AN ABSTRACT OF THE THESIS OF

Jack Revnold Carison for the degree of Master of Science in Crop Science presented on March 2O 1986

TITLE: . !st St f p soicata (Pursh) A. L&ve (Poaceae: TriticeaeJ

Abstract Approved: Signature redacted for privacy. Robert J. Metzger

Chromosome counts were determined for 152 accessions of Pseudo- roegneria spicata (Pursh) A. L6ve and, combined with existing count data, used to plot the distribution of diploid and tetraploid popula- tions. Morphological variation of 55 characters was examined in five groups totaling 205 operational taxonomic units (OTU's), using cluster, principal factor, and discriminant analyses. The five groups included diploid and autotetraploid spicata, an allotetraploid pre- viously included in spicata, a control group including four Old World Pseudoroegneria species, and a small control sample of Elymus lanceolatus (Scrib. and Smith) Gould. The analyses were not able to separate diploid from autoploid spicata nor identify any clear sub- groupings within the dip].ojds. However, the alloploid was separated from spicata and aligned with Elymus lanceolatus based on glume and spike characters. This study recommends the alloploid be included in lanceolatus as a new subspecies, Elymus lanceolatus ssp. wawawai. The chromosome count data indicate it is distributed in the canyons and tributaries of the lower Salmon and Snake Rivers of northern Idaho, northeastern Oregon, and southeastern Washington. The new subspecies keys to Elymus lanceolatus based on glume characters and is separated from subspecies .ianceolatus and albicans by its cespitose growth habit. A S11LPY OF MORPHOLOGICAL VARIATION WITHIN PSELVOROEGNERIA SPICATA (PURSH) A. LOVE (POACEAE: TRITWEAE)

By

Jack Reynold Carison

A THESIS

Submitted to

Oregon State University

in partial fulfillment of the requirements for the degree of

Master of Science

Completed March 20, 1986

Commencement June, 1986 APPROVED:

1' Signature redacted for privacy. Professor of Crop Science in charge of major

Signature redacted for privacy. - Read of Department of Crop Science

Signature redacted for privacy.

Dean of Graduate hoo 1

Date thesis is presented

Typed by Marguerita Balint-Carden for JackR.Carison ACKNOWLEDGEMENTS

I would like to extend my appreciation to my Major Professor and good friend, Dr. Robert J. Metzger, for his excellent guidance throughout the course of my graduate study. Together with Dr. Douglas R. Dewey, their expertise in cytogenetics provided an outstanding foundation for my work on this thesis problem. I am especially indebted to Dr. Dewey for his willingness to serve on my graduate committee, despite the difficulty in doing so from his position at Logan, Utah. I feel very fortunate to have received training from hint.

I thank Dr. Kenton Chambers, Dr. Warren Kronstad, and Dr. Harry Mack for accepting the membership of my graduate committee. A special thanks to Dr. Chambers for his guidance on biosystentatics and his out- standing teaching in this area. Appreciation also is extended to Dr. Mary Barkworth and Dr. Steven Broich for assistance with numerical taxonomy questions.

I thank the USDA Soil Conservation Service for the opportunity and support to complete this graduate study. Special thanks to the late Archie Fuchs and my current supervisor, Clarence Maesner, for their patience and helpfulness. I want to express my gratitude to Marguerita Balint-Carden for her excellent assistance with word pro- cessing in the development of this manuscript.

I want to acknowledge the friendship and collaboration of Dr. Luiz Gonzaga E. Vieira, my former fellow graduate student in plant gene- tics. Luiz is a wizard in the laboratory, likes to argue passionately about scientific issues (among others), and is very open and helpful to others. He created a day-to-day learning environment that was a tremendous benefit to me. I shall miss this interaction. Finally, I feel very fortunate in the love, encouragement, and support from my wife, Vicki, and children, Chris and Sara, while completing this degree. TABLE OF CONTENTS

Pace

INTRODUCTION 1

MATERIALS AND METHODS 5

RESULTS 10

DISCUSSION 21

CONCLUSIONS 24

BIBLIOGRAPHY 46

APPENDICES 49 LIST OF FIGURES

Yiqure Pane

1 Distribution of Pseudoroegneria spicata in 27 western North America, including the alloploid form.

2 Condensed dendogram obtained from average linkage 28 cluster analysis of 205 OTU's of Pseudoroegneria spicata and Elymus.

3 Scatter plot of factor scores for spike/spikelet 29 and awn characters in principal factor analysis of 205 Pseudoroegneria spicata OTU's.

4 Scatter plot of factor scores for spike/spikelet 30 and awn characters in principal factor analysis of 165 Pseudoroegneria spicata OTU's.

5 Scatter plot of factor scores for spike/spikelet 31 characters and leaf/cuim lengths in principal factor analysis of 165 Pseudoroegneria spicata OTUS.

6 Scatter plot of factor scores for spike/spikelet 32 characters and leaf/cuim widths in principal factor analysis of 165 Pseudoroegneria spicata OTU's.

7 Scatter plot of factor scores for leaf/cuim widths 33 and awn characters in principal factor analysis of 165 Pseudoroegneria spicata OTU's.

8 Scatter plot of canonical variable scores for five 34 groups of Pseudoroegneria and Elymus.

9 Scatter plot of canonical variable scores for three 35 groups of Pseudoroegneria spicata, including the a].loploid form. LIST OF TABLES

Table Paae

1 Species of the genus Pseudoroegneria (Nevski) 36 L6ve.

2 Characters used to score OTU's of five groups of 38 Pseudoroegneria and Elymus.

3 Morphological characters selected for use in 40 cluster analysis of Pseudoroegneria and Elymus OTU' s.

4 Group frequency in five major clusters formed by 41 analysis of 17 morphological characters of 205 OTU's of Pseudoroegneria and Elymus.

5 Variance explained by factors identified in 42 principal factor analysis of three combinations of morphological characteristics measured on 205 OTU's of Pseudoroegneria and Elymus.

6 Rotated estimated factor loadings in principle 43 factor analysis of 165 OTU's of Pseudoroegneria.

7 Variables selected by stepwise discriminant 44 analysis and classification functions for groups of Pseudoroegneria and Elymus OTU's.

8 Jackknifed classification of OTU's into five and 45 three groups of Pseudoroegneria and Elymus by discriminant analysis. LIST OF APPENDIX TABLES

Table Pane

1 Collection information for 205 OTU's of 50 Pseudoroegneria and Elymus.

2 Environmental conditions at four uniform 56 garden nursery sites which were sources of 192 Pseudoroegneria and Elymus OTU's.

3 Source of known diploid, triploid, auto- 57 tetraploid, and allotetraploid forms of Pseudorogenria spicata.

4 Means and standard deviations of 55 morphological 65 characters for five groups of Pseudoroegneria and Elymus. A STUDY OF MORPHOLOGICAL VARIATIONWITHIN PSEUDOROEGNERIASPICATA (PURSHJ A. LOVE (POACEAE: TRITICEAE)

INTRODUCTION

The genus Pseudoroegneria (Nevski) L5ve, was recently con- structed (L6ve 1982) to accommodate all S genome species of the Triti- ceae tribe of grasses. Several members of this new genus came from section Pseudoroegneria in the genus Elytrigia, established by Nevski (1934), hence the new generic name. The new genus includes one North American species, still commonly known as Agropyron spicatum (Pursh) Scrib. and Smith, after the treatment by Hitchcock (1951). The name Pseudoroegneria spicata (Pursh) Lve, will be used for this species, which is the focus of this study, as proposed in the new nomenclature for North American Triticeae by Barkworth et al. (1983) and Barkworth and Dewey (1985). This new treatment of North American Triticeae is based on the genomic system of classification advocated by Dewey (1982, 1983) and L6ve (1982, 1984). That is, different genomes and combinations of genomes determine generic limits and pro- vide a clearer understanding of biological relationships, a major objective of taxonomy.

Bluebunch wheatgrass, a widely used common name of P. spicata, is a major cool-season range grass, native to much of western North America. It is characterized by Hitchcock (1951) as cespitose with erect cuims, awned and distant spikelets, a continuous rachis, diver- gent awns, more than 7 spikelets per spike, spikes 8 to 15 cm long, and spikele'cs usually shorter than the internode. The awnless form, considered by Hitchcock (1951) as the separate species Agropyron inerme (Scrib. and Smith) Rydb., differs from P. spicata only in the awnless spikelets and is not considered to have much taxonomic significance (Daubenmire 1939, Dewey 1982). However, it is considered a subspecies in some treatments (LBve 1984). Bluebunch wheatgrass is a major component of many plant coirimuni- ties, ranging from arid desert sagebrush(Artemisia)habitats to mesic, subhumj.d pine/fir(Pinus/Pseudotsuga)woodlands.It is often associated or sympatric with other Triticeae such as thickspike wheatgrass,Elymus lanceolatus(Scrib. and Smith) Gould; squirrel- tail,E. elymoides(Raf.) Sweezy; slender wheatgrass,E.trachy- caulus(Link.) Gould ex Shinners; western wheatgrass,Pascopyrum smithii(Rydb.) L6ve; basin wildrye,Leymus cinereus(Scrib. and Merr.,) Love; and beardless wildrye,L. triticoides(Buckley) Pilger.Natural hybrids betweenP. spicataandE. lanceolatus, and betweenspicataandE. elymoides,are conunon and are evi- dence for introgression of specific characters.For instance, spi- catapopulations found in mesic woodland habitats often are mildly rhizomatous, a characteristic perhaps contributed bylanceolatus (Dewey 1982). E. lanceolatus spp. albicans(Scrib. and Smith) Barkw. and Dewey, with a somewhat restricted range in Montana and adjoining states, is less rhizomatous and has divergent awns resemb- lingP. spicata. Hybridizingspicatawithlanceolatuspro- duces some partially fertile progeny morphologically similar to albicans(Dewey1970). Hybridization experiments also have urecreatedu the oldAgropyron saxicola,revealing it to be a natu- ral cross betweenspicataandE. elymoides,now called X Pseudelymus saxicolus(Scrib. and Smith) Barkw. and Dewey (Dewey 1964). Bluebunch wheatgrass occurs over a wide range of habitats and hybridizes with other syinpatric Trjticeae.Because it outcrosses, considerable variation within the species is expected.However, there have been few attempts to delineate subspecies and varieties.

A. S. Hitchcock(1951)and C. L. Hitchcock(1969)divide Pseudoroegneria spicatainto var.spicata,which encompasses nearly the entire species, and var.pubescens,which is restricted to the Wenatchee Mountains in central Washington, with scattered popu- lations in eastern Oregon and Idaho.The varietypubescensis characterized by pubescence on the sheaths, the lower surface of leaves, and stems.Conceivably,E. elymoideshas contributed its pubescent characteristics to these populations through introgression. 3

Whereas pubescens may be worthy of variety rank, other populations, perhaps those influenced by lanceolatus, also warrant consideration as valid taxa below the species level. There are other apparently distinct populations as well (SCS 1947), including a lower-growing, finer leafed, awniess type confined to the Palouse prairie and arid Columbia Basin in eastern Washington and northern Idaho.

Pseudoroegneria spicata is morphologically similar to Elymus arizonicus (Scribn. and Smith) Gould. In the past, some thought arizonicus to be a more southern, robust form of spicata. Recent crossing experiments, however, show arizonicus to be an allotetraploid of SI! genome constitution (Hsiao, unpublished), and therefore, it falls into Elymus. This supports Pyrah's findings (1983) demonstrating the occurance of natural hybrids between arizoriicus and E. elymoides. Likewise, recent studies (Dewey 1982) show the cv.'Secar' of spjcata and similar forms (P1285272) to be alloploid in nature. Although usually keyed morphologically to spicata, the 'Secar' form, which is tetraploid, forms sterile hybrids with known spicata autotetraploids. Cytological examina- tion of Secar-type pollen mother cells at Metaphase I show 14 biva- lents and few to no trivalents or quadrivalents. Preliminary hybridi- zation efforts with alloploid E. lanceolatus have produced some fertile F1 progeny (Dewey, unpublished). Evidence currently points to separating the 'Secar' type from spicata and including it in Elymus.

