mi S of Hwy 389 along the Mt Trumbull Rd, small shrubs local on white Moenkopi badland exposure, growing with Artemisia tridentata and Eriogonum corymbosum, T40N R4 W S36 NENW, 22 Sept 1999, John L. Anderson 99-30 (ASU); Mt. Trumbull Loop, 350731 4077310N, Clifton 41711 (Clifton private herbarium). Apache Co., Red Valley, ca 3 mi NW of jet. between Navajo Rtes 12 and 134, fine red clay soils of Chinle formation, occasional with Artemisia bigelovii, Yucca angustissima, Sitanion hystrix, Juniperus monosperma, 12S 674218 3988143N, 2206 m (7280 ft), 12 June 2003, Roth 1600 (ASC, SJC). Previous knowledge. Artemisia pygmaea extends from northern Nevada, Utah and Colorado south to northern New Mexico and northern Arizona. In Arizona it has only been known from one locality south of Fredonia where it has been collected several times since 1945 (Darrow 3006 [ARIZ]). Significance. The above collections document the second and third localities of Artemisia pygmaea in Arizona. Anderson 99-30 is only ca 16 km (10 mi) from the previous known Arizona locality, but Roth 1600 1s ca 320 km (200 mi) east, near the New Mexico state line. Its importance is as an indicator of rare plant habitat since **...1t occurs in peculiar edaphic habitats...where it is often a component of communities that support rare plant species” (Welsh et al. 2003). At the Fredonia locality Artemisia pygmaea occurs with Pediocactus sileri Gray, a Moenkopi Formation endemic. BURSERA MICROPHYLLA A. Gray (BURSERA- CEAE).—La Paz Co., Harquahala Mts., foothills on the S side, ca 4 mi NE of Salome Hwy. below Socorro Peak, small shrubs (< | m tall) growing on steep south- facing hillsides of Paleozoic gray limestone with Parkinsonia microphylla, Carnegia gigantea, Fouquieria splendens, Ferocactus cylindraceus, Opuntia bigelovii, Encelia farinosa, Hyptis emoryi, TSN R11W S30 center, 2100 ft (640 m) 19 Mar 2001, John Anderson 2001-22 and Leanna Anderson (ASU, ARIZ). Previous knowledge. Bursera microphylla (elephant tree, torote blanco, copal) ranges throughout the Sonoran Desert from western Sonora and southern Baja California north to southern Arizona and disjunct in California in the Anza-Borrego Desert (Kearney and Peebles 1960; Felger 2000). In Arizona Bursera microphylla is known from approximately fifteen desert mountain ranges primarily just north of the Mexican border. Significance. The Harquahala Mountains location represents the northernmost known occurrence of Bursera microphylla. This site is 120 km (75 mi) north of the nearest occurrence to the south in the Mohawk Mountains (Sal/ywon 547, 551 [ASU]), east of Yuma, Arizona, and 80 km (50 mi) northwest of the White Tanks Mountains (Keil 4012, 5943, 6191 [ASU]), west of Phoenix, Arizona. The Harquahala Mountains Bursera microphylla plants are dwarf shrubs due to the harshness of the habitat at this northernmost location, but in the main part of their range in Mexico they are trees 2—6 m tall. FUIRENA SIMPLEX Vahl (CYPERACEAE).—La Paz Co., Grapevine Springs on S side of the Santa Maria River, first spring E of Mine Spring, locally common at spring, with Prosopis velutina, Salix gooddingii, Bac- charis sergiloides, Vitis arizonica, surrounding hills are Sonoran Desert, T11N R11 S21 SENE, 1400 ft (425 m), 18 Sept 1997, John L. Anderson 97-27 (ASU). Previous knowledge. Fuirena simplex is widespread from the midwest (Kansas, Missouri, and Illinois) south through Mexico and the West Indies to northern South America (Kral 2003). Imdorf (1994) documented the first record of Fuirena simplex in Arizona. There it occurs at a spring in oak-juniper woodland at 4800 ft (1450 m) in the Sierra Ancha Mountains of east central Arizona. Significance. The La Paz Co. record documents a second locality in Arizona and is 240 km (150 mi) west of the previous Arizona record. The Grapevine Springs site is also over 1000 m lower in elevation than the Sierra Ancha Mountains location. Its adjacent vegetative community, Sonoran Desertscrub, is very different from the Madrean Evergreen Woodland at the Sierra Anchas locality. The Grapevine Springs locale is also dissimilar from the usual habitats of Fuirena simplex described as ‘low open woods, savannas and prairies” (Kral 2003). PHOLISTOMA MEMBRANACEUM (Benth.) Constance (HY DROPHYLLACEAE).