Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Systematics of Saxifraga rufidula and related species from the Columbia River gorge to southwestern British.. Perkins, Walter Ethen 1978-03-01

You don't seem to have a PDF reader installed, try download the pdf

Item Metadata


831-UBC_1978_A1 P47.pdf [ 7.52MB ]
JSON: 831-1.0094491.json
JSON-LD: 831-1.0094491-ld.json
RDF/XML (Pretty): 831-1.0094491-rdf.xml
RDF/JSON: 831-1.0094491-rdf.json
Turtle: 831-1.0094491-turtle.txt
N-Triples: 831-1.0094491-rdf-ntriples.txt
Original Record: 831-1.0094491-source.json
Full Text

Full Text

SYSTEMATICS OF SJ^IFRJGA BOFIDULJ AND BELATED SPECIES FBCM THE COLOMBIA BIVEB GOBGE TO SOUTHSESTEBN BBITISH COIUUBIA By HALTER ETHEN PERKINS B.Sc. , University of Oklahoma, 1S73 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOB THE DEGREE OF DOCTOB OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES {Department of Botany) Se accept this thesis as conforming to the required standard THE UNIVERSITY OF BBITISH COLUMBIA October, 1978 (8) Walter Ethen Perkins, 1978 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the Head of my Department or by his representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. W. Ethen Perkins Department of Botany  The University of British Columbia 2075 Wesbrook Place Vancouver, Canada V6T 1W5 5 October, 1978 Research Supervisor: Dr. Ratherine I. Beamish. ABSTRACT In the Pacific Northwest, hybridization and polypoid variation have produced confusion in the relationships among Saxifraga rufidula (Small)James Macoun and its relatives. The entities from the Columbia Biver Gorge are particularly difficult to separate taxoncmically., Some authors have recognized several species from that area while others recognize one highly variable, widely distributed species with varietal components. The present study approached the systematic treatment of S. rufidula and related subspecif ic taxa of S. occidentalis Wats. with data from numerical studies, studies of meiosis, observations of artificial and natural hybrid individuals and populations, breeding system experiments, observations of pollinators, and ecological observations. Polyploid intermediates and plants with introgressant characteristics are shown to occur, many of which appear to be the result of hybridization with the S. integrifolia species complex. Sufficient correlations of morphological, ecological and geographic discontinuities exist to substantiate the treatment of S. rufidula. S. Occidentalls var. latipetiolata, S. occidentalis var., dent ata, and S. occidentalis var. occidentalis as a species. , According to the rules of nomenclature S. rufidula becomes S. aeguidentata. S. occidentalis var., latipetiolata becomes S. latipetiolata, S. occidentalis var. dentata becomes S. gormanii. and *>• occidentalis var., occidentalis becomes S. Occidentalis. ii Table Of Contents ABSTRACT . . . . .* . '/ i LIST OF TABLES iv LIST OF FIGURES . .:. ....... .. .... .. .......,>. y* w.-, .-.V.-. . V AC K NOWL E EG E H ENT S .................... ,<••• • • .... . ..... ix INTRODUCTION ......... .... .............,,....... / 1 MATERIALS AND METHODS .v.w. .V. .V. »V >v>./. .V. V. «^ . Y 6 Chromosomal Studies 13 Hybridization Studies ................................... 14 Numerical Studies ... .............. ...................... . 15 Breeding Systems ....... .... .... ...... •. .-......... 17 Habitat And Pollination Studies 18 RESULTS AND DISCUSSION 19 Morphology .....».........•Chromosomal Studies ......... ............................ 42 Hybridization .............................. ...... ....... 63 Numerical Studies ....................................... 87 Breeding System Observations • • • • 114 iii Bagging Tests •».......*.........* *.•.«.*.... •.. 114 Floral Observations 118 Geography, Habitat And Pollination Ecology Observations . 121 TAXONOMY . ...... ..................... ... ..V................ 130 Species Definition In This Complex- •..«-. i*>-..>V»«-*v»VV»-*V»»- .' 130 Key To The Species ...................................... 131 CONCLUSIONS 145 LI TEE AT 0 EE CITED 15APPENDIX .... ......,.i. ..... ...........• .. ... ..... • • 161 Insect Specimens ........................................ 161 iv LIST OF TABLES Table I, List of Collection Sites 9 Table II.List of Morphological Characters Used for Computer Analysis .. .v.. v... .v................ 16 Table III., Summary of Chromosome Number for Natural Populations ...................... .... ...... ............ 44 Table IV. Pollen Fertility in Artificial Hybrids ........76 Table V. Baw Principle Components for PCA Analysis ........99 Table VI., Bagging Tests for Automatic Self ing and Self Compatibility ..................,. 115 Table VII..Bagging Tests for Outcrossing and Apomixis .....116 Table VIII. Average Soil Depth Beneath Plants ............. 125 Table IX. Amounts of Soil Moisture Under Plants ...........126 Table X. Summary of Insect Visitors .....•....•.•........••128 V LIST OF FIGURES Figure 1 : Map of Collection Sites. .......................7 Figure 2 : Cut-away Flower and Fruit Drawings of J?» occidentalis var. latipetiolata. ••/.... .............. 20 Figure 3 : Cut-away Flower Drawings of S,.occidentalis var. occi den tali s, S. rufidula. s.--, occidentalis: Var. dentata, S. inteqrifolia var. claytoniifolia. S. intearifolia vari inteqrifolia, and S. inteqrifolia var. leptopetala . ................................... 23 Figure 4 : Sketch of Artificially Produced S.,tafidula x S. occidentalis Hybrid and .Representative Parental Plants. ... ................. ,•>• 27 Figure 5 : Sketch of Artificially Produced S. rufidula X S. Occident alls - var. dgntata Hybrid and Representative Parental Plants. ......... • 29 Figure 6 : Sketch of Artificially Produced S. rufidula x J. inteqrifolia Hybrid and Representative Parental Plants. ,....-»....... * .................................... 31 Figure 7 : Sketch of Artificially Produced S. rufidula x S. occidentalis var. latipetiolata. Hybrid and Representative Parental Plants. .................... y...33 Figure 8 : Sketch of Artificially Produced S. occidentalis x S. occidentalis var. dentata Hybrid aud Representative Parental Plants. .,......,....,.,........ 35 Figure 9 : Leaf Outlines of S. rufidula, S. occidental var. occidentalis. S. occidentalis var. latipetiojata. and S. integrifpjj-ia var. inteqrifolia vi v. . .... 38 Figure 10 : Leaf Outlines of Diploid and Tetraploid S. occ|dentalis var. dentata. ........................40 Figure 11 : Camera Lucida Drawings Showing Variation in Chromosome Number for One Population of S. rufidula in the Columbia River Gorge. ..........................47 Figure 12 : Camera Lucida Drawings Showing Variation in Chromosome Number for One Population of S. rufidula in the Columbia River Gorge 49 Figure 13 : Camera Lucida Drawings Showing Meiosis in Cryptic Hybrids and Normal Cells from Populations of S. rufidula in the Columbia River Gorge. .............. 51 Figure 14 : Camera Lucida Drawings of Meiotic Cells cf 5. rufidula and S. occidentalis var. occidentalis . .53 Figure 15 : Camera Lucida Drawings Showing Variation in Chromosome Number for S. occidentalis var. dentata. .55 Figure 16 : Camera Lucida Drawings Showing Heiotic Chromosome Numbers for S. occidentalis var. latipetiolata and S. inteqrifolia Re la t ives. «,...... y .59 Figure 17 : Camera Lucida Drawings of Meiotic Cells cf S. oregana and S. inteqrifolia var. clavtcniifclia Populations. ...........................• •.... .61 Figure 18 : Camera Lucida Drawings Showing Variation in Chromosome Number of Meiotic Figures in Artificial Hybrids. .. .... v.........v.... ..........................66 Figure 19 : Principal Components Analysis Axis 1 vs. 2 Depicting Artificial Hybrids and Parental Populations. .68 Figure 20 : Principal Components Analysis Axis 1 vs. 3 vii Depicting Artificial Hybrids and Parental Populations. .70 Figure 21 : Principal Components Analysis Axis 2 vs. 3 Depicting Artificial Hybrids and Parental Populations.,.72 Figure 22 ; Crossing Success Chart. ....................... 74 Figure 23 : Scatter Plot Depicting Character Introgression Between S. integrifolja var. clay ton i i f o1ia and Tetraploid S. occidentalis var. dentata . ........ ....80 Figure 24 : Scatter Plot Depicting Character Introgression Between S. inteqrifclia var. claytcniif olia and Sympatric S. rufidula Plants. .... ...... ................82 Figure 25 : Scatter Plot Depicting Character Introgression Between S. rufidula from the Columbia River Gorge and S. Occidentalls var. ocoidentalis . ..................84 Figure 26 : Dendrogram of Total Sample Osed in Clustering Program, UBC-CGBOUP. ..86 Figure 27 : Principal Components Analysis Axis 1 vs. 2 frco Samples of Natural Populations. ........................93 Figure 28 : Principal Components Analysis Axis 1 vs. 3 from Samples of Natural Populations. ..95 Figure 29 : Principal Components Analysis Axis 2 vs. 3 from Samples of Natural Populations. ........................97 Figure 30 : Hyperspace Plot of Grouping Relationships cf Canonical Variables in SDF Analysis Program and Point Cloud Bepresentation Of S. inteqrifclia var. leptop„etala . ......................................... 101 Figure 31 : Individual Points in Hyperspace Cloud Relationships about the Group Means for Sk integrifolja var. claytoniifclia and S. inteqrifolia var. viii leptepetala . • •.. .•.......•................. .......... 104 Figure 32 : Individual Points in Hyperspace Cloud Relationships about the Group Means for Hexaploid Hybrid Population and S. Occidentalis var. latipetiolata . ..106 Figure 33 : Individual Points in Hyperspace Cloud Relationships about the Group Means for S. cccidentalis var. dentata and S. occidentalis var. occidentalis . .................. ...... ......... ... .......108 Figure 34 : Individual Points in Hyperspace Cloud Relationships about the Group Means for Columbia River Gorge S. inteqrifclia and Vancouver Island S. inteqrifolia . ......................... ... .......... 