As indicated, bluebunch wheatgrass is the only S genome species in North America. L5ve (1984) recognizes 17 S genome Pseudoroeg- neria species outside North America, including P. strigosa (M. Bieb.) Löve, the type species (table 1). The taxonomic picture of the Old World species is no clearer than for spicata. For example, recent research shows some Pseudoroegneria tauri to be alloploid (Wang, personal communication). Few of the species have been cytolo- gically examined in detail after controlled crossing experiments. 4

Elymus differs from Pseudoroegneria in that Elymus are alloploids with SH genome structure, the H contributed by Hordeuni (Dewey 1982). North American H genome species now are considered to be Critesion (Barkworth and Dewey 1985). Consequently, natural hybrids frequently occur between Elymus and Pseudoroegneria because they share the S genome. Pseudoroegneria is not closely related to Leymus, which has a JN genome constitution. Hybrids between Pseudoroegneria and Leymus are not known to occur naturally. Pascopyrum, a SHJN octoploid, is the result of hybridi- zation (and doubling) between Elymus and Leymus. IntrogressiOn between Pseudoroegneria and Pascopyrum is conceivable but has not been reported.

Pseudoroegneria spicata is predominantly diploid 2N=14, but some populations are autotetraploid. As indicated, an allotetraploid also resides within the species. Occasional triploids can be found in mixed populations of diploids and tetraploids. No other ploidy levels have been reported to occur naturally.

Bluebunch wheatgrass is important to the livestock economy, wild- life management programs, and is widely used in mineland reclamation projects. Two cultivars, cv. 'Whitmar' and cv.'Secar', have been released and are commercially available. Understanding taxonomic relationships at the subspecies and variety levels, and resolving the confusion concerning the allotetraploid will assist in further improvement of the species and more effective reclamation recommenda- tions. The objectives of this study were to (1) determine which morphological characters can be used to distinquish diploids, autote- traploids, and allotetraploids; (2) establish the geographical distri- bution of the three groups; and (3) to modify morphological keys to separate the taxonomically significant groups. 5

MATERIALS AND METHODS

Seeds and live plants were obtained from a variety of sources to determine chromosome number, including:

Accessions from the ARS Regional Plant Introduction Station (RPI$) with assigned p1 numbers. Seeds came directly from the RPIS at Pullman (WA); from seed packets stored at the ARS Crops Research Laboratory at Logan (UT); or SCS Plant Materials Centers (PM Centers) at Pullman, Bridger (MT), Aberdeen (ID), and Meeker (CO).

T numbered accessions from evaluations underway at the above named SCS PM Centers.

Seeds collected by the author in 1983 and 1984 in California, Idaho, Montana, Nevada, Oregon, and Washington.

Seeds provided by the ARS Crops Research Laboratory at Logan of collections by Kay Asay in 1975 and 1981 and Al Davis in 1982.

Seeds of cv. 'Whitmar' and 'Secar' from the Pullman PM Center.

Prior to this study, chromosome numbers had been assigned to approximately 110 plant accessions, most specimens originating from British Columbia, Washington, and Idaho, (Hartung 1946, Bowden 1965, Dewey unpublished). Chromosome counts were determined for 152 addi- tional accessions including numerous specimens from Nevada, Oregon, Montana, and California.

Somatic chromosome counts were determined from root tips pre- treated in ice water for 24 hours and then fixed in 3:1 ethanol- glacial acetic acid. Root tips were prepared for examination by hydrolizing 10 minutes in iN HCL at 60°C and staining in Fuelgen. Meiotic chromosome counts were ascertained from pollen mother cells 6

(pmc's) at Metaphase I. Pmc's were teased from anthers taken from spikes at 1/10 heading stage and stained with acetocarmine.

Ploidy groups were plotted on a map of western North America and evaluated for possible relationship to available vegetative survey information.

The next phase of the study was to evaluate the morphological differences between different ploidy groups within P. spicata, and also compare them with other Pseudoroegneria and a small sampling ofElymuslanceolatus (thickspike wheatgrass). Each operational taxonomic unit (OTU) consisted of a complete, mature flowering cuim, including the crown and usually with a few roots attached. Field sampling involved taking at least three flowering cuims per plant, and at times as many as fifty, depending on availability. Sample culms were bundled, tied, and wrapped loosely in newspaper and allowed to air-dry.

Five groups were selected for evaluation, (1) diploid spicata, (2) autotetraploid spicata, (3) allotetraploid spicata,(4) Old

World Pseudoroegneria, and (5)Elymuslanceolatus. Group 1con- sisted of 81 OTU's of 33 accessions; Group 2,31 OTU's of seven acces- sions; Group 3, 53 OTU's of 12 accessions; Group 4, 26 OTU's of five species and nine accessions; and Group 5,14 OTU's of five acces- sions. The group 4 Old World species included three accessions of P. stipifolia (Czern. ex Nevski) Lve, two accessions of P. tauri (Boiss. and Bal.) L6ve, two accessions of P. tauriSSp. libanotica (Hackel) Löve, one accession of P. strigosa (Bieb.) L8ve, and one accession of p. cognata (Hackel) Lve. See appendix table 1. The taxonomy of p tauri is unsettled in that the tetra- ploid accessions evaluated in this study now appear to be alloploids (Wang, personal communication). Furthermore, the P. strigosa sampled at Logan, Utah is tetraploid and may not be typical of the species. Therefore, a subset, excluding all P.tauri and P. strigosa, was used for some of the analyses. 7

A total of 205 OTrJ's were selected from four uniform garden nursery environments (1) Utah State University (USU) Evans Farm at Logan, Utah;(2) USU Deer Pens Plots at Logan, Utah;(3) Oregon State University (OSU) Branch Experiment Station, Moro, Oregon; and (4) OSU greenhouse, Corvallis, Oregon. Site conditions are described in appendix table 2. Measurements were scored for 116 OTU's from site 1, 32 from site 2, 34 from site 3, and 23 from site 4.

A set of 55 characters was chosen for the morphological study after preliminary evaluation of all five groups and reviewing pub- lished keys (Hitchcock 1951, Hitchcock 1969, Holmgreri and Holnigren 1977). The character set included four two-state characters, four three-state characters, nine vestiture characters, six discrete quan- titative characters, and 32 continuous quantitative characters. See table 2. Emphasis was placed on characters relatively easy to measure in the field. Flowering cuims were visually selected from the tallest one-third cuims, with average culm and spike length within this top group of culms. Culni leaf measurements were made on the largest cuim leaf, which was usually two leaves below the flag leaf (sometimes involved the first or third leaf below the flag leaf). The number of spikelets per spike and florets per spikelet were the mean of three culms rounded to the closest integer.

Samples per accession varied from one to ten, with an average of 3.1 OTU's per accession. Atkins et al. (1984) demonstrated that at least 12 OTU'S were needed to encompass variation within a population in a study of two species ofLeymus. However, the OTU's in the Leymusstudy were taken from natural populations where soil and other micro-environmental conditions are highly variable compared to uniform garden nurseries. The number of OTU's required to accommodate variation within accessions in uniform garden nurseries is lower, but probably exceed the 3.1 average in this study. Accession OTU size was limited by the availability of material to sample. Emphasis was placed on sampling throughout the range of spicata and sampling from four nursery environments. 8

All measurements were made on mature, dry, flowering culms except for certain characters. Culm and flag leaf sections were moistened in a petri dish with a mixture of water, ethanol, and a small amount of laundry detergent. These moistened samples were used to record leaf width, ligule and auricle lengths, and leaf-blade rib number.

Data were compiled in a 205x55 matrix, edited, and stored on a Data General MV8000 II mini-computer at the USDA SCSI West National Technical Center (WNTC), Portland, Oregon. Summary statistics were calculated using BMDP2D (Enge].man 1982). All data were computed to z-scores (standardized) using BMDP1S (Ho 1982).

Three types of multivariate analysis were performed on the trans- formed data to indentify similarities between OTU's and important mor- phological variables for classification. Dendogratus and tree dia- grams, using unweighted pair-group mathematical average algorithims, were generated by BMDP2M cluster analysis for cases (Engelluan 1982). Four computer runs were conducted using (1) 205 OTU's and 17 variables selected after examination of preliminary summary statistics, (2) 165 OTU's (diploid/autoploid/alloploid groups) and 13 variables identified as important after discriminant analysis, (3) the same 165 OTU's and 11 vegetative plus two spike variables, and (4) the same 165 OTU's and 19 spikelet variables. These analyses grouped morphologically similar OTU's without regard for any group assignment.

Next the data was evaluated by BMDP4M Factor Analysis (Frane and Jennrich 1982), in which principal components were identified. Since few state characters, although ordered in this study, often can bias principal factor analysis (Srteath and Sokal 1973), computer runs were made with and without these characters.

Finally, stepwise discriminant analysis was performed on the transformed data using BMDP7M (Jennrich and Sampson 1982). In this analysis the JACKKNIFE option was chosen to compute Mahalanobis D2 and posterior probabilities for the distance from each case to the groups formed by the remaining cases. A jackknife classification 9 matrix was constructed from these computations. Incorrect classifica- tions were analyzed using accession record information.

In addition to discriminant analysis of the entire 205 case (5 group) data set, three subsets were also analyzed: (1) diploids/auto- tetraploids/allotetraploids, 165 cases; (2) diploids/autotetraploids, 112 cases; and (3) all groups with reduced Old World set, 194 cases. The first case was repeated using contrasts to order selection of variables for the classification function, maximizing differences between the two spicata groups and alloploid OTU's.

For discriminant analysis the data were further evaluated using a subset validation procedure. Subsets of data were randomly selected and assigned to five corresponding new groups. Classification func- tions were computed on the remaining data and used to classify the new groups. This procedure was performed three times on the 205 case data set. Variable selection was cross-checked among runs, and from the entire set of results, a list of useful taxonomic characters was chosen. 10

RESULTS

Chromosome Number

Pseudoroegneria spicata is predominantly diploid throughout its range. Chromosome counts for 152 new accessions revealed all were diploid except for seven tetraploids.

T7681 Clearwater County, Idaho T25668 Big Horn County, Wyoming T37193 Adams County, Washington (mixed with diploids) T37194 Franklin County, Washington T40385 Nez Perce County, Idaho T40583 Yakima County, Washington T40599 Asotin County, Washington

Of these accessions, only T7681 exhibited predominantly quadriva- lents at meiotic metaphase I and, therefore, was autotetraploid. The others formed 14 bivalents and, therefore, were alloploids. Appendix table 3 lists all known accessions in each ploidy group. Figure 1 plots their distribution in western North America.

Autotetraploid spicata appears to occur in scattered popula- tions in more mesic vegetative zones in eastern Washington, northern Idaho, and southern British Columbia. These populations usually are associated with ponderosa pine (Pinus) woodlands and inesic shrub (Symphoricarpus, Crataegus) grasslands. Very recent work (Dewey unpublished) shows that most spicata on Steptoe Butte, a relic basalt formation rising out of the Palouse grasslands in eastern Washington, appears to be autotetraploid. This formation receives more moisture than the surrounding area, in which primarilydiploids have been found. 11

Most of the allotetraploids appear associated with the lower drainages of the Salmon, Snake, and Columbia Rivers, especially preva- lent on the steep canyon slopes of the rivers, tributaries, and cou- lees. Populations are concentrated along the Snake River from Riggins, Idaho to its confluence with the Columbia River. Two acces- sions from Wyoming also are tetraploid and have alloploid characteris- tics, such as described later. Collins (1964) reported an allotetra- ploidspicatafrom central Montana. However, the status of the allotetraploid outside northern Idaho and eastern Washington, is uncertain. More collections are needed in the eastern part of the spicatarange. Thirty-four accessions collected in western Montana, in 1984, were all diploid including several within a 75 km radius of the Collins accession.

Morpholoqical Variation

Group means and standard deviations for the 55 characters are shown in appendix table 4. Comparing groups revealed possible differ- ences in spike, glume, and lemma characters. Both the alloploid and thickspike groups appeared to have more compact spikes and shorter, fewer-veined, narrower, sharper, and rougher glumes. The alloploid differed from the thickspike group in having a higher frequency of awns, longer awns, and less pubescence on the lemma. Although not measured, all thickspike OTU's were at least moderately rhizomatous. Another characteristic not scored was the presence of basal leaf sheath hairs (greater than 1 mm in length), which are only reliably observed in the live plant state, particularly in young seedlings of the alloploid and thickspike groups. This character is absent in spjcatabut has been observed in all live specimens examined in the alloploid and thickspike group.

Little difference was apparent between the diploid and autoploid groups for most characters. Autotetraploids seemed to be more often awned, with straighter awns, longer pubescence on adaxial leaf sur- 12 faces, and denser pubescence on the adaxial glume surface. However, these differences did not seem to be significant in this sample.

The Old World Pseudoroegneria, a highly variable group since it contained several species, generally lacked awns, and were more robust in overall growth habit.