—Mohave Co., in large wash about 0.1 min of junction of Boneli Landing Rd 74 and Temple Bar Rd, 550 m, 24 March 2001, Katherine Birgy s.n., Seth Thompson and Elizabeth Powell (ARIZ, UNLV); Wilson Ridge, W side, canyon N of LMNRA Rd 64 (Boundary Mine Rd), Mohave Desert, sandy wash bottom along base of narrow canyon, with Keckiella antirrhinoides ssp. microphylla, Viguiera deltoidea, Salazaria mexicana, Brickellia cali- fornica, Bebbia juncea, Penstemon bicolor, T30N R22W S26E, 3200 ft (970 m), 2 Apr 2003, John L. Anderson 2003-17 (ASU, ARIZ); Petroglyph Wash, 721059 3993362N, Clifton 43456 (Clifton private herbarium). Previous knowledge. Pholistoma membranaceum 1s widespread through the southern two thirds of Califor- nia from the coast, foothills, and desert in a variety of habitats below 4750 m. It also extends into Baja California. Significance. The above collections, all from the Wilson Ridge area, represent first records for Arizona of Pholistima membranaceum. Wilson Ridge, part of the Black Mountains, is directly east of the Colorado River and Nevada. PULICARIA PALUDOSA Link (ASTERACEAE).— Yuma Co., Mitry Lake State Wildlife Area, 0.5 mi E of AZ/CA stateline and 6.2 mi W of Hwy 95 on Imperial Dam Rd, roadside in damp soil with Phragmites, Typha, Pluchea, Polypogon, T6S R21W S3INW, 11S 0737901 3639858, 176 ft (53 m),15 July 2010, John L. Anderson 2010-14 (ASU). La Paz Co., Cibola National Wildlife Refuge, Island Unit (Unit 3), 0.5 mi W of Colorado River on Island Road, edge of flooded field with Cynodon dactylon, Typha, Salix gooddingii, Prosopis, 11S 0715264 3687034, 200 ft (60 m), 30 Sept 2010, John L. Anderson 2010-26 (ASU); Parker, east bank of Colorado River just 2011] downstream from Hwy 62 Bridge between Parker, AZ, and Earp, CA, densely vegetated mudflat with Echino- chloa lemmonii, Cynodon dactylon, Arundo donax, Typha, Pluchea purpurascens, Baccharis salicifolia, \1S 0749065 3782807, 350 ft (105 m), 30 Sept 2010, John L. Anderson 2010-27 (ASU). Previous knowledge. Pulicaria paludosa 1s a native of the Mediterranean region of Portugal and Spain, reflected in its common name, Spanish false fleabane. It is introduced in California (Preston 2006) where it was first collected in 1946 (Orange Co., Rancho Santa Ana, Santa Ana River Canyon, moist sandy bank, Munz 11554 [RSA], Preston 2006) and first documented in 1963 Raven (1963). In California, Pulicaria paludosa is primarily known from coastal southern California with a Mediterranean climate similar to its native habitat; it also occurs in the Palm Springs area and all along the Colorado River adjacent to Arizona: Squaw Lake, Imperial Co., Be// 230 (UCR); Blythe, Riverside Co., Ballmer s.n. (UCR); Earp, San Bernardino Co., McGaugh s.n. (UCR); Whipple Mts, San Bernardino Co., DeGroot et al 3348, 4367, 4382 (RSA). The collection McLaughlin 4318 has a label location of “Sand island in Colorado River, near outlet of Taylor Lake” but with different counties on different dupli- cates at different herbaria: Imperial/Yuma Cos. (RSA), Imperial Co. (UCR), and Yuma Co. (ARIZ). S. McLaughlin (Univ. of Arizona, personal communica- tion) stated that the collection McLaughlin 4318 was from the California side of the Colorado River. Significance. The above collections extend the range of Pulicaria paludosa east across the Colorado River from California into Arizona and represent the first records for Arizona. Though, as noted above, Anderson 2010-26 was actually collected west of the Colorado River but still in Arizona where the CA/AZ state line follows an old meander which is west of the present course of the Colorado River. PURSHIA GLANDULOSA Curran (ROSACEAE).