110 Figure 35 : Individual Points in Hyperspace Cloud Relationships about the Group Means for S. rufidula from Vancouver Island and the Olympic Mountains and S. rufidula from the Columbia River Gorge with Artificial Hybrids. ..........................,.«>......112 Figure 36 : Speculative Polyploid Relationships in •S. rufidula and Related Taxa. ......................... 151 Figure 37 : Distribution Map of S. gormanii. S» occidentalism and S. occidentalis x S. inteqrifolia Hybrid Population. ... •........... .... ..... ............. 162 Fiqure 38 : Distribution Map of S. rufidula and S. latipetiolata. ..... .y«. ..... . • •>•;#•• .. «... ........... .164 ix ACKNOWLEDGEMENTS "The journey need not be alone at all moments. He can and do spark one another, and carry each other on." Paulus Berensohn To many who have helped along the way simple thanks are insufficient, but, alas, they are all I have to demonstrate my appreciation. My wife, Mary Dee, knows all too well the kinds of teamwork and effort which she has applied to this paper. Her capacities as field assistant, data recorder, computer operator, illustrator, friend and mother were invaluable in the completion of this labor of love. My daughter, Rachel, provided intangible joys and much needed diversions when she arrived during the final phases of this work. Kay Beamish has been a most helpful advisor and respected friend throughout this study. Funds for travel, equipment and expenses were provided by an N.R.C, Grant (No. ,67-6901) to Dr. Beamish, a O.B.C. , Graduate Fellowship, Careers »75 Stipend, and a U.B.C. Student Travel Stipend to the author. The members of my advisory committee deserve thanks for their thoughtful suggestions along the way and careful reading of the final manuscripts, The flower close-up drawings were done by my good friend, Joan Miller. Gary Bradfield generously provided his statistical expertise, computer program, and time in reading certain sections of this paper. Johann van Seenen identified the insect specimens and John Pinder-Moss assisted in numerous herbarium duties and operational aspects of this study. So many others gave various kinds of support and encouragement alcng the way it is difficult to thank each one here. Of the non-human cohorts, the UBC computer was cf great usefulness for many aspects of this study. But mostly there are the plants. X whose spirits live on where the wild things grow. xi 1 Sort of a Song Let the snake wait under his weed and the writing be of words, slow and guick, sharp to strike, guiet to wait, sleepless. Through metaphor to reconcile the people and the stones. Compose. (No ideas but in things) Invent! Saxifrage is my flower that splits the rocks. -William Carlos Williams (1883-1963) 1 INTBODOCTION The genus Saxifraga L. is composed of herbaceous, mostly perennial, plants which are found in arctic, alpine and rocky places in temperate regions. The genus is circumboreal and restricted mostly to the Northern Hemisphere. The name, Saxifraga, comes from the Latin words, sajsi meaning rock and fragere meaning to break or fragment, and is probably a reference to the habitats which many members occupy. A less likely derivation could be attributed to the folk use of some members of the genus for the treatment of kidney stones (Spongberg 1972). Ethnobotanical uses of Saxi,f raga species are as a leaf exudate for treatment of superficial wounds and boils (Hunan 1974), as a root extract used for the treatment of a wide range of disorders from a child's teething pains to relief of dysentery (Watt 1972), as alpine rock garden plants, and as the hanging basket houseplant, S. sarmentosa Schrceder, known as strawberry begonia, strawberry geranium, or mother-of-thousands. Although several early botanists contributed greatly to the knowledge of the genus (Linneaus 1753, Don 1822, Hooker 1833), the comprehensive monograph of Engler and Irmscher (1916) brought together the earlier taxonomic information and today constitutes the standard reference for this group of plants. In North America the section Boraphila of Engler and Irmscher (1916) has two temperate centers of species diversity, one in the Eastern Appalachian Region and the other in the Rocky Mountain Region of the Pacific Northwest (Spongberg 1972). Both areas were strongly influenced by recent Pleistocene glaciaticns 2 and the evolutionary history of the plants in those areas is, in many respects, closely linked with the history of glacial advance and retreat. This is especially evident in the rapidly evolving species complexes of those areas. In such complexes single morphological characteristics which separate one entity from its close relatives are often difficult to find., The name, rufidula, was first used to describe a species of the genus Micranthes by Small and Hydberg (1905). It was subsequently treated as a part of the genus Saxifraga by James Macoun (1906) whose father, John, collected the type material from Mt. Finlayson on Vancouver Island, British Columbia in 1887. Persistent confusion has surrounded the relationships among Saxifraga rufidula (Small) James Macoun and its close Pacific Northwest relatives in the section Boraphila subsection Kjvali-virginiensis (Engler and Irmscher, 1916) up to the present time. Early monographic treatments dealt with the variability extant within this closely related group by recognizing a large number of specific entities (Small and Bydberg 1905, Johnson 1S23). More recently, these have been reduced to two or three broadly defined species complexes with subspecific or varietal components (Bacigalupi 1944, Hitchcock et al. 1961, Hitchcock and Cronguist 1973, Krause and Beamish 1972,1973). The channeling of the Columbia River through its gorge has permitted a zone of contact between drier interior flcristic elements and more mesic coastal species. The intimate contact of otherwise generally ecologically and geographically isolated plants in situations where environmentally intermediate or 3 perhaps unique micro-habitats are available apparently has led to complicated patterns of variability and evolution within several groups of plants, including the Clay_tonia perfpliata polyploid complex (Miller 1976), and the Sisvrinchium §31 gg|?*i?-53ffl- j-dahoense- littorale duodecaploid species group (Henderson 1976) . Several authors have commented on the extent of morphological intermediary between S. rufidula and S. occidentalis S.,Wats. Bacigalupi, in Shram* s flora (1944), tentatively divided S. occidentalis into two subspecific taxa; subspecies occidentalis and rufidula. Hitchcock et al. (1961), and Hitchcock and Cronguist (1S73) treated S. rufidula as a variety of S. occidentalis and considered variety rufidula transitional to variety dentata (Engl, and Irmsch.) C.L. Hitchcock and variety allenii (Small)C.L. Hitchcock in the Columbia Biver Gorge and adjacent Oregon. , Hitchcock also included variety idahoensis (Piper)C.L.Hitchcock in S. occidentalis and described a new variety, latipetiolata C.L.Hitchcock. Saxif raga rufi dnla was confirmed by Krause and Beamish (1973) as a species and S, occidentalis was considered as an extremely variable entity without subspecific taxa. They noted that the extent of intergradation bet wee n S. r ufidula and S. occidentalis in the Columbia Biver Gorge was largely unexplored. The entity idahoensis had been treated by Krause and Beamish (1972) as a subspecies of S. marshallii but they recognized that S. marshallii subsp. idahoensis has some introgressant characteristics of ,S. occidentalis. Elvander (1975) confirmed the classification by Hitchcock et j|l. (1961) , 4 and Hitchcock and Cronguist (1973) of S. occidentalis varieties idahoensis, rufidula. and latipetiolata. He considered the varieties dentata and alienii to lack morphological distinctions and vary continuously with variety occidentalis and conseguently, he treated them as synonyms of variety occidentalis. None of these previous systematic investigations have addressed the problem of the confusion that exists among the relatives of S. ruf idula in the Columbia Biver Gorge. Eeamish (1961,1967) and Krause and Beamish (1972,1973) focused their studies on relationships mainly in British Columbia although they reported several chromosome counts of the material from the Columbia Biver Gorge and other areas of Oregon and Washington. Elvander (1975) based his study cn numerical, cytological, and chromatographic evidence in plants largely from Idaho and Rontana. He briefly discussed S. rufidula. S. occidentalis var. d£JLt§il# and S. occidentalis var. latipetiolatabut did not include data from populations of those entities in his basic study. The relationships among S. rufidula. S. occidentalis var. occidentalis. S,,occidentalis var. dentata and S. occidentalis var. latipetiolata in the Columbia Biver Gorge and other areas extending northward into Southwestern B.C. are dealt with in the present paper. Correlations of hybridization studies and ecological observations with cytological studies and morphological analyses based on three different numerical approaches are presented in an effort to provide a more natural classification. Since S. rufidulaand its close relatives are 5 often found growing sympatrically with members of S. Integrif olia (sect. Integrifcliae of Engler and Irmscher 1916), the varieties inteqrifolia. cla ytoniifolia, and leptopetala are also included as potential sources of hybridization and character introgression in the studies presented here. Several taxa peripheral to the main thrust of this study may have some close relationships with the taxa discussed in this paper. Evidence that S. oreqana has close ties to S.