Coefficients of variation were about 30 percent lower in the autotetraploid group, a possible indication of the buffering effects of higher ploidy. The variability within the diploid, Old World, and thickspike groups is due in large part to the diversity of types within the group. The Old World group contains several species, while the other two contain subspecies, and varieties. The alloploid group, although restricted in range and expected to be relatively uniform, probably varies due to introgression from diploid spicata.

Cluster Analysis

Based on the apparent variation among groups for the 55 charac- ters scored, a subset of 17 characters was selected for use in the cluster analysis of the 205 cases. See table 3. Characters were selected which indicated variation among groups and also were not limited to a single part of the plant anatomy. Nevertheless, spikelet characteristics, which appeared to vary the most among groups, included most of those selected.

The analysis divided the OTU's into two major clusters and seve- ral minor ones. See figure 2 reading left to right, and table 4, reading down. After two diploid accessions, a small cluster of nine OTU's was formed, all from the 14 OTU thickspike group. Of the five thickspike OTU'S not included in this cluster, two were var. albi- cans and three were artificial crosses with spicata. 13

Next came a region of 33 OTU's with several miscellaneous clus- ters of no apparent importance. Included in this region were 17 dip- bid, one autotetrapboid, eight allotetraploid, five Old World, and two thickspike OTU's. The three p. strigosa OTU's were positioned in this region, as were the two thickspikealbicans.

The next group of 44 OTU's formed a major cluster. This cluster included five diploid, 35 allotetrapboid, one Old World, and three thickspike OTU's. The cluster was 80 percent alloploid and contained 66 percent of the allopboids evaluated, evidence for the distinctness of the this type.

Following was a major cluster of 104 OTU's in which 56 diploids (69%) and 30 autotetrapboids (97%) were located. Also included were seven allotetraploids and 11 old World OTU's. The Old World OTU's were concentrated towards the right side of the cluster. Within the entire cluster, seven subclusters were apparent, but did not seem to be strongly associated with any particular ecotype, variety, or sub- species of spicata. Differences within populations or geographic areas seemed as great as differences between populations. Two pos- sible exceptions are implied by the data. First, 27 of the 31 autote- traploids clustered in the left three subclusters of 64 OTU's. Secondly, the putative awniess, fine-leafed ecotype (SCS 1947) found primarily in eastern Washington, seemed concentrated in the right four subclusters of 40 OTU's. One 22 OTU subcluster contained 12 of the 30 OTIJ's of this ecotype. However, these delineations were not clear enough in this data set to separate as distinct taxa.

The final cluster of 13 OTU's at the right side of the dendogram was composed of one diploid, three allotetrapboids, and nine old World accessions. All Old World accessions in this cluster were P. stipi- folia.

Overall, this cluster analysis (table 4) revealed two large clus- ters, one which contains 80 percent albotetrapboid and another which contains 83 percent dipboid or autotetrapboid spicata. Within the 14 largespicatacluster there is a tendency for the autotetraploids to group together and an awnless, fine-leafed diploid to cluster. The awnless type, confined primarily to eastern Washington, comprised 37 percent of the diploids. Other groupings withinspicatamay be possible but await a more complete sample of OTU's from Montana, Utah, Colorado, Oregon, and Nevada.

Three cluster analyses also were run on the smaller diploid/auto- ploid/alloploid data set, using (1) variables identified as important by discriminant analysis (discussed later) in separating the three groups, (2) vegetative and spike characters, and (3) spikelet charac- ters. In the first run, diploid and autoploidspicataseparated from the alloploid group with about the same level of distinctness as in the 205 case run with 17 variables. Seventy percent of the allo- ploids fell within four well defined clusters. Sixty-five percent of the autoploids were found in a large cluster that contained many dip- bids as well. The awnless dipboids tended to cluster together, but less clearly than in the 205 case run.

In the second three-group run featuring vegetative characters, the clusters formed did not correspond to the three cytological groups. In the third run, using only spikelet characters, most of the alloploid group segregated into two major clusters. About half of the autoploids clustered into one group, probably based on adaxial glume vestiture.

Principal ComPonent Analysis

All characters scored in this study were ordered. However, 17 were few (two to four) state characters, which tend to be over empha- sized in a principal factor analysis (Sneath and Sokal 1973). Three principal factor computer runs were made: (1) all 55 characters included, (2)11 of the less important few-state characters omitted, and (3) all few-state characters omitted. 15

In all three analyses, the first factor explaining the highest percentage of total variance was a combination of spikelet dimensions and spike compactness. This had the effect of separating the allo- ploid and thickspike groups from the other three (figure 3). The two groups had shorter, narrower, fewer veined, more acute glutues, shorter and narrower lemmas and paleas, shorter spikelets and shorter rachis internodes. The correlation matrix revealed a relatively strong association of these characters. This factor explained 21 to 30 percent of the total variance depending on the run, the larger per- centage resulting when all few-state characters were omitted.

Other important factors varied with the run. However, lemma awn length and angle rated high in all three analyses. This factor sepa- rated the relatively awniess thickspike, old World, and awnless to short-awned diploids from the other OTU's. Between eight and eleven percent of the total variance was explained by this factor.

A more complex factor involved leaf, culm, and spike length measurements. This factor was rated second in run two accounting for ten percent of the total variance. Although there undoubtedly are differences between groups, the importance of this factor results from differences in growth and vigor between locations. Plants sampled at Moro were younger, smaller, and subject to greater moisture stress; whereas, those sampled at Evans were, generally, robust and fully developed. The Deer Pens and greenhouse samples were intermediate in growth and vigor. Environmental conditions make this factor somewhat unreliable for taxonomic purposes.

Another factor associated leaf blade width and rib number with cuim diameter and spike dimensions, accounting for seven to eight per- cent of the total variance. In this factor, the Old World group seemed to have wider leaves, thicker culms, and larger spikes. It argued for the general robustness of this group. It also identified 10 to 15 fine-leafed diploid OTU'S. Moreover, greenhouse samples tended to have wider leaves. 16

The remaining six to seven factors contained more or less logical combinations of traits, but accounted for less than five percent of the total variance in each case. Table 5 summarizes the contribution of each factor for the three analyses.

In all three runs, the four most important factors did not explain more than 59 percent of the total variance. Pseudoroeg- neria and Elymus .Zanceolatus are cross-pollinating and subject to introgression from closely related species. Considerable mixing of traits occurs between spicata and lanceolatus, so introgression no doubt explains much of the variance by creating many small factor combinations. The lumping of several species into the Old World group and the variability of the small thickspike group also contributed to the complexity of the analysis.

Principal factor analysis also was performed on the smaller dip- loid/autoploid/alloploid data set, with (1) 11 few-state variables removed, and (2) all few-state variables omitted. In both runs, as above, spikelet and spike characters were the primary factor, explain- ing 25 to 29 percent of the total variance. As with the other analy- sis, leaf and cuim measurements, awn length and angle, and leaf and culm widths were the next most important factors. The first four factors accounted for 52 to 60 percent of the total variance. Rotated factor loadings are shown for run two in table 6.

Scatter plots of factor scores for the four most significant factors in the 165 OTU data set are shown in figures 4 through 7. The first factor, spikelet and spike characters, tends to separate the alloploid from the other two groups. As with the 205 OTU data set, leaf and cuim lengths do not contribute to increased separation (figure 5) and reflect the differing environmental conditions of the four locations where samples were taken. On the other hand, both awn characteristics and leaf/culm widths indicate differences within groups, particularly within diploid spicata. There is evidence for the awnless, fine-leafed diploid form in the rotated factor scores for 17 awn and width characters. Of the 19 diploid OTU's scoring high for narrow leaves, only five are awned; out of 33 awniess (or nearly awn- less) diploids, only six scored high for width measurements. However, the sample of diploids is not yet large enough to substantiate these tendencies. Furthermore, sampling within populations must be increased to adequately examine differences among diploids.

Overall, principal factor analysis identified a combination of spikelet and spike characters as the most important factor, which aligns the alloploid and thickspike groups and separates them from the rest. Awn length and leaf blade width and rib number may be useful in separating a distinct group within the diploids, but more sampling is needed to verify this.

Discriminant Analysis

Analysis of the entire 205 case data set separated the OTU's into four groups. Diploid and autoploid spicata, were combined into a single group. The thickspike OTU'S were the most distinct as shown in figure 8. The other groups were canonically aligned along a single axis with the Old World Pseudoroegneria grading into spicata, which graded into the alloploid group. The plot of canonical vari- ables grouped the alloploid OTU's more closely to spicata than to thickspike. The thickspike OTU's that appear closer to spicata and the alloploids actually were artificial crosses between thickspike and spicata.

In the discriminant analysis, nine variables were identified as useful for classification. Lemma pubescence was the most significant and reflected the main difference between the thickspike group and the others. Table 7 gives the classification function for the nine characters. Second glume width was recognized as the next most impor- tant variable and represented associated glume and lemma characters in separating the thickspike and alloploid groups from the others. In 18 the stepwise procedure, when second glume width was selected, F values for these other gluine and lemma variables fell dramatically. This apparent correlation was confirmed by principal factor analysis reported earlier. Likewise, awn length helped to separate the awned alloploids, autoploid, and diploid spAcata from the relatively awn- less Old World and thickspike groups.

Most variables chosen were spikelet or spike characters. Only adaxial palea vestiture, which was greatest in the Old World group, seemed an anomaly. For the purpose of a morphological key, discrimi- nant analysis showed that the alloploid and thickspike group separate from the others on glume characteristics, with second glume width the most reliable trait. Thickspike separates from alloploid based on lemma pubescence and, although not included in this analysis, rhi- zomes.

Spikelet width initially was identified by discriminant analysis as an important variable in separating groups. Widths were greater in the alloploid and Old World groups. However, this trait was difficult to measure on mature specimens because spikelets exhibited different stages of opening. Therefore, spikelet width perhaps reflected stage of maturity rather than its intended value. This variable subse- quently was removed from further analysis and discounted. It may be a character worth closer examination under more standard conditions as an aid in separating the alloploid from the diploid/autoploid group.

The jackknifed classification table (table 8) showed that 59 per- cent diploids, 64 percent autoploids, 87 percent alloploids, 89 per- cent Old World species, and 79 percent thickspikes were correctly grouped. When diploid and autoploid spicata are grouped together, 85 percent are correctly classified as spicata. Of the 17 misclas- sifications in this combined group, many had apparently logical expla- nations. T5247, from Park County, Wyoming, was classified as an alloploid. This accession was vegetatively collected in a uniform garden nursery in Montana which included several alloploids. The specimen examined may have been a contaminant in the row sampled. Two 19 samples of T37252 from Box Elder County, Utah were classified as thickspikes, which seems logical since all specimens of this example were rhizoinatous in the greenhouse. The entire accession probably was mislabeled as a diploid. One specimen each of D2837 and DSI17 were classified as an alloploid. Cluster analysis showed the eight samples of these accessions, from northern Idaho and eastern Washington, tobe highly variable and not clustered together. This may reflect intro- gressjon from lanceolatus or the a].j.oploid, which grow naturally in the area.

The other misciassifications are not easy to explain. In the other groups, seven alloploid accessions were classified in one of the Pseudoroegneria groups. Four of the niisclassifications belonged to P737, a strain originally collected in 1934 from northern Idaho. All but one of the specimens examined originated from seed several genera- tions removed from the original collection. P737 is a superior strain that has been increased for seed over a 50-year period at Pullman, Washington, where it has been exposed to other Pseudoroegneria and Elymus on a continual basis. In fact, there probably has been a concerted effort to isolate it from spicata, resulting in it being planted on occasion next to Elymus lanceolatus. It is no wonder that all three multivariate analyses show this accession to be highly variable. Factor analysis (figure 3) illustrates that some specimens have acquired the g].ume characteristics of spicata and have lost their awns either to introgression by lanceolatus or spicata. However, one specimen of P737 was collected from the original site in 1984 (OTtJ 158) and was not ].ong-awned, showing possible introgression as well. It was classified by discriminant analysis as an diploid. It was placedbycluster analysis in the relatively polymorphic group between the thickspike and alloploid clusters.

The three Old World pseudoroegneria misclassifications pro- bably have little significance and only reflect the similarities within the genus. The three thickspike misclassifications are three 20 of the four spicata x lanceolatus crosses that were arbitrarily placed in the thickspike group.

The discriniinant analysis validation procedure had little effect on classification of all groups except the diploids, whichtended to place more dip].oids into the alloploid and autoploid groups.