—Mo- have Co., near Whitney Pass, gravelly, sandy loam, locally common, with Yucca brevifolia, Y. baccata, Thamnosma, Encelia, Hymenoclea, 3925 ft (1190 m), 22 Apr 1980, Ralph Gierisch 4714 (ARIZ, ASC, ASU); Black Mountains, ca 7 air mi N of Union Pass, near radio facility and ca 2 mi N of Air Ranch, small shrubs growing on light-colored volcanic tuff (and extending onto adjacent rhyolitic hillsides), associated species include Juniperus californica, Coleogyne ramosissima, Ericameria linearifolia, Salazaria mexicana, Salvia dorii, Cylindropuntia acanthocarpa, T 22N R20W S2 SWSW, 4100 ft (1242 m), 26 Apr 1994, John L. Anderson 94-5 (ASU). Previous knowledge. Purshia glandulosa is not includ- ed in Arizona botanical references (Kearney and Peebles 1960; Shreve and Wiggins 1964; McDougal 1973; Lehr 1978; Benson and Darrow 1981) as part of the Arizona flora. Several floras of adjacent states do include Arizona in its range (Munz 1973; Hickman 1993; Cronquist et al. 1997; Welsh et al. 2003). Significance. The above collections document the occurrence of Purshia glandulosa in Arizona. Mature fruits are needed in collections to make a positive identification of Purshia glandulosa because it resembles P. stansburiana vegetatively. Early season collections from other localities farther east in Mohave Co., Arizona, Quail Canyon (21 Apr 2000 Higgins (NY Accession Number 848039) and Cedar Pockets (2 May 2000 Higgins (NY Accession Number 848038), were NOTEWORTHY COLLECTION 65 identified as Purshia glandulosa. The author visited these localities on Aug 30 and Sept 1, 2010, respectively, and found the Purshia plants present to be P. stansburiana, having the multiple plumose-tailed achenes of P. stansburiana, not the single non-plumose achene of P. glandulosa. In Arizona Purshia glandulosa is a peripheral species of limited distribution, present only in the westernmost mountains in Mohave Co, Arizona, the Black Mountains (Anderson 94-5) and the Virgin Mountains (Gierisch 4714), adjacent to Nevada; nonetheless, it is a welcome addition to the Arizona flora. As Lester Rowntree (1939) said, “‘Although there is a great deal of Purshia glandulosa growing with Desert Artemisia on the mountains slopes bordering the desert, it seems never to be tiresome.”’ —JOHN L. ANDERSON, Bureau of Land Manage- ment, 21605 N. 7 Avenue, Phoenix, AZ 85027. jlanders@blm.gov. LITERATURE CITED BENSON, L. AND R. DARROW. 1981. Trees and shrubs of the southwestern deserts, 3rd ed. University of Arizona Press, Tucson, AZ. CRONQUIST, A., N. H. HOLMGREN, AND P. K. HOLMGREN. 1997. Intermountain Flora, Vol. 3, Part A. The New York Botanical Garden, Bronx, NY. FELGER, R. S. 2000. Flora of the Gran Desierto and Rio Colorado of Northwestern Mexico. University of Arizona Press, Tucson, AZ. HICKMAN, J. C. (ed.). 1993. The Jepson manual: higher plants of California. University of California Press, Berkeley, CA. IMDoRF, G 1994. Noteworthy collections: Fuirena simplex. Madrono 41:330. KRAL, R. 2003. Fuirena. Pp. 32—37 in Flora of North America Editorial Committee (eds.). 2003, Flora of North America North of Mexico, Vol. 23. Oxford University Press, New York, NY. KEARNEY, T. H. AND R. H. PEEBLES. 1960. Arizona flora, 2nd ed., with supplement, by J. T. Howell and E. McClintock. University of California Press, Berkeley, CA. LEHR, J. H. 1978. Catalogue of the flora of Arizona. Desert Botanical Garden, Phoenix, AZ. McDOUGAL, W. B. 1973. Seed plants of northern Arizona. Museum of Northern Arizona, Flagstaff, AZ. MuNnNz, P. A. 1973. Flora of California with supple- ment. University of California Press, Berkeley, CA. PRESTON, R. E. 2006. Pulicaria. Pp. 471— in Flora of North America Editorial Committee (eds.). 