•..occidentalis var. latipetiolata is brought forth aisd as a result it is included in the key to species. A systematic study of S. integrifolia and S. oreqana sensu lato is currently in progress at the University of Uashington (Elvander, PhD dissertation, in preparation)., Saxif raga •• occidentalis var. dentata may introgress to S. marshallii subsp. marshallii. Other taxa mentioned here include S. marshallii subsp. idahoensis, S. reflexa, S. nivalis, S« califcrnica* J» integrifolia var. columbiana. S. rhomboidea. S, ferruginea and a group of Eastern North American species. Saxifraja isiiflfifsiia var. columbiana is treated here as equivalent to S. j,ntegrjfclja var., leptppetala. The present paper uses the terminology of Hitchcock for S. .,occidentalis but 5. rufidula and S. marshallii subsp. , idahpensis are treated as recognized by Krause and Beamish (1972, 1973), and ,S. occidentalis var. allenii as synonymous with var. Proposed changes in the classification and revised comecclature of the taxa studied appear in the Taxonomy and Conclusions sections of this paper. 6 MATERIALS AMD METHODS Live plants, pressed specimens, flower buds and several soil samples were collected from populations from the mouth of the Columbia River east to beyond the Columbia River Gorge. Some collections were also made in areas extending north and south in the lower Columbia River area and adjacent Coast Range. Populations from the Olympic Mountains, Southeastern Vancouver Island, and Cascades in Washington and the Southern Mainland of British Columbia (Fig. 1, Table I) were also sampled. Many collections came from localities with two sympatric taxa. Fifteeen to twenty rosettes from a large number of the populations sampled were transplanted to ccldframes filled with a peat-vermiculite bedding mixture which were located at the University of British Columbia. Mass collections of buds in meiosis were sampled. Buds from individual plants were collected and a large number of plants were pressed as voucher material. Specimens are on deposit at the university of British Columbia, Specimens studied for morphological comparison, classification, annotation, and geographic distribution included a total of 1,507 herbarium sheets representing about 5,000 plants from the following herbaria: Brigham Young University (BEY), University of Minnesota (MIN), U.S. National Herbarium (NA) ,University of Oregon (ORE) , Oregon State University (OSC) , University of British Columbia (UBC), University of California at Berkely (UC), British Columbia Provincial Museum (V)w Washington State University (MS) , and University of Washington (WTU). Several hundred additional herbarium sheets were also 7 Figure 1: Collection sites in the Pacific Northwest for the present study. One collection from Berthoud Pass, Colorado, is not shown. 9 • Table L. 7 Collections for Study. Collection Tjaxpn Code Number location S. rofidula 616 Viento, rest area on Interstate 8, 0.9 mi w. of Viento St. Pk., Hood Biver Co., Ore. 608 Troutdale, 1.3 mi s. of Sandy B. Bridge, Multnomah Co., Ore. 610 Yeon Pk., ca. 1.5 mi along trail to McCord Falls, Multnomah Co. Ore. , 612A Mayer Pk., 0.8 mi n. of Mayer St. Pk., Wasco Co., Ore. 613A The Dalles, 2. 0 mi w. of Chynoweth Cr. on Old U.S. , Hwy. 30, Wasco Co., Ore. 637 Mosier, 1.0mi e. cf Hosier on Old Columbia 8. Hwy., Wasco Co., Ore, 673 Lake crescent, 8.5 mi e. of Fairholm on U.S. Hwy.,101, Clallam Co., Wash. . 675A Mt. Pleasant, 1.9 mi e. of the Clark-Skamania Co. bdry. on St, Hwy. 14, Clark Co. , Wash. 647A Washougal,1.0 mi w. of Clark-Skamania Co. bdry. on St. Hwy. 14, Clark Co., Hash. 76-1 Marmot Pass, ca. 5 mi s. of Camp Mystery, s.e. of Marmot Pass, Jefferson Co,, Wash. 617A Bingen, 5.6 mi e. of Bingen, on St. Hwy 14, Klickitat Co., Wash. 669A Bingen Lk., 7.4 mi e. of w. bdry., of Klickatat Co. on St. Hwy. 14, Wash. 672 Cock, 9.8 mi w, of e. bdry, of Skamania Co. on St. Hwy. 14, Bash., 6 18A Skamania Co., 3.5 mi e. of Clark-Skamania Co. bdry. on St. Hwy. 1 Wash, 645A Collins, 0.9 ni e. of Collins Depot Bd. on St. Hwy. 14, Skamania Co., Wash. NMO-8 Nanaimo, ca, 5 mi's, of Nanaimo cn White Rapids Rd., V.I., B.C. 10 Table It JconJMt) Taxon S. occidentalis var. denata Collection Code Number NOOS-B UCL-R 623A 624B 627A 629A 606 630 607 S. occidentalis 629C var. latipetiolata S. occidentalis 605 var. occidentalis BMR 626 666 COEN 687 S. inteqrifolia 641 var. claytoniifolia Nanoose Hill, ca. 15 mi n. of Nanaimo on Nanoose Hill, V.I., B.C. Upper Campbell Lk., 7.4 mi w. of Campbell Lake Bridge, V.I., B.C Sooke, s. side of Sooke B. at Sooke Potholes Pk., V.I.,B.C.. Hill Hill, Capital Distr. Pk., Victoria, V.I., B.C. Mt. Finlayson, Goldstream Prov. Pk., V.I., B.C. Saddle Mt., cn trail to summit near westernmost Peak, Clatsop Co., Ore. Delena, Beaver Creek Falls, 2.8 mi w. of Delena, Columbia Co. Ore, Tillamook Co., 2.0 mi w. of e. bdry. of Tillamook Co. cn St, Hwy. 6, Ore. Kalama B., 3 mi e. of Interstate 5 on Kalama B. Bd., Cowlitz Co., Wash. Saddle Mt., on trail to summit near top, Clatsop Co. , Ore. Chehalis 8. , (hybrid) , 3.3 mi w. of Littel, Lewis Co. Wash. , Mt. Baker Natl, forest. Yellow Aster Meadows, Whatcom Co., Wash, Yale, ca, 1.5 mi e. and n. of Yale below Can. Hwy. 1, B.C. Liumchen Bidge, s. of Sardis, B » C» Cornwall Lookout, se. of Hat Creek, B.C. Bootanie Valley, Skwaha Mt. area n. of Lytton, B.C. Giliam Co., 1.7 mi e. of w, bdry. Giliam Co. on Interstate 80 n., Ore. 11 Table I. (con't) Collection HiLSSLE Code Number Location 609 Trcutdale, 1.3 mi s. of Sandy 8. Bridge, Multnomah Co., Ore. 611 Bowena, 2,4 mi w. of Bowena overpass on Interstate 80-N,, Hasco co., Ore, 612B Mayer Pk., 0.8mi n. of Mayer St. Pk., Wasco Co., Ore. 613B The Dalles, 2.0 mi w, of Chynoweth Cr. on Old U.S. Hwy. 30, Wasco Co. Ore. 667 Biggs, 1.25 mi e. of jet. with U,S Hwy. 97 on U.S. Hwy. 30, Sherman Co., Ore. 692 Stephen's Pass, 11 mi e. of Stephen's Pass Ski Lodge on U.S. Hwy. 2, Chelan Co., Hash. 675B Mti Pleasant, 1.9 mi e. of the Clark-Skamania Co. bdry. on St.Hwy. 14, Clark Co., lash. 668 Lyle, 1.6 mi n. of jet, with St. Hwy. 142, Klickitat Co., Wash. 669B Bingen Lk. , 7.4 mi e. of w. bdry. of Klickatat Co.:on St. Hwy. 14, Wash. ., 642 Benchmark, U.S.G.S. V425,1S68, 5.2 mi e. of The Dalles bridge, Klickatat Co., Wash. 617B Bingen, 5.6 mi e. of Bingen, on St. Hwy 14, Klickitat Co., Bash. 619 Clark Co. line, bdry. with Skamania Co. on St.,Hwy, 14, Sash. 5A integrifolia 671 Grizzly Lk., about 0.5 mi var. integrifc^ia • downstream from Grizzly Lk. Siskiyou Co., California MIMA Mima Mounds, w. of Little Bock on Bd. to Mima, Thurston Co., Wash. 6 23B Sooke, s. side of Sooke B. at Potholes Pk., V.I., B.C. 624B Mill Hill, Mill Hill Capital Distr. Pk., Victoria, V.I.,B.C. 627B Mt. Finlaysoh, Goldstream Prev, Pk. , V.I. , B.C. 12 Table Ja. (con* t) Ccllecticn Tax on Code Number Location NJ30-B Nanaimo, ca, 5 mi s, of Nanaimo on White Bapids Bd. , V. I. , E. C, NOOS-B Nanoose Hill, ca. 15 mi n. of Nanaimo on Nanoose Hill, V.I., B.C. 625 Yale, ca. 1.5 mi e. and n. of Yale below Can. Hwy 1, B.C., 617B Bingen, 5.6 mi e. of Bingen on St. Hwy. 14, Klickatat Co., Wash. 78-1 Harrison Lk., 1.5 mi w. of Harrison, Hot Springs, B.C. 682 Elk Balls, below Campbell Lk. Dam, V.I., B.C. S. integrifolia 648 Princeton, off Hwy. e. of var. leptopetala Princeton, road tc golf course, B. C. S*. orjejgajja BEBH Berthoud Pass, Bdry. of Grand Co. Colorado 13 inspected from material on loan to the University of Washington., Chrcmosomal Studies Squashes of anther tissue nere made by first fixing bud material in Carnoyfs 3:1,absolute ethanol:glacial acetic acid fluid. Buds were then stained using Snow*s (1963) bulk method and warmed in a 40 C oven for about 24 hours. Slide preparations were preserved in Hoyer*s permanent medium (filexopculcs and Beneke 1952) . Pollen stainability of live, preserved and dried flowers was measured by counting up to 200 pollen grains which had been stained and mounted in lactophenol-aniline blue stain (Sass 1958). Large, dark-staining grains were considered fertile. Preliminary investigations of pollen grain size confirmed Sokolovskaya•s (1958) reports that pollen size does not appear to be correlated with chromosome number and further investigation was discontinued. 14 HYbridlzatiGjp Studies Plants used as female parents were emasculated, bagged with water-resistant parchment bags prior to flowering and artificially cross-pollinated under a dissecting microscope. Only crosses involving simultaneously blooming plants were successful. Stigmatic receptivity was estimated from the onset of a slight watery appearance and papillate condition of the stigmatic surface. Increased stainability using lactophenol-aniline blue stain (Sass 1958) also indicated stigmatic receptivity. Whole anthers were used tc transfer pollen masses to the stigmas. Samples of hybrid and outcrossed seed were planted in a well-drained sandy soil mixture in early December in rows beneath about 1-2mm of sand and a 1cm layer of pea-sized gravel in large-mouthed pots set in a shallow pan of water. Plantings were placed outside under ambient temperature and light conditions and misted periodically to maintain a moist environment. By April seedlings of nearly every cross had germinated and later were transplanted to sandy propagation beds where they were allowed to grow for one season. In the spring of the second season seedlings were transplanted into pots. As they flowered, samples were examined for their morphology and behaviour in meiosis, then preserved as pressed specimens. 