When only diploid and autoploid spicata were discriminated, adaxial glume vestiture and presence of awns proved to be the most important variables in separating the two groups. About 94 percent of the autoploids were awned, compared to 62 percent for diploids. This, no doubt, Indicates that most of the autoploids in this study have come from awned diploid parents, so awn presence is not a useful character. Adaxial glume vestiture showed a relatively weak correla- tion to adaxial palea vestiture and was virtually uncorrelated to any other spikelet trait. Therefore, reliance on this character for clas- sification seems inadvisable.

Discriminant analysis on the 165 OTU diploid/autoploid/alloploid data set separated the alloploids based on glume characters (table 9). There seems to be good morphological justification (figure 9), when combined with the cytological data, to separate the alloploid group out as a separate taxon. The data show mixing of traits through introgression but enough separation exists to correctly classify most of the alloploid group. 21

DISCUSSION

The results of this morphological study confirm the preliminary cytological conclusion that the alloploid group currently residing in Pseudoroegneria spicata belongs in Elymus. While the alloploid group is similar to spicata in its caespitose habit,long divergent awns, general growth form, and occupies similar habitats,its spike and spikelet characters are more closely related to Elymus lanceola- tus. In particular, its glumes are characteristically Elymus, short, narrow lanceolate, and scabrous. Furthermore, young seedlings are wider leaved, with basal leaf sheath pubescenceexceeding 1 mm in length, similar to lanceolatus. Glume characters often are impor- tant at the generic level (Stebbins 1982, Gould and Shaw 1983); there- fore, the alloploid group should be placed in Elymus.

Once in Elymus, the alloploid group must be properly situated in relationship to other species. As indicated, preliminary data shows it at least partially interfertile with lanceolatus, so it must closely relate to it. Natural hybrids at Logan, Utah (Dewey unpublished) are meiotically regular, fertile, and produce meiotically regularF11S,demonstrating the cross-compatability of the two types. Furthermore, chromosomes of the alloploid appear similar to the SR karyotype (Hsiao, unpublished). However, to correctly classify it, the alloploid group should be crossed with several forms of the lan- ceolatus complex, and with other SR species, such as E. trachy- caulus, E. glaucus, E. arizonicus, and E. elymoides, then compared cytologically. A numerical taxonomic study involving these species also would be helpful.

With the present state of knowledge about the alloploid group in a preliminary stage, several evolutionary scenarios arepossible. Both diploid P. spicata and Critesion (formerly Hordeum), the progenitors of the SR species (Dewey 1982), come from arid environ- ments. Characteristically, spicata contributed drought tolerance, cross-pollination, long life, and possibly the capability to form 22 rhizomes, whereas, Critesion contributed rapid growth, self- pollination, long awns, and a short life. Elymus trachycaulus, glaucus, and canadensis were spawned in mesic environments with rapid growth as its primary adaptive feature. All are relatively short-lived. Elymus elymoides probably evolved to inhabit open, at least somewhat disturbed, infertile sites, even more arid than those occupied by spicata. Such sites require a plant with a good dis- persal mechanism and the ability to establish quickly. These traits were provided by Critesion. The result was a sort of "perennial cheatgrass" in the form of the low growing, rapid developing, long- awned elymoides.

The discovery of the alloploid group perhaps provides a clue to the evolution of Elymus lanceolatus. This species occupies stabi- lized inland sand dunes during early seral stages of succession, often forming large colonies. It also is found along river courses and other sites that are semi-disturbed and show evidence of deposition. Such a plant requires good seedling vigor, rhizomes, and drought tole- rance. A large colony of this species often is a single plant with an extensive rhizome system, an adaptive feature on sand dunes where seedling competition can be detrimental for survival. Elymus lance- olatus is similar to the sand-binding species of Leymus, which also form large single plant colonies and are cross-pollinators. It seems logical that Pseudoroegneria spicata contributed long anthers and cross-pollination, drought tolerance, the abscence of awns, and possibly the genes for rhizome formation. On the other hand, Crite- sion provided seedling establishment, vigorous growth, and leaf sheath pubescence for added drought tolerance. It seems plausible that ancestors of lanceolatus were alloploids that occupied habi- tats similar to spicata. and that .tanceolatus evolved character- istics that enabled it to occupy sandier sites. The alloploid and sympatric lanceolatus could have evolved from a common ancestor, SR hybrid, and the alloploid group may closely resemble the ancestor. Some of the alloploid group are mildly sod-forming. 23

Much of the alloploid group grows in association with the awniess form of spicata, so its awns may not have derived from spicata in the manner E. lanceolatus ssp. albicans acquired awns (Dewey

1970). The awns of the alloploid could have been contributed by Critesion. Lanceolatus could have lost this characteristic through introgression from awniess spicata. Some of the alloploid group is awniess. However, this does not explain the vast expanses of long-awned spicata, which are associated with the usually awniess lanceolatus. As stated earlier, awns may be a trivial character on which to base evolutionary discussion of these grasses. However, Old World Pseudoroegneria appears to be predominantly awnless, so spicata either has evolved awns or has acquired them from related species in the Triticeae tribe.

The cluster analyses in this study were not able to clearly sepa- rate groups within spicata. Environmental influences appear to confuse any differences that may exist between diploids and autotetra- ploids. Most autotetraploids appear to have been derived from awned diploids and they appear to occupy moister habitats, but practical taxonomic separation does not seem to be possible. Furthermore, the fine-leafed awniess phase that seems to occupy much of the Palouse grassland in eastern Washington also was not adequately delineated in this study. More field work is necessary to determine if there is a clear ecological distinction between the awned and awnless types. No other subspecies or varieties of spicata were apparent and possibly await discovery after examination of more specimens from the eastern and southern portion of the range of species. 24

CONCLUS IONS

Since the cytological data, showing the relationship to lanceo- latus, is only preliminary, the aUoploid group will be conserva- tively recognized as a subspecies of Elymus lanceolatus.

ELYMUS LANCEOLATLYS (Scribner & J.G. Gould) Gould subspecies WAWAWAI J.R. Carlson & D.R. Dewey -- TYPE: Washington, Whitman County, breaks of Snake River Canyon on slopes along road from Pullman to Wawawai, north side of Snake River; site of numerous collections and source of P1-285272, voucher specimen currently in preparation.

Cespitose, sometimes weakly rhizomatous, perennial; culms (including inflorescence) 50 to 130 cm tall. Basal leaf sheaths usually moderately pubescent (with hairs greater than 1 mm) when young, becoming glabrous when mature. Culm leaf sheaths glabrous or rarely sparsely pubescent 6 to 18 cm long; blades adaxially densely pubscent often with hairs about 0.5 mm in length, abaxially glabrous or rarely sparsely pubescent, up to 26 cm long, 1.8 to 4.8 mm wide, flat with 10-20 ribs. Ligules truncate, 0.1 to 1.1 mm long. Flag leaf sheath 10 to 25 cm long, blade to 13 cm long and 1.1 to 3.9 mm wide. Inflorescence 6 to 20 cm long, spikelets solitary at the nodes, 10 to 22 mm long with 4 to 10 florets. Third rachis internode 5.0 to 13.7 mm. Glumes lanceolate, acute, often tapering somewhat below mid- length to the tip, unequal to subequal, occasionally awned up to 1.1 mm, the first 3.9 to 9.8 mm long, 0.4 to 1.6 mm wide, 2-5 veined at mid-length; the second 4.1 to 10.5 mm long, 0.5 to 1.8 mm wide, 3-6 veined; backs moderately to very scabrous; often curved to one side of first lemma. Lemmas (excluding awn) 7.0 to 12.2 mm long, 1.6 to 2.8 mm wide; awn usually present, up to 28 mm long, longer awns occurring in top half of spikelet; lemmas dorsally glabrous to slightly sca- brous, sparsely pubescent along edges towards base near attachment to rachil].a. Second rachilla internode 1.0 to 2.0 mm long with very short stiff appressed hairs. Palea two-keeled, sparsely to moderately strigulose inside and sparsely so outside usually toward the tip. 25

Anthers 3.7 to 5.8 mm at maturity. Chromosome number: 2n28 (Dewey 1982, this study). Distribution: Snake River Drainage, along the breaks and tributaries of the Salmon, Snake, and Yakima Rivers in northern Idaho, northeastern Oregon, and southeastern Washington.

The formal description of this new subspecies of Elymus lanceo- latus and location of the type specimen subsequently will be pub- lished in a recognized taxonomic journal.

For identification of field specimens, glume, spike and awn characters and cespitose habit should key to Elymus. Glumes are not as acuminate as in Pascopyrum and are not obtuse or truncate as in Pseudoroegneria. Some confusion could arise, because in many specimens of ssp. wawawai, the glumes begin tapering below mid- length and could be confused with Pascopyrum. However, Pasco- pyrum glumes usually begin tapering near the base and glumes often equal the first lemma in length, whereas, in ssp. wawawai the glumes taper nearer the midlength than the base, are subequal to lemmas, and are moderate to very scabrous.

One change is recommended in the generic key of North American Triticeae by Barkworth and Dewey (1985). In it cespitose Elymus must have anthers 2-3 nun long. However, subspecies wawawai has long anthers, 3-6 mm long.

Within Elymus, using the key of Barkworth et al. (1983), ssp. wawawai keys to step seven where changes to include the new sub- species become apparent. The following is a revised version of certain lines of the key on page 567 of the article.

7(1). Plants rhjzomatous or cespitose; anthers 3-6 mm long 8

Plants cespitose; anthers 1-3 mm long 10 26

8(7). Plants rhizomatous, sterile; anthers not well filled at

anthesis, not dehiscent E x pseudorepens

Plants fertile, anthers well filled at anthesis, dehiscent

(E.lanceolatus) 9

9(8). Plants cespitose or only weakly rhizomatous

E lanceolatus ssp. wawawai

Plants at least moderately rhizomatous 10

10(9). Lemmas awnless or with awn-tip less than 5 mm long

E lanceolatus ssp. lanceolatus

Lemmas with a divergent awn 5-12 mm long

E lanceolatus ssp. albicans

The remainder of the key is the same except for adjustments in numbering.

The data in this study confirm the morphological similarity bet- weenElymuslanceolatus ssp. wawawai and Pseudoroegneria spicata. If obvious grouping errors are removed, more than 90 per- cent of the alloploid wawawai can be correctly classified using morphological characters. However, there is still evidence in the data of introgression which is not surprising since natural triploid hybrids between the two taxa are occasionally found in the field. It is recommended that the entire set of spikelet and spike characters be examined during identification of specimens. Glumes of wawawai should be short, narrow, sharp-pointed, and scabrous. Although over- all spikelet dimensions are smaller, spikelets overlap rachis inter- nodes by about 50 percent in wawawai compared to 10 to 20 percent in spicata. As a result wawawai spikes appear more compact. Furthermore, live plants should be examined for basal leaf pubescence greater than 1 mm long. 27

FIGURE1. Distribution ofPseudoroegneria spicata inwestern North America, including thealloploid form;dots = diploids,circles =autoploids, x marks -alloploids. FIGURE 2. Condensed dendogram obtained from average linkage cluster analysis of 205 OTUS of Pseudo- roegneria and E.Zymus; A = P. spicata diploid; B = P. spicataautoploid; C = unknown alloploid;D =Old World Pseudoroegneria; E = E. lanceolatus. Numbers after the let- ters represent OTU frequency of each group in the cluster. Scale is expressed in euclidean distance.