2006, Flora of North America North of Mexico, Vol. 19. Oxford University Press, New York, NY. RAVEN, P. H. 1963. Pulicaria hispanica (Compositae: Inuleae), a weed new to California. Aliso 2251-253: ROWNTREE, L. 1939. Flowering shrubs of California. Stanford University Press, Stanford, CA. SHREVE, F. AND I. WIGGINS. 1964. Vegetation and flora of the Sonoran Desert. Stanford University Press, Stanford, CA. WELSH, S. L., N. D. ATwoop, S. GOODRICH, and L. C. HIGGINS. (eds.). 2003. A Utah flora, 3rd ed., revised. Brigham Young University Press, Provo, UT. MADRONO, Vol. 58, No. 1, p. 66, 2011 NOTEWORTHY COLLECTION CALIFORNIA POLEMONIUM CARNEUM A. Gray (POLEMO- NIACEAE).—Siskiyou Co., Klamath National Forest, W of Yreka, uphill from USFS road 44N33, 1.1 air mi SE of Deadwood historical marker, 41°42.217'N, 122°47.645’'W, 1332 m. Indian Creek Baldy quadrangle, T45N, RO8W, Sec. 32, plants growing along a seasonal watercourse on a north-facing slope, with Pseudotsuga menziesii, Isatis tinctoria, Berberis aquifolium var. repens, Amelanchier sp., Symphoricarpos sp., Agastache sp., Prunus virginiana, Ceanothus integerrimus, Lathyrus sulphureus, the plants extended up the creek for 8 m, plants were ascending with multiple stems arising close together and almost a half meter tall, 25 June 2010, Rebecca Stubbs 10 (SFSU, CAS). A few additional plants were found extending E from the ravine, and there was a single plant growing at a lower switchback south of the larger population. D. Reed originally collected from this location in 1982 and deposited the specimen in the Scott River Ranger District of Klamath National Forest (CNDDB 2010). Previous knowledge. Polemonium carneum had been found in the late 1800’s and early 1900’s as far south as the San Francisco Bay area. In the Consortium of California Herbaria the most recent collection was from 1950 with no other collections being deposited in the last six decades. The majority of the collections were made coastward. Polemonium carneum is common in Oregon and state-threatened in Washington. Significance. Second report since 1950, the first by Reed in 1982 (CNDDB 2010). There was concern that Pole- monium carneum had been extirpated from California. —REBECCA STUBBS, Department of Biology, San Francisco State University, 1600 Holloway Avenue, San Francisco, CA 9413; ROBIN FALLSCHEER, Depart- ment of Fish and Game, 601 Locust Street, Redding, CA 96001. stubbsrl1@gmail.com. LITERATURE CITED CALIFORNIA NATURAL’ DIVERSITY DATABASE (CNDDB). 2010. Polemonium carneum. California Department of Fish and Game, Sacramento, CA. Website http://www.dfg.ca.gov/biogeodata/cnddb/ [accessed 01 June 2010]. MADRONO, Vol. 58, No. 1, p. 67, 2011 IN MEMORIAL ISABELLE I. TAVARES 1921-2011 Dr. Isabelle Tavares died on May 21, 2011. Dr. Tavares had been associated with the University Herbarium for 59 years. Her academic career began at City College of Los Angeles, was interrupted by service in the Women’s Army Corps during World War II, and continued at Berkeley after the War. She received B.A., M.A., and Ph.D. degrees from the University of California. For her doctoral dissertation, con- ducted under the guidance of Professor Lee Bonar, she investigated the Laboulbeniales, an order of minute fungi that parasitize insects. This research continued after her dissertation and eventually resulted in her magnum opus: The Laboulbeniales, published in 1985. She began working in the University Herbarium while still in graduate school, and continued uninterrupt- edly until long past her retirement in 1993. She curated the fungi (including lichens) and the bryophytes, and participated in all day-to-day operations of the Herbarium. Curating lichens led her to her second