15 Numerical Studies Forty-three quantitative characters or character combinations (Table II) were measured on 263 live, flowering plants (from natural populations and artifically produced hybrids) for numerical taxonomic treatments. Periodic infestations of rust, aphids, mildews, slugs and root weevils not only increased plant mortality but occasionally altered the developmental morphology of some growing plants. Such individuals were avoided wherever possible in the study but a certain amount of spurious variation may be the result of responses to minor infestation or subsequent pesticide treatment. Population samples consisted of 8-10 plants chosen at random from well-established transplantations grown in cold frames. Leaf measurements were taken of a mature leaf of each plant after it was carefully removed from the outer rosette whorl and pressed. Flower measurements were standardized by choosing flowers as close as possible to the sequential midpoint of anther dehiscence and by trying to obtain flowers from equivalent positions in the inflorescence. Linear variables were converted to metric values, counted variables were transformed using the square root transformation and the resultant data matrix was subjected to three different available computer programs. The hierarchical clustering analysis program which was used is called UBC-CGBOUP, available at the University of British Columbia Computing Center. The alogarithm used is that of Hard (1963) and the coding is modified from Veldman (1967). ft Principal Components analysis (PCA) program called PBINCOMPS 16 Table IIX Characters used in Numerical Studies No. Character 1. petal In. 3. sepal In, 5. gland wd, 7, style ht. 9. perianth wd. 11. filament wd. at bottom 13, hair no. at leaf tip 15, hair no, on pedicel 17, angle of lowest bract to stem 19, hair no. at lowest branch 21. hair no. on petiole 23. lowest bract ln./wd. ratio 25. scape diam. at lowest branch 27. plant ht. 29. inflorescence ln./wd. ratio 31. inflorescence In. from widest part to top 33. leaf ln./wd. ratio 35. hair In. petiole 37. angle leaf shoulder 39. petiole ln./wd. ratio 41. style In. • gland ht./ gynoecium In. 43, leaf In./petiole In. ratio No. Character 2. petal ln./wd. ratio 4. sepal ln./wd. ratio 6. gland ln./wd. ratio 8. filament wd. at middle/wd. at bottom ratio 10. gynoecium ht. 12. pedicel In. 14.,hair In. leaf tip 16.,angle of lowest inflorescence branch 18. hair In. at lowest branch 20. hair no. at scape bottom 22. lowest bract In. 24. teeth no.; on lowest branch 26. inflorescence angle 28. filament In. 30. plant In./In. to lowest branch ratio 32. leaf In. 34. teeth no. on leaf 36. angle of leaf tip 38. petiole wd. 40. In. of tooth at leaf tip 42. gland wd. sguared * filament In. sguared/ (style In. + gland In.) sguared. abbreviations: ht.=h eight, ln.=length, no.-number, wd.=width 17 (available from Brad field. University of British Columbia, Department of Botany) was used to obtain a standardized correlation matrix which was then operated upon to form an eigenvector matrix. First vs. second, first vs. third, and second vs..third eigenvector plots were also produced. Ten groups including hybrid and intermediate groups were assigned and analysed using a stepwise discriminate function analysis program (BMD07M, Dixon 1970) to determine categorizing functions, important discriminating variables and group hyperspace relationships. Breeding Systems Tests for autogamous pollination were carried out by; 1) bagging flower buds for the flowering season and 2) hand pollinating emasculated, bagged flowers with pollen from other flowers on the same plant. Tests for asexual seed production which also served as procedural controls for hand pollination experiments were conducted by allowing emasculated, bagged buds to mature within the pollination bags. Undehisced seed and unfertilized ovules were recorded in a subsample of each inflcresence. Seed number was estimated where high numbers of seeds were produced. The number of seeds in immature carpels was counted from several plants of each taxon. Repeated counts on a single inflorescence were grouped according to the determinate order and position of the flowers on the inflorescence branches. 18 Habitat find Pollination Studies Observations of insect behavior on Sax.jfraqa flowers were taken over a period of 21 hours on 5 days at 3 locations. The locations included two on Vancouver Island where S. rufidula and S. inteqrifolia are sympatric and a S. occidentalis population at Yale, British Columbia. Observations of behaviour included the time spent foraging (for nectar and/or pollen) or resting, the numbers of and distances between flowers and inflorescences visited, and if possible, the visitors' preferences for flowers in various stages of anthesis. Insects observed visiting Sasifraga flowers were collected and identified tc assist in identification of similar uncaptured visitors. Pollen was washed or scraped from the collected insects to determine the amount of gaxifraga pollen present. Soil depth to rock substrate (or in rare cases to a oaximum depth of 30 cm) was measured from beneath 15 to 30 plants chosen at random and representing populations on Vancouver Island, the Lower Mainland of British Columbia, and a transect of stations along the Columbia Biver Gorge. About 20 cc samples of soil were taken from beneath 15 plants chosen at random from 7 populations |6 cn Vancouver Island, 1 at Yale, British Columbia), Three locations received repeated sampling over the course of one flowering season. The samples from each population were homogenized, weighed and dried in a 60 C vacuum oven for 24 hours before dry weight measurements were taken. 19 RBSCLTS ftMP DISCUSSIGN Horphological features which distinguish between the taxa include both floral and vegetative characteristics in most taxonomic treatments, /The position of the ovary has been used as an important key characteristic, ft developmental feature which has created problems for taxcnomists is the disparity between the degree of ovary inferiority at early anthesis (judged by midpoint of anther dehiscence) and later at maturity of the fruit. For instance, in S. occidentalis var, latipetiolata development proceeds from a greater than half inferior ovary as in S. inteqrifolia and S, oregana to an almost completely superior ovary in fruit (Fig.2ft,B). Evidently the difference between flowering and fruiting conditions is difficult to detect cn pressed specimens and has led to some ambiguities in previous taxonomic treatments if an early fruiting condition where petals are persistent is interpreted as anthesis. Pressing and drying processes may distort precise ovary relationships. Even in live material that is standardized by using midpoint of anther dehiscence as a precise event in anthesis, variability in stylar elongation, receptacle width and gland positioning create difficulties in guantitive measurement. Nonetheless, as a characteristic which separates the major species complexes, ovary position is useful provided that anthesis is clearly defined ty discrete events such as anther 20 Figure 2: Drawing of S. occidentalis var, latipeticlata flower (A) and fruit (£). Flower is at midpoint of anther dehiscence. Sepals, nearest petals, filaments and a pie-shaped segment of ovary are removed to show the position of the ovary. Note the differences in ovary position and gland structure between flower and fruit. .21 22 dehiscence. On herbarium material it is often possible to interpret flowers which are past anther dehiscence as being those from which pollen has been removed by pollinators. The breadth and shape of the nectar gland are useful characteristics in separating some species and species complexes. ., • In S. rufidula and many populations of S« Occidentalis var. occidentalis it is a small ringlike band encircling the lower portion of the ovary (Fig. 3,A,B). The gland and its secretions are hidden at the base of the appressed petals and filaments in many of these plants. In S. occidentalis var. dentata (Fig. 3,C), S. occidentalis var. la t i pe tie1at a (Fig. 2A), and members of the S. integrifolia - oregana- complex (Fig. 3,D-F) the gland at anthesis is an obconic disc which covers a considerable portion of the top of the ovary and exudes nectar in diffuse, glistening droplets. Filament shape varies widely among relatives of S. occidentalis var., cccidentalis (Fig. 3,A-C). In S. rufidula and 5. occidentalis var., dentata filaments are linear or subulate. A few plants from high mountain populaticrs of S» rufidula on Vancouver Island and the Olympic Mountains have slightly clavate filaments., In S. occidentalis var. occidentalis the filaments are usually at least scuewhat clavate but on some pressed specimens shrinkage in drying produces apparently linear filaments. All members cf the S. inteqrifolia -oregana complex have linear or subulate filaments (Fig. 3D-F) . It is interesting that in other relatives of S. Occidentaljs such as S. marshallii,, which have yellow or green petal spots the filaments are broadly clavate and may have ure 3: Drawing of S. occidentalis var., occidentalis (A), £J3JLi£LsIa (B) # S. occidentalis var., dentata (C), §* inteqrifolia var. claytoniifolia (D) , S. inteqrifolia var. inteqrifolia IE). S. inteqrifolia var. lggtoeetala (F), flowers at midpoint of anther dehiscence. Nearest sepals, petals, filaments and a pie-shaped seqment of the ovary wall are removed to show the position of the ovary. Note differences in nectar qland features, ovary position, filament structure and petal shape in the flowers. The qland is the swollen structure immediately above the bases of the filaments. 25 a special function in pollinator attraction. Correlations of these features as well as gland characters with diverse pollination strategies cculd perhaps provide an interesting evolutionary story, , Petal size, color, and shape vary considerably among the taxa (Fig. 2,3). Saxifraga occidentalis var. occidentalis differs from its relatives in this study by having usually elliptic petals which often are somewhat narrowed into a clawlike base. Certain dwarfed alpine forms of S. occidentalis var. occidentalis as well as S. rufidula may be nearly apetalous with darkly anthocyanic inflorescences and flower parts. Saxifr a g_a rufidula. 