1.3 A-2 A-0 A-17 h-5 A-17 A-3 A-13 A-2 A- 15 A-4 A-2 A-i B-0 B-0 B-i 8-0 8-12 B-8 B-i B-0 B-3 B-0 8-0 B-0 C-0 C-0 C-8 C-35 C-2 C-i C-i C-2 C-i C-0 C-0 C- 3 D-0 D-0 D-5 D-1 D-0 D-0 D-0 D-0 D-3 D-2 D-6 D-9 E-0 E- 9 E-2 E-3 E-0 E-0 E-0 E-0 E-0 E-0 E-0 E-0 2.0 2.1 2.1 2.2 2.3 2.3 2.4 2.5 2.5 2.5

2.9

3.0 3.1

3.3

37

4.5 FIGURE 3. Scatter plots of factor scores for spike/spikelet and awn characters in principal factor analysis of 205Pseudoroegneriaand Elymus OTU's; dots = diploids, squares - autoploids, x marks = alloploids, triangles = Old World species, and + marks = thickspike. 3.5

2.5 w zCD x XX X - El Z X '4.4 : A. x TEl: :. A '* A ii: :

-4.5 -3.0 -1.5 0.0 1.5 3.0 4.5J

FACTOR I(SPIKE/SPIKELET DIMENSIONS) FIGURE 4. Scatter plots of factor scores for spike/spikelet and awn characters in principal factor analysis of 165Pseudoroegneria OTU's;dots = diploids, squares - autoploids, x marks = alloploids. 3.5fl 2.5 0 0 * 0 0 0 zxx z0x ., z

.5 X W z 0 000 0 * 0 e x * -5; x 0 000 00 00 0 -1 5 0 00 0 00 0 0

-2.5

-4.5 -3.0 -1.5 0.0 1.5 3.0 4.5

FACTOR I(SPIKE/SPIKELET DIMENSIONS) FIGURE 5. cuimScatter lengths plots in of principal factor scores factor for analysis of 165 spike/spikelet characters and leaf! 3.5 OTU's; dots = diploids, squares - autoploids, x a marks = alloploids.Pseudoroegneria x a 0 a 0 0 1.5 5 x a a a x_Jw x 0 : a 0 0 ¶5 0 0 00 a a 0 ° a -5. . a a L -3.5 L -4.5 . FACTOR-3.0 I (SPIKE/SPIKELET DIMENSIONS) -1.5 0.0 1.5 3.0 4.5 FIGURE 6. Scatter plots of factor scores forspike/spikelet characters and leaf/ cuim widths in principal factor analysis of 165Pseudoroegneria OTU's; dots = diploids, squares - autoploids, xmarks = alloploids. 3.5 a

2.5: x D x a a 1.5 x a a a0' a xc a z x a z

a x a x 0 a

0 a a x

a

-3.5L 4.5 3.0 1.5 0.0 1.5 3.0 4.5

FACTOR I(SPIKE/SPIKELET DIMENSIONS) FIGURE 7. Scatter plots of factor scores for leaf/cuim widths and awn characters in principal factor analysis of 165PseudoroegneriaOTU's; dots = diploids, squares - autoploids, x marks = alloploids. 3.5__

2.5 x x 0 0 Ox 0 x I ci . xci ci 01ri0 0 00! 1% -.5 000 zx ci x i0e 0 0 I

-1.5 0 ci o ci Ci -. ci L

-4.5 -3.0 -L5 0.0 i.5 3.0 4.5I

FACTOR 3 (AWN LENGTH/ANGLE) FIGURE 8. Scatter plots of canonical variable scores for five groups of Pseudo- roegneria and Elymus. dots = diploids, squares - autoploids, x marks = alloploids, triangles = Old World species, and + marks =thickspike.

2.0; A A £ .a#AAA A 1.0 : ', 0.0 A A A Q , x x x

x I -2.0 Xe II I -3.0 . -4.0

-5.0 . +

-6.0 + .. L -6.0-4.5-3.0-1.5 0.0 1.5 3.0 4.5 6.0

CANONICAL VARIABLE I FIGURE 9. roegneriasquaresScatter -plots autoploids,spicata, of canonical x marks variable = alloploids. scores for three groups including the alloploid form; dots = diploids, of Pseudo- 3.0 T 0 2.0 x X X 0 0 0 3 0 3 3 IO x x X x 0 z 08 $ 0 0 0 0.0 XX XZ I 00 0 0 0 0 00 : 0 0 -:1.0 I X 11 I I XX X 0 O 0 0 0 0 8 -2.0 1 I 1 0 x 0 0 0 0 0 ° 3°L -4.0 -3.0 -2.0 -1.0 CANONICAL VARIABLE I 0.0 1.0 2.0 3.0 4.0 36

TABLE 1. Species of the genus Pseudoroegner.ia (Nevski) L6ve.

Species Name Ploidy Distribution

Pseudoroegneria strigosa (Bieb.) LSve SSp.strigosa 2N=14 Crimea (Black Sea) ssp. aegilopoides (Drobov) L6ve 2N=14 Siberia, N. China ssp. amgumensis (Nevski) Love unk Far East (USSR) ssp. jacutorum (Nevski) Lve unk Far East (USSR) ssp. kanashiroi (Owhi) Love unk ssp. reflexiaristata (Nevski) unk Ural Mtns (USSR) LOve

P. divaricata(BOiSS. and Bal.) Lve ssp. divaricata unk Turkey ssp. attenuatiglumis (Nevski) unk Caucasus Live

P. dsinalica (Sablina) LOve unk Caucasus P. kotovii (Dubovik) LOve unk P. ninae (Dubovik) L'dve unk P. cretacea (1(10k. and Prok.) Lôve 2N=28 SE Ukraine (USSR) P. stipifolia (Czern. ex Nevski) 2N=14 SE Ukraine (USSR) L5ve

P. tauri (Boiss. and Bal.) Lve ssp. tauri 2N=14 Turkey ssp. libanotica (Backel) LOve 2N=14 Asia Minor (Iran)

P. pertenuis (C.A. Mey.) Lve 2N28 Caucasus P. kosaninii (Nabelek) LOve unk P. sosnovskyi. (Backel) LOve unk Caucasus P. heidemaniae (Tsvelev) Ldve unk Caucasus 37

TABLE 1. Species of the genus Pseudoroegneria (Nevski) Lve(cont'd)

Species Name Ploidy Distribution

P. cognata (Hackel) Löve ssp. cognata 2N=14 Central Asia (USSR) ssp. shingoensis (t4elderis) Löve 2N=14

P. geniculata (Trin.) L6ve ssp. geniculata 2N28 Siberia, Mongolia ssp. nevskii (N.Ivanova in unk Grub.) L3ve ssp. pamirica (Melderis) Lve 2N28 ssp. pruinif era (Nevski) L6ve unk Ural Mtns (USSR) ssp. sythica (Nevski) LSve 2N=28 Crimea (USSR)

P. setulifera (Nevski) Lve unk Central Asia (USSR) P. armena (Nevski) L'óve unk Caucasus P. gracillima (Nevski) L6ve unk Caucasus

P. spicata (Pursh) Love ssp. spicata North America ssp. inermis (Scrib. and Smith) 2N=14 North America LOve

P. stewartii (Melderis) L6ve unk 38

TABLE 2. Characters used to score 205 OTU's of Pseudoroegneria and Elymus.

Computer Two-state Characters: Code Presence of longer hairs 0.5 nun) on adaxial leaf LHR blade. Presence of first glume awns. FAW Presence of second glume awns. SAW Presence of first lemma awns. LAW

Three-state Characters: First glume tip shape (blunt, sharp, very sharp). FSH First glume abaxial texture (smooth, scabrous, FGT very scabrous). Second glume tip shape. SSH Second glume abaxial texture. SGT

Vestiture Characters: Culm leaf blade adaxial (none, sparse, moderate, dense). VAD Cuim leaf blade abaxial. VAB Culm sheath. SHV First glume adaxial. FGV Second glume adaxial. SGV First lemma adaxial. LVI First lemma abaxial. LVO First palea adaxial. PVI First palea abaxial. PVO

Discrete Ouantitative Characters: Cuim nodes. NOD Cuim leaf blade ribs. RIB Spikelet nodes. SPN First glume nerves. RVN Second glume nerves. SVN Third spikelet florets. FLO 39

TABLE 2. Characters used to score 205 OTUs ofPseudoroegneriaand Elymus. (continued)

Computer Continuous Quantitative Characters: Code Culm leaf ligule length. LIG Cuim leaf length. CLN Culm leaf blade width. BLW Culm leaf sheath length. SilL Cuim length (crown to bottom of spike). ClAN Culm diameter at bottom internode. CDM Culm leaf auricle length. AUR Top internode length (top node to bottom of spike). IND Flag leaf blade length. FBL Flag leaf blade width. FBW Flag leaf sheath length. FSL Spike length. SPK Spikelets per cm spike. SCM First rachis internode length. FRI Third rachis internode length. TRI Third spikelet, second rachilla internode length. RAC Third spikelet length. SPL Third spikelet width. SPN First glume length. FGL First glume awn length. FAL First glume width. FGW Second glume length. SGL Second glume awn length. SAL Second glume width. SGW First lemma length. LML First lemma awn length. LAL First lemma width. LMW Longest lemma awn length in bottom half of third LBA spike let. Longest lemma awn length in top half of third LTA spikelet. First palea length. PAL First palea width. PAW Average lemma awn angle of third spikelet. LAA 40

TABLE 3. Morphological characters selected for use in cluster analysis of Pseudoroegneria and Elymus OT(J's.

COMPUTER RUN

1 2 3 4

Groups 5 3 3 3 OTU's 205 165 165 165

Characters VAD VAD NOD TRI BLW LHR RIB SCM TRI BLL LIG RAC SHV BLL FGL SCM FBW TRI SHL FSH FGL SCM CLN FGW FGW FAL IND FGV FVN FSH SPK FVN SSB FGW FBL FGT SGW SGV FBW SGL SVN LAW FSL SSH LAW LTA SPL 5GW LMW PAW SGV LVO SVN LBA SGT PAW LML PVD LMW

Character Mixture Mixture Vegetative Spikelet Type Type

NOTE: Compare computer codes above to description of characters in table 2. discriminant analysis and classificationfunctions TABLE 7. Variables selected by stepwise for groups of Pseudoroegneria and ElymusOTU's; all data standardized.

Autote- Old Thick- Alloploid World spike Variables Diploid traploid

5 Groups. 205 OTU's

-0.083 6.518 Abaxial lemma vestiture (LVO) -0.486 -0.201 -0.821 -1.362 1.093 -1.184 Second glwne width (SGW) 0.590 0.404 1.826 -1.020 Adaxial palea vestiture (PVI) -0.461 -0.559 0.405 -0.941 0.532 Rachilla internode length (RAC) 0.531 0.319 -0.677 -1.975 -0.207 Second glume abaxial texture (SGT) 0.216 0.121 0.622 -1.889 -0.687 Lemma awn length (top i) (LAL) 0.092 0.483 0.685 -0.310 1.460 0.735 Lemma width (TAlL) -0.530 0.360 0.180 -0.764 Second glume adaxial vestiture (SGV) -0.189 0.934 -0.144 0.859 0.007 0.224 Second glume shape (SSH) -0.364 -0.633 -3.592 -5.862 -11.656 Constant -2.051 -2.425

3 Groups, 165 OTU's

-1.669 First glume width (FGW) 0.787 0.796 0.069 Second glume adaxial vestiture(SGV) -0.347 0.792 Rachilla internode length (RAC) 0.574 0.388 -1 . 104 1.116 Culm length (CLN) -0.650 -0.211 -1.350 Third rachis internode (TRI) 0.936 -0.138 1 .046 Lemma length (TAIL) -0.423 -0.684 0.757 Adaxial palea vestiture (PVI) -0.348 -0.385 -0.568 Cuim nodes (NOD) 0.468 -0.253 -0.243 Palea width (PAW) -0.265 1.107 -3.003 Constant -1.640 -2.175 42

TABLE 5. Variance explained by factors identified in principal factor analysis of three combinations of morphological characteristics measured on 205 OTU's of Pseudoroegneria and Elymus lanceolatus.

CHARACTER COMBINATIONS 1

Factor A B C spikelet and spike dimensions 21.6 25.6 30.1 lemma awn length and angle 8.9 8.3 11.3

leaf and culm lengths, spike length 6.8 10.5 9.7

leaf blade width, number of ribs, cuim diameter, spike dimensions 7.8 7.9 8.2 glume shape 5.8 5.7 glume adaxial vestiture 2.8 4.5

glume texture 3.0 4.2

glume awns 4.6 3.4 4.1

lemma vestiture 3.9 2.4

abaxial leaf blade and sheath vestiture 1.9

cuim length and number of nodes 1.8 2.6 3.5

Total 68.9 75.1 66.9

1 A - all 55 character variables included in the analyses B - 9 few-state characters omitted, 46 characters analyzed C - all 17 few-state characters omitted, 38 characters analyzed. 43

TABLE 6. Rotated estimated factor loadings in principal factor analysis (PFA) of 165 OTU's ofPseudoroegneria.