major research interest: the taxonomy of Usnea, a widespread and notori- ously taxonomically difficult genus of lichens. Dr. Tavares was a founding member of the California Lichen Society, and an active promoter of California lichenology. Her involvement with the California Botanical Society was extensive, and she made major editorial contributions to Madrofio over a period of many years. She was President of the Society from 1983—1984. —Richard L. Moe, Collections Manager, University and Jepson Herbaria, University of California, Berkeley Images used with permission, University & Jepson Herbaria Archives, University of California, Berkeley, CA. Volume 58, Number 1, pages 1—68, published 31 August 2011 SUBSCRIPTIONS — MEMBERSHIP The California Botanical Society has several membership types (individuals ($35 per year; family $40 per year; emeritus $27 per year; students $27 per year for a maximum of 7 years). Late fees may be assessed. Beginning in 2011, rates will increase by $5 for all membership types except life memberships, for which rates will increase by $100, and student memberships, which will not show a rate increase. Members of the Society receive MADRONO free. Institutional subscriptions to MADRONO are available ($70). Membership is based on a calendar year only. Life memberships are $750. Applications for membership (including dues), orders for subscriptions, and renewal payments should be sent to the Membership Chair. Requests and rates for back issues, changes of address, and undelivered copies of MADRONO should be sent to the Corresponding Secretary. INFORMATION FOR CONTRIBUTORS Submit manuscripts to http://www.edmgr.com/madrono/. If you are submitting a manuscript for the first time to the MADRONO PeerTrack website, you must first register by clicking "register now" and following the instructions. Manuscripts by authors having outstanding page charges will not be sent for review. Manuscripts may be submitted in English or Spanish. English-language manuscripts dealing with taxa or topics of Latin America and Spanish-language manuscripts must have a Spanish RESUMEN and an English ABSTRACT. For all articles and short items (NOTES, NOTEWORTHY COLLECTIONS, POINTS OF VIEW, etc.), follow the format used in recent issues for the type of item submitted. Detailed information for contributors, including instructions for manuscript preparation, is available at the California Botanical Society website, www.calbotsoc. org; follow the link for MADRONoO. Authors are encouraged to include the names, addresses, and e-mail addresses of two to four potential reviewers with their submitted manuscript. All California Botanical Society members current in the volume year that their contributions are published are allowed five free pages per volume year. Additional pages will be charged at the rate of $40 per page. Joint authors may apply all or a portion of their respective five-page allotments to a jointly-published article. Partial pages will be charged as full. The purpose of this fee is not to pay directly for the costs of publishing any particular paper, but rather to allow the Society to continue publishing MADRONO on a reasonable schedule, with equity among all members for access to its pages. Printer’s fees for color plates and other complex matter (including illustrations, charts, maps, photographs) will be charged at cost. Author’s changes after typesetting @ $4.50 per line will be charged to authors. Page charges are important in maintaining Madrono as a viable publication, and timely payment of charges is appreciated. At the time of submission, authors must provide information describing the extent to which data in the manuscript have been used in other papers that are published, in press, submitted, or soon to be submitted elsewhere. SMITHSONIAN INSTITUTION LIBRARIES WNL 455