5. occidentalis var. dentata and S. occidentalis var. latipet j.olat a usually have ovate petals with broad bases. Varieties of S. intggx^fojLia range from apetalous or with small greenish-petaled forms to large, white-petaled forms, Saxifraga rufidula has a flat-topped or broadly convex, diffuse-flowered inflorescence (Fig. 4, ft, 5#B) . Some Columbia Biver Gorge plants have broadly obconic inflorescences. In contrast, S. occidentalis var. occidentalis (Fig. 4,B), S. occidentalis var. latipetiolata(Fig. 7.ft). S. occidentalis var. dentata (Fig. 5,ft) and most varieties of S. integrifolia (Fig. 6,B) usually have conic or interrupted-conic inflorescences which range in flower density from open to congested, Montane forms of S, occidentalis var, occidentalis as well as S. integrifolia var. apetala are usually few-flowered, dense, capitate panicles. Flower number ranges per inflorescence from few (4-42) in S. ...-rufidula, especially those 26 on Vancouver Island and the Olympic Mountains, to several (13-81) in S. occidentalis var., occidentalis, S. occidentalis var. dentata and •• ~s. inteqrifolia var. inteqrif clia. Saxifraga occidentalis var. latipetiolata and S. inteqrifolia var. clavtoniifolia have a large number of flowers, usually exceeding 75. Saxifraga occidentalis var. dentata and * j§. inteqrifolia var. clavtoniifolia have fine, brittle, vertically penetrating networks of rhizoires. Such networks may also be present in S. jar shal Iii subs p. mar shall ii and S. inteqrifolia var. leptopetala. The other taxa have short, stout, horizontal rhizomes which do not form deep networks, althouqh rosette replacement from short branches and branchlike basal bulblets appears to be a common means of veqetative qrowth in this qroup cf plants. Many features of the leaves in most taxa vary within a rosette. The terminal leaves are usually more pubescent with fewer teeth and are much smaller in size than the leaves from the outer whorl of the rosette. The leaves may also demonstrate a plastic response to shaded conditions in which they become elcnqate and have a qreater surface area. Seme leaf characteristics are of taxonomic siqnifieance. S. occidentalis var. 1atipetiolata, as the name implies, has broad, short, rather indistinct petioles (Fiq. 9D). In that feature it resembles S. oreqana and differs from the other taxa which have distinctly narrowed, evident petioles (Fiq. 9,A-C,E). The transition from blade to petiole is most abrupt in S, inteqrifolia var.. clavtoniifolia and S. occidentalis var. 27 Figure 4: Habit sketch cf artificial hybrids and representative parental plants. Tetraploid S. rufidula (A), tetraploid S. occidentalis var. occidentalis IB), and an artificially produced S. rufidula x occidentalis var. occidentalis F1 hybrid plant (C)., Magnification x 2/3. 28 29 Figure 5: Habit sketch of artificial hybrids and representative parental plants. Tetraploid S. occidentalis var. , dentata {&), diploid S. rufiduja (B) , and an artificially produced S. rufidula x S. occidentalis var. dentata F1 hybrid plant (C). Magnification x 2/3. 31 Figure 6: Habit sketch of artificial hybrids and representative parental plants. Diploid S. rufidula-(fl). tetraploid S. inteqrifolia (B), and an artificially produced S. rufidula x S. integrifolia F1 hybrid plant (C) . Magnification x 2/3. . 32 gure 7: Habit sketch of artificial hybrids and representative parental plants. N=38 S. occidentalis var. 1 atipetiolata (A) , tetraploid S. rufidula (B) , and an artificially produced S. rufidula x S. occidentalis var. latipetiolata FI hybrid plant <C). Magnification x 2/3. 34 35 Figure 8: Habit sketch of artificial hybrids and representative parental plants., Tetraploid S. occidentalis var. dentata (A), tetraplcid S. occidentalis (B) , and an artificially produced 5* occidentalis x S. occidentalis var. dentata F1 hybrid plant (C). Magnification x 2/3. 36 37 dentata (Fig. 10) and usually more gradual for members cf the other groups (Fig.9). The rather distinct rounded or somewhat squared teeth of S. rufidula and S. occidentalis make the blade-petiole transition evident but the actual angle cf the blade base is usually rather obtuse in these taxa (Fig. 10). Marginal teeth of Si rufidula tend to be deeply rounded and acute whereas S. occidentalis var. occidentalis teeth are usually right-angled or obtuse. However. S. rufidulaplants from the Columbia Biver Gorge are apparently introgressant for this character (Fig. 9,A-C). Leaves of S. rufidula are usually reddish tinged on the lower surfaces while those of S. occidentalis var., occidentalis and S. occidentalis var. dentata vary from green to reddish. Saxifragaoccidenta 1 is var. latipetiolata leaves are usually light green on both surfaces, a feature resembling members of the S. integrifolia-oreqana group. Seme morphological features which require a broader survey and further attention include: leaf thickness in cross section, stcmate characteristics, fine structural differences in seed coat sculpturing, and papillcsity of petals, filaments, and nectar glands. Although variation was observed in these features, definite correlations with most groups were not possible on the basis of limited material and data concerning the extent of variability in ether members of Saxifraga. These characteristics may be of considerable usefulness in definition of subsections within the genus. It was noted that S. inteqrifclia var. leptopetala apparently is unique amonq the observed taxa in that its petals are without adaxial 38 Figure 9: Leaf outline drawing shewing variability in shape and margin characteristics in samples of diploid S. rufidula (A), tetraploid S. occj.qentalis var. occidentalis (B) , tetraploid and hexaploid 5. rufidula in the Columbia Biver Gorge (C), n=38 S. occidentalis var. latipetiolata (D), and tetraploid S. integrifolia var. integrifolia (E) leaves. Each sample is from one population. Each leaf is taken from the lower whorl of leaves from a separate plant. Line is 5 cm long. Note differences in petiole length and width, petiole-blade transition and dentition. increased variability of Columbia Biver Gorge polyploid S. rufidula plants and resemblance to S. occidentalis is also apparent., 39 40 Figure 10: Leaf outline drawing showing variability in shape and dentition in samples of diploid (A) and tetraploid IB) S. occidentalis var* dentata. Samples are from four different populations. Each leaf is taken from the lower whorl of leaves from a separate plant. Line is 5 cm long. Note the variation in leaf size and extent of dentition. .41 42 papillae and the fine structure of seed coat surfaces is without ridges or projections. It would be of interest to study differences in floral ultra-violet reflectance patterns for this taxcn and its relatives. Use of these characteristics may prove especially helpful in studies of Eastern and Western North American species-pairs. Chromosomal Studies Polyploid-aneuplcid series are common for many groups of plants (Tobgy 1943, Lewis and Raven 1958, Jackson 1962, Kyhos 1965, Carr 1975, Subhasi 1975) including Saxifraga{Dambclt and Pcdlech 1965, Dambolt 1968, Elvander 1975) and are confirmed in the present study as well. Diploids with 10 pairs, tetraploids with 19 pairs and hexaploids with 29 cr 28 pairs have been reported for S. occidentalis var. occidentalis and S. rufidula (Krause and Beamish 1972,1973). It is interesting that in 5. ferruginea a 10-paired diploid and 19-paired reduced tetraploid situation fRandhawa and Eeamish 1972) has evolved in an apparently parallel } polyploid-aneuplcid pattern to the one seen in S. rufidula and its relatives. The same situation may also be the case in S. inteqrifolia var. ijjtejrj,folia where 19-paired populations are common. Chromosome counts for Saxifraga populations are summarized in Table III. Populations of S« rufidula from Vancouver Island, the Olympic Peninsula and the Opper Willamette River dxainaqe are, with one exception, composed of diploid individuals with 10 43 pairs of chromosomes (see Fig. 14C). Krause and Beamish (1913) have reported a hexaploid population (n=29) at Elk Falls on Vancouver Island. Attempts tc relocate that population were unsuccessful, possibly because of local extinction resulting from hydroelectric water diversion.. Saxifraga rufidula populations from the Columbia River Gorge are more complex and variable in their chromosome numbers. Diploid individuals (n=10) occur in some mixed populations with the more common tetraploid (n=19) (Fig. 11). Other populations had only tetraploid or hexaploid representatives (Fig. 12), but the possibility of rare diploid individuals cannot be ruled cut considering the small number of counts made. , Some of the S. rufidula plants from most populations along the Columbia River Gorge undergo an abnormal meiosis typical of hybrids between plants of different levels of polyploidy (Fiq. 13A,B). Diploid, tetraploid and hexaploid plants and their putative natural hybrids are morphologically almost identical. A population from along the Chehalis River in Southwestern Washington previously referred to S. occidentalis (Krause and Maze No. 690001 in Krause and Beamish 1972) is morphologically intermediate between S. occidentalis and S. inteqrifolia. Individuals sampled were uniformly hexaploid (n=29, Fiq. 13C,D). This population probably represents a rare, stabilized allopolyploid hybrid entity. Attempts to locate similar populations elsewhere in the area were unsuccessful, althouqh there are resemblances tc certain S. integrifolia specimens of the Fort Lewis and Olympia, Washinqton reqions. One specimen frcm near Mima, Washinqton had an uncertain chromosome count of 44 Table III. Summary of Chromosome Numbers. Taxa, locations ^collection no.) occidentalis No. No. Plants Cells Meiotic Chromosome No. Ill o_r II+I) * Yale, B.C. 2 6 19 (626) Cornwall Lookout, B.C. 3 10 19 (COBN) Ht. Baker, Hash. 2 5 19 (BAK) ata Saddle Mt., Oregon 3 8 10 (629A) Tillamook, Oregon 11 44 10 (630) Kalama Biver, Hash. 2 7 20 (607) Delena, Oregon (606) 8 23 20(6) ,19+2(2) . dula Mt.Finlayson, V.I., 3 20 10 B.C. (627A) Nanoose Hill, V.I., 4 7 10 B.C. (NOOS-B) Nanaimo,V.I.,B.C. , 4 31 10 (NMO-R) Upper Campbell Lake, 3 5 10 V.I.,B.C. (UCL-R) lake Crescent, Wash. 4 16 10 (673) Bingen Lake, Hash. , 2 10 29 (669A) Ht.Pleasant, Wash. 1 3 10 (675A) Yecn Park, Ore. 6 21 10(2) ,19(3) ,ca.20 (610) Vientc, Ore., 5 13 19(4), ca.19 (616) Troutdale,Ore. 5 7 ca.20*1, ca.27+2. (608) 29, 19+12, 19+13 Bayer Park, Ore. 4 11 ca.19,ca.16+8, <612A) ca.29, ca.20+5 The Dalles, Ore. 4 16 ca.29 (3) ,ca. 26+6 Skamania Co., Wash. 4 26 19"(3) ,ca. 14+8 (618A) •Numbers in parentheses indicate the number of plants with tha particular chromosome number. 