Rotated Estimated Factor Loadinqs1 Factors Fl F2 F3 F4 Communalities2

Spike, Spikelet characters (El)

FGW .867 - - - 0.89 SGW .849 - - - 0.87 SGL .778 - - - 0.85 FGL .759 - - - 0.80 SVN .750 - - - 0.66 FVN .720 - - - 0.58 LNW .715 - - - 0.68 LML .680 - - - 0.78 SCM -.658 - - - 0.87 PAL .611 - - - 0.80 SPL .602 - - - 0.84 TRI .591 .463 - - 0.85 RAC .437 - - 0.65

Height, Length Measurements (F2)

FSL - .834 - - 0.82 SHL - .797 - - 0.72 SPK .447 .768 - - 0.90 FRI .496 .557 - - 0.83 CLN - .548 - - 0.83 IND - .469 - - 0.67 SND - .439 - 0.77

Awn Lenqth, Angle (F3) LBA - .939 - 0.93 LTA - - .925 - 0.90 LAL - - .922 - 0.90 LAA - - .815 - 0.73

Thickness. Width Measurements fF4)

BLW - - - .862 0.84 RIB - - - .776 0.75 FBW - - - .773 0.84 CDM - - - .743 0.68

Cumulative % of total sample variance explained 29.4 42.2 51.8 59.7

1 Rotated factor loadin9s are shown f or the first four factors out of nine chosen by PFA of 36 morpho1oglca characters and 165 DID's. 2Communalities obtained from nine factors chosen by PFA. TABLE 4. Group frequency in five major clusters formed by analysis of 17 morpho- logical characters of 205 OTU's of Pseudoroegneria and Elymus

Cluster Number Awniess Autote- Allote- Old Thick- Group Subgroup OTU's Diploid Type traploid traploid World spike

1 a 2 2 (0) 0 0 0 0 b _Q (0) _Q

Group 1- Total 13 2 (0) 0 0 0 9

2 33 17 (4) 1 8 5 2

3 44 5 (1) 0 35 1 3

4 a 31 17 (5) 12 2 0 0 b 12 3 (0) 8 1 0 0 c 21 13 (5) 7 1 0 0

ci 4 2 (0) 0 2 0 0 e 22 15 (12) 3 1 3 0 f 6 4 (3) 0 0 2 0 g 8 (0) 0 0 6

Group 4 - Total 104 56 (25) 30 7 11 0

5 13 1 (0) 0 3 9 0 45

TABLE 8. Jackknifed classification of 205 OTU's into 5 and 3 groups of Pseudoroegneriaand Elymus by discriminant analysis.

NumberofCases Classified into GrouPs Percent Auto- Allo- Old Thick- Group Correct Diploid ploid ploid World spike

5 Groups, 205 OTU's

Diploid 59.3 48 17 10 3 3

Autoploid 64.5 10 20 0 1 0

Alloploid 86.8 5 1 46 1 0

Old World 88.5 2 1 0 23 0

Thickspike 78.6 2 1 0 0 11

Total 72.2 67 40 56 28 14

3 Groups, 165 OTU's

Diploid 75.3 61 17 3

Autoploid 83.9 5 26 0

Alloploid 88.7 3 3 47

Total 81.2 69 46 50 46

B I BL IOGRAPHY

Atkins, R. J., N. E. Barkworth, and D. R. Dewey. 1984. A taxo- nomic study of Leymus ambiguus and L. sai.inus (Poaceae:- Triticeae). Systematic Botany 9(3) :279-294.

Barkworth, N. E. and D. R. Dewey. 1985. Genomically based genera in the perennial Triticeae of North America: Identifica- tion and membership. Amer. 3. Bot. 72(5):767-776.

Barkworth, N. E., D. R. Dewey, and R. 3. Atkins. 1983. New generic concepts in the Triticeae of the Intermountain region: Key and comments. Great Basin Naturalist 43(4):561-572.

Bowden, W. N. 1965. Cytotaxonomy of the species and inter- specific hybrids of the genus Agropyron in Canada and neigh- boring areas. Can 3. Bot. 43:1421-1448.

Collins, D. D. 1965. Ecological, biosystematic, and biochemical studies of species in the genus Agropyron Gaertn. native to Montana. Ph. D. Thesis Montana State University, Bozeman, Montana.

Daubenmire, R. F. 1939. The taxonomy and ecology of Agropyron spicatum and A. inerme. Bull Torrey Bot. Club 66:327-329.

Dewey, D. R. 1964. Natural and synthetic hybrids of Agropyron spicatum x Sitanion hystrix. Bull. Torrey Bot. Club 91:396-405.

Dewey, D. R. 1970. The origin of Agropyron albicans. Amer. 3. Bot. 57(1):12-18.

Dewey, D. R. 1982. Genomic and phylogenetic relationships among North American perennial Triticeae. In 3. R. Estes, R. 3. Tyrl, and 3. N. Brunken (eds) Grasses and grasslands: systematics and ecology. University of Oklahoma Press. pp. 52-87.

Dewey, D. R. 1983. Historical and current taxonomic perspec- tives of Agropyron, Elymus, and related genera. Crop Science 23:637-642.

Engelman, L. 1982 rev. BMDP2M - cluster analysis of cases. BMDP Statistical Software, Inc. University of California, Los Angeles.

Engleman, L. 1982 rev. BMDP2D - detailed data description, including frequencies. BMDP Statistical Software, Inc. Univer- sity of California, Los Angeles. 47

Frane, J. and R. Jennrich. 1982 rev. BMDP4M - factor analysis. BMDP Statistical Software, Inc. University of California, Los Angeles.

Gould, F. w. and R. B. Shaw. 1983. Grass systematics. Texas A&M University Press, College Station.

Hartung, M. E. 1946. Chromosome numbers in Poa, Agropyron, and Elymus. Amer J. Sot. 33:516-531.

Hitchcock, A. S. 1951. Manual of the grasses of the United States. 2nd edition revised by Agnes Chase. USDA Misc. Pub. 200. U.S. Gov't. Printing Office, Washington, D.C.

Hitchcock, C. L. 1969. Gramineae. In C. L. Hitchcock, A. Cronquist, and N. Ownbey (eds).Vascular plants of the Pacific Northwest, part 1. Univ. of Washington Press, Seattle. pp. 383-725.

Ho, C. 1982 rev. BMDP1S - inultipass transformation. BMDP Statistical Software, Inc. University of California, Los Angeles.

19 Holmgren, A. H. and N. H. Holmgren. 1977. Poaceae. In A. Cronquist et al (eds) Intermountain Flora, Vol. 6. Columbia University Press, New York. pp. 175-460

20 Jennrjch, R and P. Sampson. 1982 rev. BMDP7M - stepwise dis- criminant analysis. BMDP Statistical Software, Inc. University of California, Los Angeles.

21 L$ve, A. 1982. Generic evolution of the wheatgrasses. Biol. Zentra].bl. 101:199-212.

22 L6ve, A. 1984. Triticeae conspectus. Feddes Repert. 95:425- 521.

23 Nevski, S. A. 1934. Tribe Hordeae Benth. pp. 469-578. In V. L. Komarov (ed.) Flora of the U.S.S.R., Vol. II. Israel Program for Scientific Translations. Jerusalem.

24. Pyrah, G. L. 1983. Agropyron arizonicum (Grammeae:Triticeae) and a natural hybrid from Arizona. Great Basin Naturalist 43(1):131-136.

25 Sneath, P. H. A. and R. R. Sokal. 1973. Numerical taxonomy. The principals and practices of numerical classification. Freeman, San Francisco.

26. Stebbins, G. L. 1982. Major trends of evolution in the Poaceae and their possible significance. In J. R. Estes, R. J. Tyrl, and J. N. Brunken (eds) Grasses and grasslands: systematics and eco- logy. University of Oklahoma Press. pp. 3-36. 48

27. USDA, Soil Conservation Service (SCS). 1947. Annual technical report of the Pullman Plant Materials Center. 49

APPENDICES 50

APPENDIX TABLE 1. Collection information for 205 OTU's ofPseudo- roegneriaand Elymus.

Accession OTU OTU Number Origin Source

DIPLOIDS

1 D1252 Cache County, Utah Evans 12-53 2 D1252 " Evans 12-51 3 D1252 I' Evans 12-55

4 D1252 " Evans 12-55 5 T7686 Kootenai County, Idaho Moro 1

u 6 T7686 Moro 2 7 D2839 Whitman County, Washington Evans 12-57 8 D2839 'I Evans 12-60 9 D2839 " Evans 12-59 10 D2839 Evans 12-56

11 P1232140 Glacier County, Montana OSU GM (L) 12 T7697 Davis County, Utah OSU GM (L) 13 T40584 Kittitas County, Washington OSU GM (L) 14 T5252 Lincoln County, Montana OSU GM (L) 15 T7699 Colorado OSU GM CL)

16 T40666 Baker County, Oregon OSU GM (L) 17 T40449 Austin, Nevada OSU GM (L) 18 T37190 Columbia County, Washington OSU GM (L) 19 T40451 Eureka, Nevada OSU GM (L) 20 T40553 White Pine County, Nevada OSU GM CL)

21 T5247 Park County, Wyoming OSU GH (L) 22 T40455 White Pine County, Nevada OSU GM (L) 23 T40454 Ely, Nevada OSU Gil CL) 24 T40450 Austin, Nevada OSU Gil CL) 25 T40453 White Pine County, Nevada OSU GM CL)

26 T37252 Box Elder County, Utah OSU Gil CL) 27 T37252 " OSU Gil (L) 28 T37252 0 OSU GM CL) 29 T37189 U OSU GM (L) 30 P739 Asotin County, Washington Evans 13-16

31 P739 1 Evans 13-18 32 P739 0 Evans 13-17 33 p739 U Evans 13-20 34 D2837 Adams County, Idaho Evans 13-23 35 D2837 " Evans 13-25 51

APPENDIX TABLE 1. Collection information for 205 OTU's ofPseudo- roegneria and Elymus. (continued)

Accession OTU OTrJ Number Origin Source

36 D2837 Adams County, Idaho Evans 13-22 37 D2837 it Deer Pens 33-35 38 D2837 " Deer Pens 33-34 39 D2837 " Deer Pens 33-38 40 D2837 " Deer Pens 33-33

41 D2837 Deer Pens 33-37 42 Whitmar Whitman County, Washington Moro 3 43 Whitmar " Moro 1 44 Whitmar U Moro 10 45 Whitmar Moro 9

46 Whitmar 'I Evans 12-41 47 Whitmar (I Evans 12-45 48 Whitmar U Evans 12-44 49 Whitmar U Evans 12-43 50 P1236670 Nelson, British Columbia Evans 12-48

51 P1236670 U Evans 12-46 52 P1236670 Evans 12-50 53 DS117 Asotin County, Washington Deer Pens 10-19 54 DS117 " Deer Pens 10-25 55 DS117 " Deer Pens 10-18

56 DS117 " Deer Pens 10-26 57 DS1I7 (I Deer Pens 10-21 58 DS117 U Deer Pens 10-20 59 DSI2O Whitman County, Washington Deer Pens 8-30 60 DS12O " Deer Pens 8-25

61 DS12O U Deer Pens 8-46 62 DS12O U Deer Pens 8-40 63 DS12O 'I Deer Pens 8-26 64 DS12O Deer Pens 8-27 65 D2840 U Evans 12-65

66 D2840 " Evans 12-62 67 D2840 " Evans 12-64 68 D2840 " Evans 12-61 69 D2838 Big Horn County, Montana Evans 14-22 70 D2838 U Evans 14-25 52

APPENDIX TABLE 1. Collection information for 205 OTU's ofPseudo- roegneriaandElymus.(continued)

Accession OTU OTU Number Origin Source

71 D2838 Big Horn County, Montana Evans 14-23 72 D2838 " Evans 14-25 73 BB1600 Custer County, Idaho Evans 14-15 74 BBI600 " Evans 14-14 75 BB1600 u Evans 14-11

76 D2836 Strevel, Utah Evans 14-19 77 D2836 u Evans 14-20 78 BB7031 Colorado Evans MA 7-31 79 BB7031 Evans MA 7-32 80 P1232128 Idaho County, Idaho Moro 7

81 P1232127 U Moro 3

AUTOTETRAPLOIDS

82 P1236669 Ainsworth, British Columbia Moro 3

83 P1236669 " " Moro 1 84 P1236681 Coleman, Alberta Moro 4 85 P1236681 " Moro 1 86 P1236681 " Evans 13-46

87 P1236681 I' Evans 13-49 88 P1236681 " Evans 13-47 89 P1236681 " Evans 13-48 90 P1286198 Whitman County, Washington Evans 13-54 91 P1286198 U Evans 13-55

92 P1286198 Evans 13-51 93 P1286198 " Evans 13-48 94 P1286198 U Moro 8

95 P1232124 U Evans 13-8 96 P1232124 " Evans 13-7

97 P1232124 Evans 13-10 98 P1232124 " Evans 13-9 99 P7845 Lewis County, Idaho Evans 14-10 100 P7845 II Evans 14-7 101 P7845 " Evans 14-6

102 P7845 U Evans 14-8 103 P7845 " Moro 104 P7845 Moro 105 P7845 Moro 106 P7845 Moro 5 53

APPENDIX TABLE 1. Collection information for 205 OTU's ofPseudo- roegneriaand Elymus. (continued)