45 Table III. (con*t) Taxa, locations jcollection noA) Ho. No. Meiotic Chromosome No Plants Cells • jll or II+I)* occidentalis x inteqrifolia pop, Chehalis R., Hash. 4 (605) 2£ti£ejkiciata Saddle Mt., Ore. 5 (629C) integrifolia Mt.Finlayson, V.I., 3 B.C. (627) Nanaimo, V.I.,B.C, 1 (NMO-I) Elk Falls, V.I.,B.C. 1 (682) Harrison Lake, B.c. 1 (78-1) Yale, B.C. 1 (625) Mayer Park, Ore. 4 (612B) Grizzly Lake, Ca. 1 (671) Bingen, Hash. 1 (617) Mima Bounds, Hash. 1 (MIMft) SlSltcniifolia Troutdale, Ore. 5 (609) Bowena, Ore. 5 (611) Mayer Pk., Ore. 4 (612) The Dalles,Ore. 2 (613) Biggs, Ore. 10 (640) Clark-Skamania Co. line 3 Hash. (619) Gilliam Co., Ore. 1 (641) Cape Horn, Hash. 1 (646) Lyle, Hash. 1 Bingen, Hash. 6 (617) oreaa^a Berthoud Pass, Colo. (BEE) 11 23 3 2 3 3 2 20 4 1 4 16 12 27 6 55 20 2 10 1 30 28,29 (3) ca.36,38j(3) ,ca.40 19 19 19 19 19 19,ca. 17+5(3) 19 19 ca.47 10(4) ,19 10,8 + 11 (4) 10 (3) ,9+5 10 10 10 10 10 10 10 (2) ,14+10, ca.14+13, ca. 18+9, ca.23+10 ca. 36 46 n=ca. 47 (see Fig. 16D) and may be related to the Chehalis hybrid population or to S. oregana (Elvander, personal communication), The chromosome counts for S. occidentalis var. occidentalis agree with the majority of earlier counts (Beamish 1961, Krause and Beamish 1972) of n=19 for all populations sampled in the present study (Fig. 14A,B,0)• Packer (1968) has reported a diploid n=10 population from Blakeston Mt., Alberta and Elvander (1975) records two mixed diploid and tetraploid populations from Trapper Peak in the Bitterroot Mts., Montana and from the Storm Lake Pass in the Anaconda Range, Montana, where individuals apparently are intermediate between S. marshallii sufesp. idahoensis and S. occidentalis var, o cc i de ja t a lis. Krause (Krause and Beamish 1972) has recorded higher numbers of n=28 or ca, 29 for populations in Northern British Columbia and the mountains of Idaho. Plants identified as S. occidentalis var. dentata (Hitchcock et al, 1973) were collected from four locations west of the Columbia River Gorge and toward the mouth of the Cclumbia River. Plants of the Coast Range of Oregon, including those from Saddle Mountain, Clatsop County, are diploid (n=10) (Fig. 15C-E). Plants from Columbia County (EP606)# Oregon, and Cowlitz Co.,Washington (EP607), have 20 or sometimes 19 pairs of chromosomes (Fig. 15A,B). These tetraploids as well as other plants from the Willamette River Valley probably represent allotetraploid hybridizations between S. inteqrifolia var. clavtoniifolia and diploid S. occidentalis var. dentata as judqed by their morphological resemblance and the proximity of 47 Figure 11: Camera lucida drawing of pollen mother cells (PMC*s). Nucleolar organizers are grey circles. Drawings here are from four plants in one population along the Columbia River Gorge. Parts D and E are from the same plant.; &=PMC at diplotene-diakinesis; S. rufidula- (EP610) n=20 B=PMC at diplotene-diakinesis; S. rufidula (EP610) n=19 C=PMC at diakinesis; S. rjgfjjuia (EP610) n=19 D=PMC at diplotene-diakinesis; S. rufidula (EP610) n=10 E=PMC at metaphase II; S. rufidula (EP610) n=10, each daughter cell. C. 10JJ 4* v« 49 Figure 12: Camera lucida drawings of PMC* s. Saxif raga rufidula plants from the Troutdale population. A=PMC at diplotene-diakinesis: S. ruf idula (EPeoSMe) n=ca. 27 bivalents + 2 univalents in the lower left B=PMC at metaphase I; S. rufidula (EP6 0 8-24) n=ca.20 bivalents + 1 univalent C=PMC at diplotene; S. .rufidula (EP608-26) n=29 51 Figure 13: Camera lucida drawings of PMC's., Cryptic hybrid plants from Troutdale (A,B) and Chehalis River (C,D)• a=PMC at late metaphase I; S, rufidula cryptic hybrid (EP608-14) n=ca.19 bivalents +13 univalents (2 bivalents in upper right possibly multivalents?) B=PMC at late metaphase I; S, rjifidula cryptic hybrid (EP6G8-5) n=ca.19 bivalents • 12 univalents C=PMC at diplotene-diakinesis: S. occidentalis var., occidentalis hexaploid hybrid (EP605-12) n=29 D=pcllen grain mitosis; metaphase; S. occidentalis var. occidentalis hexaploid hybrid (EP605-26) n=29 53 Figure 14: Camera lucida drawings cf PMC»s. Plants from Yale, Mt. Baker and Cornwall populations of S. .occidentalis var, occidentalis and Mt. Finlayson population of S. rufidula . Upper scale is for cells A-C, lower scale is for cells D and E. A=PMC at diplotene-diakinesis; S. occidentalis var. occidentalis (EP626) n= 19 I=PMC at metaphase I; S. ^ccideatalis var. occidentalis (BAKEB) n= 19 C=PMC at diplotene-diakinesis; S. rjjfidaia (EP627) n=10 D=PMC at metaphase I; S. occidentalis var. occidentalis lEPCOBN) n=19 ~E=PMC at~~metaphase I; S. rufidula (EP618-17) n=19 55 Figure 15: Camera lucida drawings of PMC*s. S. occidentalis var, dentata plants from Kalama Biver, Delena, Tilamook, and Saddle Mountain. A=FMC at diplotene-diakinesis; occidentalis var. dentata tetrapicid (EP607) n=19 B=EMC at diplotene-diakinesis; S. occidentalis var. dentata tetraploid (EP606-24) n=20 C=PMC at diplotene-diakinesis; S. occidentalis var. dentata diploid (EP630-25) n=10 D,E=PMC at diakinesis and diplotene-diakinesis; S. occidentalis var. dentata diploid (EP629A-25) n=10 57 S. inteqrifolia var. claytqniifoliaplants. However, seme of the S. occidentalis var. dentata plants from the Willamette Biver intergrade in features such as clavate filaments to S.,marshallii subsp. aarshallii but the filaments are less expanded and petal-like than those of S. marshallii subsp. marshallii. Plants known as S. occidentalis var., 1atipetiolata (Hitchcock et al, 1973) from isolated peaks of the Northern Oregon Coast Bange (Chambers 1974), were found to have a hiqh nuaber of chromosomes. Counts varied somewhat, with most being around 38 or 39 pairs (Fig. 16A,C). This corresponds with counts of S. oreqana from at. Hood, Oregon (n=ca. 38, Elvander, personal communication) and from Berthoud Pass, Colorado (n=ca. 36) (Fig. 17,A). Saxifraga inteqrifolia has 19 chromosome pairs on the Lower Mainland of British Columbia, or Vancouver Island ( Fiq. 16,B,E) and southward into Washinqton. Plants from the Columbia Biver Gorqe and Upper Willamette Biver are more variable and further cytological studies are necessary to detect correlated chromosomal differences. At least some of the plants from the Columbia Biver Gorge have the 19-paired complement. Populations from the prairies south of Tacoaa and Olympia Washington may be related to S. oregana and one individual had a high number of chromosomes (2n=ca. 47,51) (my counts. Fig.16,D and Elvander, personal communication). Saxifraga integrifolia var. clay, to n ji f ol ia populations have been repeatedly counted as diploids with 10 chrcuoscme pairs (Fig. 17,B-F). One 19-paired individual was recovered from 58 the otherwise 10-paired population at Troutdale, Oregon. Plants from near Bingen, Washington were mostly sterile hybrid forms with abnormal meiosis although a few 10-paired individuals were found. Further cytelogical and morphological analyses of the Eastern Columbia River Gorge populations are necessary to unravel the complicated relationships between these plants and the closely related S. integrifolia var. inteqrifolia and S. integrifolia var. leptppetala which also occur there. Northward in its range S. inteqrifolia var. l€Ptopeta,la has 19 pairs of chromosomes but populations from the Eastern Columbia River Gorge may be more complex chromosomally where they intergrade with S. integrifolia var, clavtoniifolia and l. integrifolia var. inteqrifolia, In Saxifraga close diploid relatives with 9 chrcioscme pairs are notably absent from literature reports and frcm the present study. The lack of such entities in nature argues against the production of 19-paired plants directly from chromosome doubling of a hybrid between 10 and 9-paired parental plants. A more plausible hypothesis for the 10,19,20,28,29,38 polyploid-aneuploid seguence based on the available evidence is that 10-paired progenitors gave rise to polyploid offspring with 20 pairs of chromosomes followed by subsequent loss of a chromosome pair. Some pairing instability is apparent in one 20-paired population of S. occidentalis var. dentata which may be an indication that it is of recent origin and may eventually conform to the 19-paired polyploid-aneuploid pattern found in many populations of the other taxa. Contact and crossing between 19 and 10-paired plants followed by doubling of chromosome 59 Figure 16: Camera lucida drawings of PMC,s of S. occidentalis var. latipetiolata and S. integrif oljj, a pi a nt s. , &=PBC at diplctene-diakinesis; S. occidentalis var. l^iieetiolata <EP629C) n=ca. 38 B=P0C at metaphase I; S. in^aEiJpJLia var. ifitearifolia (EP617B) n=19 C=PMC at early metaphase I; S. occidentalis var. latipetiolata (E0629C) n= 3 8 D=PMC at diakinesis; S.- . integrjfo^ja-S. oreqana? (EPMIHft) n=ca.47 E=PHC at diakinesis; S. integrif olia var. inteqrifolia (EPNMO-I) n=19 A. B. M X 60 c. D. y \ a 3^ 10JJ 61 Figure 17; Camera lucida drawings of PHC*s. Plants from Berthoud Pass, The Dalles, Lyle, Bingen, Troutdale, and Bowena populations. The upper scale is for cells A-C, the lower scale is for cells D-F. A=PMC at anaphase I; S. oregana (BPBEfi) n=ca. 36 Upper pole shows 34 bodies of which at least 2 are overlapping. lower pole shows 35 bodies of which at least 2 are overlapping, B=PMC at diplotene-diakinesis; S. integrifolia var. clavtoniifolia (EP613-34) n=10 C=PHC at diakinesis; 5. /integrifolia var. clavtoniifolia (EP668-14) n=10 D=PMC at metaphase I; S. integrifolia var. cjjjjtoniifolia (EP617-21) n=10 E=PMC at diakinesis; S. integrj-f olia var. clajtoniifolia (EP6C9-310) n=10 F=PHC at late diakinesis; S. integrifolia var. claytoniifolia (EP611) n=10 62 63 number produced 29-paired populations, and the n=28 counts may be further aneuploid reductions., Probably counts of 38 pairs represent hybrid combinations of two 19-paired entities or combinations involving diploid and hexaploid components. Although the polyploid processes in this group appear to involve mostly allopolyploidy, the possibility of autopclyploidy cannot be eliminated. Reports ic S. reflexa (Krause and Beamish 1973) where diploids, tetraploids, and hexaplcids were found in one population seem to indicate that autopolyploidy occurs in related taxa. Moore (1959) and Taylor and Mulligan (1968) report an apparently autopolyploid derivation for populations of S. taylori. an endemic to the Queen Charlotte Islands and the Brooks Peninsula on Vancouver Island, British Columbia. Mlbridization Most artificial F1 hybrids among S. rufidula. S. occidentalis var. occidentalis. S. occidentalis var. dentata, and S. integrifolia, are generally morphologically intermediate between the two parental entities {Fig. 4-8). However, the multivariate computer analyses placed certain hybrids closer to their polyploid parents than to their diploid parents, especially • S. integrifolia• and S. occidentalis (Fig. 19-21). Pollen fertility of hybrids is summarized in Table IV. Artificial hybridizations were successful among all combinations of taxa which were attempted (Fig. 22). Differences in seed set per attempt possibly reflects variations brought about by inefficient crossing techniques as well as small sample 64 sizes. Similar variation in seed germination rates (.18-.90) may be the result of difficulties in assessment of seed maturity, improper germination conditions, or small sample sizes, apparently it is possible to obtain F1 hybrids from crosses among most plants in this related group without difficulty. The artificial F1 hybrids which survived to maturity demonstrated various meiotic behaviours. Bivalent formation was restricted to a fraction of the genome and multivalents were absent in nearly all crosses (Fig. 18). a possible exception is the cross between S. rufidula (n=19) and S. occidentalis var. latipetiolata (n=38) where larger figures, possibly multivalent associations, could Jse seen in seme cells. Bivalent formation was high (15-19 pairs), possibly indicating that S. occidentalis var. latipetiolata had a hybrid origin involving tetraploid S. rufidula. The morphological affinities and ploidy level (n=ca. 38) of S. occidentalis var. latipetiolata suggest that S. oregana (n=19?.ca. 38) was probably the other parent. analysis of meiosis in artificial intervarietal and interspecific crosses indicates that there has been considerable genomic divergence among the taxa. When diploid S. rufidula is crossed to tetraploid S. occidentalis, the resultant hybrid has 10 or fewer bivalents. Hybrids between S. rufidula (n=19) and -S. ,. occidentalis. var., occidentalis (n=19) produced about 9 bivalents indicating that they have only a partial chromosome complement in common (Fig. 18,C). In contrast, the cross between pcGi den talis var., Occident a lis and S. occidentalis var. dentata (n=20) produced F1 individuals which were almost completely sterile and failed to undergo meiotic division. When 65 S» ruf idula (n=10) is crossed with S. inteqrifolia (n=19) usually 10 or fewer bivalents are forced (Fig* 18). The resultant hybrids closely resemble the sterile intermediates found in many field situations (see Fig. 6). The cross between S. occidentalis var. latipetiolata and polyploid S. rufidula produced several vigorous F1 hybrids which had a higher pollen fertility than other artificial hybrids (Table IV). Occasionally an artificial F1 hybrid deviated entirely from either parental or intermediate morphologies (Fig. 19-21, plant no. 3*). Crosses between S. inteqrifolia and S. inteqrifolia var. leptcpetala resembled S. integrifolia in petal characteristics and had from 6 to nearly 19 bivalents but even in the higher pairing individuals anaphase abnormalities produce highly sterile pollen. S. rufidula x S. occidentalis F1 hybrids resembled §» occidentalis in inflorescence shape and size (see Fig. 4). Although biological relationships and affinities may be elucidated by hybridization experiments, taxonomic decisions in this group probably should be based on the more pragmatic similarities and discontinuities of morphological and merphcraetrie data., Since all possible crosses were not completed (Fig. 22), many aspects of the hybrid relationships are as yet unclear. Even if all hybrids were available for analysis considerable caution in interpretation would be necessary in light of literature reports that in wheat, oats, rye, and ethers, homeclogous pairing is under genetic control (alley 1960# Riley and Law 1965, Rajhathy and Thomas 1972, Hossain 1977) and that bivalent or multivalent formation may be a function ef the 66 Figure 18: Camera lucida drawings of PMC*s from artificial hybrids. The upper scale is for cells A-B, the lower scale is for cells C-D. A=PMC at metaphase I; S. occidental's var. latipetiolata x S. rufidula tetraploid n=ca.18 bivalents + 14 univalents E=PMC at late metaphase I; S. rufidula diploid x S. inteqrifolia var. inteqrifolia n=ca.7 bivalents + 14 univalents C=PMC at metaphase I; S. rufidula tetraploid x S« occidentalis var. occidentalis n=ca.9 bivalents • ca.19 univalents D=PMC at metaphase I: S. occidentalis var. dentata tetraploid x S. rufjidu^a diploid n=ca. 10 bivalents + 10 univalents 68 Figure 19: Principal components analysis graph of axis one vs. two showing artificially produced hybrid individuals and representative parental populations., Symbols are as follows: D= S. occidentalis var. / dentata: 1= S. Inteqrifolia var. inteqrifolia L- S. occidentalis var. 1atipetiolata o= S. occidental's B= S. rufidula (from Vancouver Island and the Olympic Mountains) r= S. rufidula (from the Columbia Biver Gorge) 1= S. rufidula (B) x S.- inteqrifolia (I) 2= S. rufidula (B) x S. occidentalis (o) 3= S, occidentalis (o) x S. occidentalis var. dentata {D) 4= S. rufidula (B) x S, occidentalis var. dentata (D) 5= S. rufidula (r) x S. occidentalis var. dentata (D) 6= S. rufidula (B) x S. occidentalis var. 1§M BftA-ol§ia (L) 3* occupies a point to the right of its placement within the plot. It is probably an aberrant individual., The percentage of the total variance accounted for by the first three axes is 21.36, 11.73, and 7.87 percent respectively. Eigenvector 2 U) O U> H 5 1 + "2° S3 ro-ro ro ro ro m CD < CD o-n <-+ O ~5 Or D MM O M OOM ~ 01 * o ro 0) ro 05 (Jl o o o M Of LO ro-DO 70 Figure 20: Principal components analysis graph of axis one vs. three showing artificially produced hybrid individuals and representative parental populations. Symbols and total variance figures are as in Figure 19. .71 .24-c_ O +-> u £ 0 §. Li R R R 1 r 4 r R R 1 o r6 -.2-h D I L Dxl1 L D o D °I D o o D D D D D D D D D D D D • 1 0 Eigenvector 1 3 72 Figure 21: Principal components analysis graph cf axis two vs. three showing artificially produced hybrid individuals and representative parental populations. Syabcls and total variance figures are as in as in Figure 19. The actual point for hybrid plant 3* lies further to the lower right corner than drawn here. !L L 2 L c I D Ri 11 R o R R R K I D r r 2 1 ° r~ 3 o° r 6 £ J 5 R L 1 L D 23 LR 0 0 • 04 L D D D D D D D DD 0 _ ^Eigenvector 2 74 Figure 22: Crossing success chart for the taxa considered in the present study., Line number and thickness indicates relative success of seed set. Numbers outside the hexagons are the haploid chromosome number. Numbers adjacent to the lines are the average number of seeds set per attempt.,Numbers inside parentheses are total seed set followed by number of crossing attempts, .Letters represent as follows: C=S, occidentalis•• var, dentata. L=S. occidentalis var, la^ipetjolat^, 0=S. occidentalis var, occidentalis, and B=S, rufidul,a. The absence of connecting lines indicates that those crosses were not attempted. 75 Crossing 19 76 Table IVj. Hybrid Pollen Fertility. Cross Between: ruf. (10) ruf. 119) ruf. (10) ruf. (19) occ. (19) lati. (ca. int. (19) int. (19) occ. (19) occ. (19) dent.|19) dent.(19) dent. (19) 38) x ruf. x ruf. (10) * x lept. (19)* No. Individuals 4 4 3 1 3 3 4 2 (19) average J lertiliti 9.9 19.9 37.1 1.0 3.0 31. 0 2.4 1.8 Bange (0.8-24.G) (4. 1-33. 1) (0.7-73.4) (1.0) (0.0-8. 2) (2.3-78.5) (0.0-4.6) (1.7-1. ?) * These crosses were made earlier by K.I. Beamish and are included here for comparison. Varietal names are abbreviated to the first three or four letters. Numbers in parenthesis indicate haplcid number. 