Accession OTU OTU Number Origin Source

107 T7681 Clearwater County, Idaho Moro 4 108 T7681 'I Moro 9 109 T7681 U Moro 6 110 T7681 'I Moro 3 111 P1236671 Boswell, British Columbia Moro 4

112 P1236671 U Moro 5

ALLOTETRAPLOIDS

113 Secar Nez Perce County, Idaho Moro 6 114 Secar 'I MOrO 3 115 Secar " Moro 2 116 Secar 'I MorO 4 117 Secar u Evans 14-70

118 Secar U Evans 13-72 119 Secar I, Evans 13-75 120 Secar I' Evans 14-68 I' 121 Secar Evans 14-66 122 T19774 Sweetwater County, Wyoming OSU GH (L)

123 T19774 H OSU GH (L) 124 T25668 Big Horn County, Wyoming OSU GH (L) 125 P1285272 Whitman County, Washington Moro 10 126 P1285272 'I Moro 6 127 P1285272 Moro 2

128 P1285272 I' Moro 7 129 P1285272 I' Evans 13-35 130 P1285272 I' Evans 13-32 131 P1285272 U Evans 13-34 132 P1285272 U Evans 13-33

133 D2842 Idaho County, Idaho Evans 13-37 134 D2842 " Evans 13-40 135 D2842 Evans 13-39 136 DS122 Whitman County, Washington Deer Pens 7-24 137 DS122 Deer Pens

138 DS122 " Deer Pens 7-27 139 DS122 " Deer Pens 7-28 140 DS122 " Deer Pens 7-23 141 DS129 " Deer Pens 3-37 142 DS129 " Deer Pens 3-36 54

APPENDIX TABLE 1. Collection information for 205 OTU's of Pseudo- roegneria and Elymus. (continued)

Accession OTU OTU Number Origin Source

143 DS129 Whitman County, Washington Deer Pens 3-30 144 DS129 'a Deer Pens 3-23 145 DS129 H Deer Pens 3-26 146 DS129 H Deer Pens 3-34 147 DS129 I' Deer Pens 3-32

148 Ds129 II Deer Pens 3-33 149 DS129 Deer Pens 3-19 150 DS129 " Deer Pens 3-35 151 D2841 Adams County, Idaho Evans 12-67 152 D2841 " Evans 12-68

153 D2841 " Evans 12-70 154 D2843 Whitman County, Washington Evans 13-45 155 D2843 " Evans 13-44 156 D2843 " Evans 13-40 157 D2843 " Evans 13-42

158 P737 Latah County, Idaho OSU GH (L) 159 P737 H Evans 14-5 160 P737 " Evans 14-1 161 P737 11 Evans 14-2 162 P737 II Moro 5

163 P737 H Evans 14-3 164 T40388 Garfield County, Washington Moro 165 T37193 Adams County, Washington Moro 6

OLD WORLD SPECIES

166 P1440095 Pseudoroegneria stipifolia -USSR (4x) Evans 11-47 " 167 P1440095 'I Evans 11-48

11 " 168 P1440095 Evans 11-49 169 P1440095 0 " Evans 11-50 170 P1325181 U -USSR (2x) Evans 11-51

171 P1325181 'a Evans 11-52 172 P1325181 U H Evans 11-53 173 BB1161 H -USSR (4x) Evans 11-61

174 BB1161 a' Evans 11-62 175 P1401329 P.,tauri 2- Iran (4x) Evans 12-20 55

APPENDIX TABLE 1. Collection information for 205 OTU's of Pseudo- roegneria and Elymus. (continued)

Accession OTU OTU Number Origin Source

'176 P1401330 p. tauri - Iran (4x) Evans 12-25 177 P1401330 II Evans 12-23 'I 178 P1401330 I' Evans 12-22 179 P1401330 I' 'I Evans 12-21 180 P1380649 p. tauri ssp. libanotica - Iran (2x) Evans 12-36

'I IS 181 P1380650 Evans 12-29 I' H 182 P1380650 Evans 12-26 183 BB1272 P. strigosa - USSR (4x) Evans 12-75 184 BB1272 SI II Evans 12-74 185 BB1272 II Evans 12-72

186 BB6573 P. cognata - USSR (2x) Evans 15-17 187 BB6573 SI II Evans 15-20 188 BB6573 5' Evans 15-15 189 BB6573 I' Evans 15-16 190 BB6573 I' II Evans 15-12

191 BB6573 'S 5' Evans 15-13

ELYMUS LANCEOLATUS

192 BB0138 PSSP x ELLA Cross Evans 138-13 193 BB0138 H H Evans 138-i 194 BB0138 Il U Evans 138-22 195 BB0138 I' Evans 138-6 196 BB5001 E. 1. albicans - Canada Evans 13-31

197 BB5001 H 'S Evans 13-32 198 P14943 Sherman County, Oregon Evans 7-37 199 P14943 Evans 7-40 200 P14943 Evans 7-41 201 Sodar Grant County, Oregon Evans 14-9

202 D2846 E.1. griffithsii - Wyoming Evans 13-42 203 D2846 H Evans 13-41 204 D2846 I, I, Evans 13-45 205 D2846 Evans 13-44 56

APPENDIX TABLE 2. Environmental conditions at four uniform garden nurseries sites, which were sources of 192 Pseudoroegneriaand 14 Elymus OTtJ's.

OTU's Age of Annual Plant Location Repre- Plants Latitude Precip. Hardiness sented (years) (degrees) (cm) Zone Soils

Evans Farm DIPL-37 3 42° 40 4b loam Logan, UT AUTO-16 0% slope ALLO-23 OLDW-26 THIC-14

Deer Pens DIPL-17 3 42° 40 4b gravelly Logan, UT AUTO-0 loam ALLO15 droughty OLDW-0 3% slope THIC-0

Moro Exp. DIPL-8 2 45° 28 6a silt loam Station AUTO-15 10% slope Moro, OR ALLO-li OLDW-0 T}1IC-0

450 Greenhouse DIPL-19 1 N/A N/A OSU Corvallis, OR AUTO-0 potting ALLO-4 mix OLDW-O THIC-O 57

APPENDIX TABLE 3. Source of OTU of known diploid, triploid, auto- tetraploid, and allotetraploid forms of Pseudoroegneria spicata.

State! AccessionRefe Town County Province Number ence

DIPLOIDS

Cypress Hills Park Saskatchewan Senn 2435 1 Senn 6202 1 Waterton Lakes NP (Mt.Crandall) Alberta Senn 2613 1 Waterton Lakes NP (Banff) Senn 46-200-6,'7 1 Grand Forks British Columbia Senn 5937 1

Contact Elko Nevada Senn 5921 1 Yreka Siskiyou California Keck 6354 1

Robinette Baker Oregon Senn 5928 1 Monarch Cascade Montana Senn 5928 1 Yellowstone Park Wyoming Senn 6227 1

Wapiti C' Senn 6226 Shell Big Horn Senn 6230 1 Cambridge Washington Idaho Senn 5929 1 Penticton British Columbia Senn 5940-41 1 I' Hedley Senn 5813 1

I' Lytton Senn 5841 1 I' Vernon Calder 10171 1 Idaho City Boise Senn 5925 1 Susanville Lassen California 2 Canby 'I P6408 2

Shaniko Oregon P27 19 2 Mabton Washington P738 2 Walla Walla Walla Walla P9180 2 Lind Adams P7413 2 Soap Lake Grant P60 18 2

Coulee City I, P742 2 Spokane Spokane 'I P4915 2 Oneyka I' P64 11 2 Colton Whitman 'I P3537 2 Pullman I' P2522 2

He is e Idaho P27 24 2 Osgood Range Nevada 2 Anatone Asotin Washington P739 3,4 Dayton Lyon Nevada P7692 4 Park Wyoming T5247 4 58

APPENDIX TABLE 3. Source of OTU of known diploid, triploid, auto- tetraploid, and allotetraploid forms of Pseudoroegneria spicata. (continued)

State! Accession Ref er- Town County Province Number ence1

Big Horn Wyoming T25670 4 Jefferson Montana T25750 4 Helena Lewis & Clark U T25780 4 S ti 1 lwater ,. T25918 4 Madison T25976 4

U U T25982 4 Wildwood Dam Park Wyoming T26080 4 The Dalles Wasco Oregon T37170 4 Biggs Junction Sherman T37171 4 Phillipi Canyon Gilliani T37172 4

Arlington T37173 4 Juniper Canyon Umatil la I, T37174 4 Warm Springs Wasco T37175 4 Warm Springs Jefferson T37176 4 Madras T37177 4

Prineville Crook T37178 4 Taylor Butte T37179 4 Glass Butte Lake T37180 4 Squaw Butte Harney U T37181 4 Riley Harney Oregon T37182 4

Burns 'I U T37183 4 Princeton 'I U T37184 4 Indian Crk Butte Ma 1 heur U T37185 4 Sheephead Mtns U T37186 4 Bend Deschutes U T37187 4

Fossil Wheeler I, T37188 4 Columbia River Walla Walla Washington T37189 4 Delany Columbia T37190 4 Central Ferry Garfield T37191 4 LaCrosse Whitman It T37192 4

Hooper Adams I, T37193 4 Kahlotus Franklin U T37194 4 Sulfur U T37195 4 El topia 'I T37196 4 Kennewick Benton U T37197 4 59

APPENDIX TABLE 3. Source of OTU of known diploid, triploid, auto- tetraploid, and allotetraploid forms of Pseudoroegneria spicata. (continued)

State/ AccessionRefer- Town County Province Number ence1

Plymouth Benton Washington T37198 4 Crow Butte U T37199 4

The Dalles Klickitat 'I T37200 4 Kaniloops British Columbia T37236 4 T37237 4

Toe Jam Creek Elko Nevada T37238 4 Willow Lake II 11 T37239 4 Blackfoot Bingham Idaho A35 4 Nephi Juab Utah A6 4 New Meadows Adams Idaho K26 3,4

Colton Whitman Washington K36 3,4 K34D2839. 4 Rangely Rio Blanco Colorado T37544 4

Glade Park " T37545 4 Buford 'I T37546 4

I' T37547 4 T37548 4 Eagle-Glenwood T37570 4 Wolcott T37571 4 Virginia City Montana T39094 4

Colton Whitman Washington K35 4 Hinkley Summit Humboldt Nevada A26 4 National Forest A28 4 Satus Pass Klickitat Washington T40425 4 Levan Juab Utah T40427 4

Grand Ronde Rvr Wallowa Oregon T40432 4 U " " T40433 4

Troy " T40434 4 II II U T40435 4 Ft. Rock Lake U T40436 4

Silver Lake U T40437 4 " Summer Lake " T40438 4

Paisley " " T40439 4 Chandler Wayside U T40440 4 Goose Lake Modoc California T40441 4 60

APPENDIX TABLE 3. Source of OTU of known diploid, triploid, auto- tetraploid, and allotetraploid forms of Pseudoroegneria spicata.(continued)

State! AccessionRefer- Town County Province Number ence1

Alturas Modoc California T40442 4 Likely U T40443 4 Ravendale Lassen U T4O444 4 II T40445 4 II U T40446 4

Doyle ft It T40447 4 Austin Lander Nevada T40448 4 II U T40449 4 Scott Summit II II T40450 4 Eureka Eureka T40451 4

Little Antelope White Pine T40452 4 I, It (1 T40453 4 Ely U ft T40454 4 Connors Pass ft T40455 4 Current Summit Nye T40456 4

Yreka Siskiyou California T40457 4 Grass Valley Sherman Oregon T40458 4 Shaniko Wasco T40459 4 Clarno U T40460 4 Service Creek Wheeler a, T40461 4

Kimberly Grant U T40462 4 Dayville U U T40463 4 John Day U T40464 4 Unity Baker T40466 4 Irons ides Malheur It T40467 4

Brogan II It T40468 4 Ely White Pine U T40469 4 Idaho Falls Bonneville Idaho T40474 4 Dubois Clark T40475 4 Spencer II II T40476 4

Monida Pass II T40477 4 Dell Beaverland Montana T40478 4 U II T40479 4 Barrets Dam U U T40480 4 Dillon II U T40481 4 61

APPENDIX TABLE3. Source of OTU ofknowndiploid, triploid, auto- tetraploid, and allotetraploid forms of Pseudoroegneria spicata. (continued)

State! Accession Ref er- Town County Province Number ence1

Melrose Madison Montana T40482 4 Divide Silver Bow T40554 4 Butte N 'S T40555 4 Homestake Pass Jefferson U T40556 4 Cardwell U T40557 4

Three Forks Broadwater N T40558 4 Towns end U T40559 4 Winston U N T40560 4 Helena Lewis & Clark U T40561 4 Avon Powell T40562 4

Deer Lodge U T40563 4 Gold Creek N U T40564 4 Maxville Granite 'I T40565 4 Bearmouth Exit N U T40566 4 Miltown Missoula T40567 4

Missoula N N T40568 4 Arlee Lake T40569 4 Polson N T40570 4 Big Fork Flathead T40571 4 St. Mary's Lake Glacier H T40572 4

Dupuyer Pondera U T40573 4 Chateau Teton U T40574 4 Vaughn Cascade T40575 4 Augusta Lewis & Clark T40576 4 Wolf Creek 'I U T40577 4

Boulder Jefferson N T40578 4 Logan Gal latin I' T40579 4 Norris EXP. Sta. Madison N T40580 4 Park City Stillwater I' T40581 4 El lens burg Kittitas Washington T40582 4

Virden Kittitas Washington T40585 4 Swank Pass Chelan T40586 4 Wenatchee I' T40587 4 Chelan T40588 4 Brewster Douglas U T40589 4 62

APPENDIX TABLE 3. Source of OTU of known diploid, triploid, auto- tetraploid, and allotetraploid forms of Pseudoroegneria spica ta.(continued)

State! AccessionRefer- Town County Province Number ence'1

Hwy 17 Douglas Washington T40590 4 Grand Coulee Dam T40591 4 Wilbur Lincoln T40592 4 Reardon T40593 4 Spangle Spokane T40594 4

Bozeman Gallatin Montana T40598 4 Steptoe Cn Whitman Washington DS-101 4 Schlee Cn U DS-102 4 Snake River ft DS-109 4 Hellers Bar Asotin DS-117 4

Wawawai Cn Whitman U DS-120 3 N DS- 121 3 U DS-126 3 Penewawa Cn U U DS-127 3 II U DS- 130 3

N II DS-134 3 Schlee Cn U DS-103 3 Wawawai Cn ft U DS-125 3 Penewawa Cn U DS- 128 3 Idaho Idaho P1232127 3,4

P1232128 3,4 El ko Nevada P1232131 3 Wind River Mtns Sublette Wyoming P1232134 3 II P1232135 3 Yellowstone NP Gallatin Montana P1232138 3

Glacier NP Glacier U P1232139 3 N U P1232140 3 Nelson British Columbia P1236670 3 Cranbrook U P1236672 3 Livengood Alaska P1371690 3

Central P1372642 3 Richardson Hwy U P1372641 3 Colton Whitman Washington Whitmar 3 Logan Cn Cache Utah D1252 3 Colton Whitman Washington K35 (D2840) 3

Ashton Hill Fremont Idaho D28 15 3 Yellow jack Custer BB1600 3 Strevel Utah D2836 3 63

APPENDIX TABLE 3. Source of OTU of known diploid, triploid, auto- tetraploid, and allotetraploid forms of Pseudoroegneria spicata. (continued)

State/ Accession Ref er- Town County Province Number ence1

TRIPLOIDS

Cypress Hills Pk Saskatchewan Senn 6202 1 Wawawai Cn Whitman Washington DS 123 3

AUTOTETRAPLOIDS

Moscow Latah Idaho Senn 5933 1

Chewelah Stevens Washington Senn 5759 1 Winchester Lewis Idaho P7845 2,3,4 Sunnys ide Clearwater H T7681 4 Willow Springs Toole Utah K183 3

Wawawai Cn Whitman Washington DS120 3 Steptoe Butte U P1232123 3 H H P1232124 3 H 'I P1232125 3 14 II P1232126 3

H U U P1286198 3 Ainsworth British Columbia P1236669 3 Boswel 1 P1236671 3 Coleman Alberta P1236681 3

ALLOTETRAPLOIDS

Soap Lake Grant Washington P6025 2 Joel Latah Idaho P737 2,3,4 Lewiston Nez Perce Secar 2,3,4 Big Horn Wyoming T25668 3,4 New Meadows Adams Idaho K27(D2841) 3,4

Co 1 ton Whitman Washington K42(D2843) 3,4 Lewiston Nez Perce Idaho K32 4 Yakima Yakima Washington T40583 4 Joel Latah Idaho T40595 4 Silicot Asotin Washington T40599 4

Steptoe Cn Whitman II DS-104 3 U DS-105 3 II U DS-106 3 'I I, 'I DS-107 3 Snake River II DS-108 3 64

APPENDIX TABLE 3. Source of OTU of known diploid, triploid, auto- tetraploid, and allotetraploid forms of Pseudoroegneria spica ta.(continued)

Statef AccessionRefer- Town County Province Number ence1

Clarkston Whitman Washington DS-11O 3 Howell Creek Garfield I' DS-111 3

I' DS-112 3 Alpowa Creek U DS113 3 Asotin Asotin ' DS-114 3

U DS-115 3

Clarkston Whitman " DS-118 3 Wawawai Cn it H DS122 3 'I 'I DS-124 3 U Penawawa Cn " DS-129 3

H " DS-131 3 N " I' DS-132 3 H H U DS-133 3 Wawawai 'I 1 P1285272 3 0 " P1285273 3

Riggins Idaho Idaho K29(D2842) 3 Kahiotus Franklin Washington T37 194 4 Hooper Adams T37193 4

1 1 - see Bowden 1965 2 - see Hartung 1946 3 - unpublished data, D. R. Dewey, ARS, Logan, Utah 4 - results from this study 65

APPENDIX TABLE 4. Means (and standard deviations) of 55 morphological characters for S groups ofPseudoroegneriaand Elymus.

2X 4X auto 4X allo Old Thick- Character spicata spicata spicata World spike

Vecietative Characters

Cuim nodes 3.2 3.1 3.4 3.6 3.4 (0.9) (0.7) (0.7) (0.6) (0.5)

Leaf blade ribs 14.7 15.4 15.2 17.4 15.0 (3.7) (2.1) (2.7) (5.3) (1.7)

Cuim leaf blade 2.98 3.00 2.77 2.19 3.00 adaxjal vestiture (0.16) (0.00) (0.64) (1.17) (0.00) (O=none, 3=dense)

Presence long hairs 58 81 40 31 14 adaxial leaf blade (%)

Culm leaf blade 0.38 0.13 0.28 0.12 0.57 abaxial vestiture (0.58) (0.34) (0.50) (0.33) (0.85) (0=none, 3=dense)

Leaf blade width (nun) 3.3 3.6 3.3 4.6 2.9 (1.0) (0.7) (0.7) (1.4) (0.4)

Leaf blade length (mm) 17.2 19.0 17.5 20.1 19.1 (4.1) (4.8) (4.2) (4.9) (3.0)

Leaf sheath length (mm) 11.4 10.9 11.5 11.5 11.4 (2.3) (2.0) (2.6) (3.1) (2.1)

Leaf sheath 0.14 0.00 0.15 0.46 0.14 vestiture (0-3) (0.52) (0.00) (0.36) (0.86) (0.53)

Culm length (cm) 62.8 63.4 70.0 65.8 68.7 (11.1) (14.7) (13.3) (10.7) (13.2)

Culm diameter (mm) 1.5 1.6 1.6 1.7 1.5 (0.3) (0.3) (0.3) (0.3) (0.3)

Top internode 33.4 34.4 32.2 29.1 33.5 length (cm) (6.9) (7.9) (6.7) (6.5) (6.4)

Auricle length (mm) 0.35 0.55 0.36 0.45 0.46 (0.34) (0.42) (0.30) (0.40) (0.34)

Ligule length (mm) 0.50 0.48 0.60 0.60 0.72 (0.21) (0.19) (0.25) (0.33) (0.29) 66

APPENDIX TABLE 4. Means and standard deviations of 55 morphological characters for 5 groups of Pseudoroegneria and Elymus. (continued)

2X 4X auto 4X allo Old Thick- Character spicata spicata spicata World spike

Flag leaf blade 7.9 7.7 6.5 9.2 6.5 length (cm) (4.4) (3.2) (3.3) (4.8) (3.3)

Flag leaf blade 2.6 2.8 2.5 3.8 1.9 width (mm) (0.9) (0.7) (0.7) (1.4) (0.7)

Flag leaf sheath 17.9 17.0 17.5 17.4 19.9 length (cm) (3.7) (3.9) (3.7) (5.4) (3.8)

Spike Characters

Spike length (cm) 15.3 14.9 12.0 18.2 14.4 (3.7) (4.5) (3.3) (7.4) (2.8)

Spike nodes 11.5 11.3 12.5 11.8 13.6 (2.6) (2.6) (3.1) (3.5) (1.6)

Spikelets/cm spike 0.79 0.80 113 0.72 1.02 (0.28) (0.15) (0.29) (0.16) (0.21)

Third spike 13.7 12.3 8.5 14.9 9.1 internode (mm) (4.5) (2.8) (2.6) (5.4) (2.5)

First spike 20.3 18.8 12.1 18.8 13.2 internode (mm) (8.9) (6.3) (4.5) (7.9) (4.0)

Spikelet Characters

Spikelet length (mm) 18.5 18.5 15.2 18.8 15.4 (4.0) (3.5) (3.4) (2.9) (2.2)

Spikelet width (mm) 3.1 3.0 4.1 4.8 3.7 (0.9) (0.7) (1.2) (1.9) (1.2)

Florets/spikelet 7.6 7.5 7.4 8.2 6.8 (1.8) (1.4) (1.6) (1.5) (1.1)

Rachilla length (mm) 1.83 1.81 1.53 1.65 1.87 (0.24) (0.27) (0.28) (0.31) (0.21)

First Glume Characters

Length (mm) 7.9 7.7 6.5 9.2 6.5 (1.5) (1.2) (1.6) (1.8) (1.2) 67

APPENDIX TABLE 4. Means and standard deviations of 55 morphological characters for 5 groups ofPseudoroegneria and El ymus.(continued)

2X 4X auto 4X allo Old Thick- Character spicata spicata spicata World spike

Width (nun) 1.49 1.65 0.93 1.86 1.09 (0.33) (0.20) (0.34) (0.39) (0.14)

Shape (0=blunt, 1.59 1.52 2.28 1.42 2.21 2=very sharp) (0.52) (0.51) (0.60) (0.50) (0.58)

Presence awns (%) 19 25 15 0 36

Awn length (mm) 1.30 0.56 0.70 0.00 0.58

Adaxial vestiture (0-3) 1.88 2.55 1.92 2.27 1.93 (0.81) (0.77) (0.81) (0.83) (0.92)

Abaxial texture 0.84 0.97 1.19 0.46 0.86

(0=smooth, 2=very (0.46) (0.18) (0.44) (0.51) (0.53) scabrous)

Veins 4.5 4.6 3.3 4.8 3.5 (0.9) (0.7) (0.8) (1.0) (0.8)

Second Glume Characters

Length (mm) 9.2 8.8 7.3 10.5 7.2 (1.8) (1.4) (1.6) (2.1) (1.3)

Width (mm) 1.67 1.76 1.06 2.06 1.26 (0.36) (0.20) (0.39) (0.32) (0.20)

Shape (0-2) 1.49 1.45 2.30 1.42 2.21 (0.53) (0.51) (0.61) (0.50) (0.43)

Presence awns (%) 15 32 15 0 36

Awn length (mm) 1.36 0.75 0.64 0.00 0.64

Adaxial vestiture (0-3) 1.90 2.71 2.15 2.34 1.86 (0.80) (0.64) (0.82) (0.80) (0.77)

Abaxial texture (0-2) 0.92 0.97 1.17 0.46 0.86 (0.38) (0.18) (0.38) (0.51) (0.53)

Veins 5.0 5.3 3.9 5.4 3.9 (1.1) (0.7) (0.8) (1.0) (0.7) 68

APPENDIX TABLE 4. Means and standard deviations of 55 morphological characters for5groups ofPseudoroegneria and Elymus.(continued)

2X 4X auto 4X allo Old Thick- Character spicata spicata spicata World spike

First Lemma Characters

Length (mm) 10.5 10.6 9.6 11.4 9.6 (1.6) (1.3) (1.3) (1.6) (1.6)

Width (mm) 2.4 2.6 2.2 2.8 2.3 (0.3) (0.3) (0.3) (0.2) (0.3)

Presence awns (%) 62 90 94 12 50

Awn length (mm) 8.8 7.5 10.4 4.9 2.7

Adaxial vestiture (0-3) 1.17 1.42 1.32 1.19 1.50 (0.47) (0.50) (0.47) (0.40) (0.52)