77 presence of certain genes or gene combinations. Furthermore, numerous authors have shown that meiotic events in anther tissue can be environmentally modified {Skovsted 1934, Sax 1937, Singh 1975, Beamish 1961) . Ornduff (1969) has presented several cases which argue against rigid species definitions based on crossability and/or interfertility in plant hybridization studies. Naturally occurring hybrids between locally sympatric S. ruf idula and S. -, inte<jrif olia var. intejr.if.plia on Vancouver Island and in places along the Columbia Biver Gorge can be tentatively recognized by their intermediate morphology which resembles the artificially produced hybrid (Fig. 6), abnormal meiosis, and their high degree of pollen sterility . Hybrid intermediates consistently fail to set good seed but are vegetatively vigorous and may exist in nature for many years. Artifical F1 hybrids between S.rjif idula and S. integrifolia have flourished under cultivation in the cold frames at the University of British Columbia for 12 years. Judging from a large population south of Nanaimo, B.C., sterile intermediate individuals are, in nature, usually rare in comparison to their parental entities. Only about 30 putative hybrids were seen compared to about 3,000 S. rufidula plants and considerably more S. inteqrifolia var. integrifolia individuals. In other populations, somewhat aberrant individuals resemblinq S. rufidula in habitat and morpholoqy are perhaps more common. The exact nature of hybrid relationships is much more difficult to determine in the Columbia Biver Gorge area where 78 S. rufidula and up to three varieties of S. integrifolia may occur in close proximity. Variability in levels of ploidy within populations of S, rufidula as well as S. integrifolia var. claytoniifolia also confound precise determinations of hybrid origins. Individuals which are morphologically close to one or the other of the parental entities may prove to be meiotically irregular and pollen sterile. Samples for certain Columbia River Gorge locations, notably the S. integrifolia population near Bingen, Washington, contain a large proportion of these cryptic hybrids. Judging from herbarium material, hybrid swarms are common in several areas., Wilhelm Suksdorf in 1916 made a mass collection of one such area of Eastern Washington near Spokane and Spangle where putative S, occidentalis {or S. marshallii subsp., idahoensis) x S. integrifolia var. leptopetala {or var. columbiana?) hybrids were found. Parental plants were either rare or inadeguately collected. Many of the putative hybrid plants had reduced pollen fertility as well as being somewhat intermediate in appearance. Another area with high numbers of putative hybrid individuals between S. occidentalis var. occidentalis and S. integrifo1ia var. leptopetala is the Petes Point area of the Wallowa Mountains of northeastern Oregon. Hybrid swarms probably involving S. integrifolia var. claytoniifolia and S. marshallii subsp. marshallii have been recorded for the Mary's Peak area of the Coast Range and Opper Willamette region of western and central Oregon. In these areas short-term hybrid fitness, longevity, vegetative reproduction, and large areas of possibly intermediate habitat may explain the 79 apparent abundance of hybrid forms and scarcity of parental forms., Fertile polyploid populations of probable hybrid origin are common in several locations in the Lower Columbia River area. Two tetraploid populations of S. occidentalis var. dentata which conform to Engler and Irmscher's (1916) description and which are near the type locality for S. occidentalis var. dentata, are morphologically intermediate between diploid Coast Range S. occidentalis var. dentata and S. integrifolia var. claytoniifolia (Fig. 23). These resemble the type specimens of §. occidentalis var. dentata from Elk Rock, near Oswego, Clackamas County, Oregon (the first specimen cited in Engler and Irmscher 1916, is Heller 10059) ., Krause and Beamish (1972) have previously referred the Clatsop Co. Plants to S. occidentalis but synonymized S. occidentalis- var. dentata with S. marshallii subsp. ,marshal,lij. Introgression of S. intejrifolia var. clavtoniifolia characteristics into S. rufiduia in the Columbia River Gorge can be seen by comparing S. r of id,a la plants from sympatric and non-sympatrie sites (Fig. 24) . Since the introgressant §• E2£i&£±& plants are also completely introgressant to S. occidentalis (Fig. 25) it is unclear whether intermediacy resulted from past hybridization between S. rufidula and S. occidentalis followed by the local extinction of S. occidentalis from the Columbia River Gorge area or is the result of the leakage of S. integrifolia genes, especially those of S. integrifolia var. clavtoniifolia. into S. rufidula. The hybrid origin of S. rufidula in the Columbia Biver Gorge may 80 Figure 23: Scatter diagram showing introgressicn of certain characteristics between S. integrifolia var. claytoniifolia (sguares) and tetraploid 5. occi dentalis var. dentata plants (circles). Diploid S. occi dentalis var. dentata plants are triangles. 81 50" £ 25 + C 05 CL 5 + H y y + Gland Wcl.(mm) O > .6 4~.5 9 < .39 A A A A A A A A + 2 3 Petal Ln.(mm) 82 Figure 24: Scatter diagram showing introgression of characteristics between S. jntegr.jfolia var. clai+^iiifolia (squares) and sym pa trie" S. rufidula (circles) individuals. Plants in lower right are J' SJfiduia plants from a nearby population which is not locally sympatric with S. inteqrifolia var. clajftoniifolia (triangles) (EP608,609,610). ~ 83 51 T E U i'27 c _CTJ CL Gland Wcl.(mm) • ^.5 mm ; © .3- .49 mm O K29mm. © o 3 + 12 2.5 AA A A ^ AAA A . •3.7 Petal Ln.(mm) 84 Figure 25: Scatter diagram showing intermediacy of certain characteristics between S. occidentalis var, occidentalis <sguares, EP626,Yale, B.C.) and Columbia Biver Gorge S. rufidula plants (circles). Triangles represent presumably non-introgressant diploid S. rufidula plants from Vancouver Island. Leaf Ln.(cm) 0-2.9 0 3-4.5 O 4.6-7.6 O A O A A A + 75 125 Inflorescence Angle (degrees) 86 have led to a secondary resemblance between S. rufidula and S. occidentalis. The more variable introgressant morphs of S» Ofidula are mostly those of higher ploidy levels and thus have probably arisen via doubling of the chromosome complements of sterile hybrid individuals. Variability within S. occidentalis var. dentatacan be demonstrated to correlate with a higher ploidy level which probably reflects a hybrid origin. Tetraploids appear to be completely introgressant with S. integrifolia var. claytoniifolia (Fig. 23). Scatter diagrams in this group are perhaps less useful than multivariate analyses as tools to document introgression since they utilize a rather small number of characteristics. Further evidence of introgression is provided by both Principle Components Analysis and Stepwise Discriminant Function programs. The programs show that the Columbia Biver Gorge S. rufidula plants are more variable than diploid S. rufidula populations from Vancouver Island and the Olympic Mountains, The Columbia River plants also form intermediate spatial relationships between the S, occidentalis or S. inteqrifolia groups and the Vancouver Island S. rufidula groups (see Fig. 27,28,30-35). The latter are apparently not introgressant with either sympatric S. inteqrifolia- var. integrifolia or allopatric S. occidentalis var. occidentalis. 87 Numerical Identification of major groups and populations was possible using the hierarchical clustering program called OBC-CGBOUP. Relationships among larger groups in the phenogram {Fig. 26) were similar regardless of whether artificial hybrids are included or excluded from the analysis with the exception that all S. inteqrifolia plants group together at greater levels of similarity in the absence of hybrids. Artificial hybrids were usually clustered within one parental group although a few clustered independently or with a nonparental group. S. rufidula separated into northern and Columbia River Gorge populations while S. occidentalis grouped with the larger body of S. ruf idula plants for the Columbia River Gorge. Diploid S. rufidula populations clustered together as a group with phenetic similarity to their Columbia River Gorge counterparts. It is interesting that S. occidentalis var. dentata and §• Occidentalis var., latigeti.plata show closer affinities to each other and to varieties of S. i.nteqrj|glia than tc the S.rjifidula and S. occidentalis var./ occidentalis group. Perhaps this is because of the several characteristics which J. occidentalis var. latipetiolata and S.„. occidentalis var. dentata have in common with S. integrifolia- such as gland features and inflorescence shape.; Alternatively it may be a distortion of the relationships between phenetically dissimilar groups of OTU»s (Crovello 1S70). More precise measurement of ovary position could perhaps have increased the phenetic similarity between S. occidentalis var. dentata and the ure 26: Dendrogram using DBC-CGBOUP clustering program. Forty-three standarized variables from 262 individuals were used. Plot is the log of the error value at a given clustering step. Individuals are clustered under one line below an error value of 15. Beference to the numbers is given in the accompanying key. Varietal names are abbreviated to the first 3 or 4 letters. B.C=British Columbia, 0.N.—Olympic Mountains, C.E.G.=Columbia River Gorge, V.I.=Vancouver Island, Hybs.=hybrids. Key to line numbers: 1. rufidula x occidentalis, 2. 2ati£eticlata x rjjiidula 3. rufidula x dentata 4. integrif olia x Oliiuia 5. rufidula x 605 qccidentaljs (Chehalis fi.) 6. integrifclia x rufidula 7. rjafi^ula (618) ; ififitata {629) 8. dentata (630D) 9. 2 rufidula ; integrifolia x rufidula: dentata x rufidula 10. occidentalis x dentata 11• rufidula x occidentalis 12. occidentalis x dentata 13. rufidula x occidentalis 14. 2 rufidula x occidentalis 15. rufidula x occidentalis 16. dentata x rufidula: rufidula x occidentalis 17. occidentalis: rufidula 18• occidentalis 19. 3 rufidula: clavtoniifolia 20. EP605 (Chehalis B.) 21. 1 £te^rifs 1 ia; le£topetaja 22. integrifolia 23. occidentalis x dentata 24. rufidula 25. rufidula 26. rufidula 27. rufidula ; .integrifolia 28. rufidula 29. occidentalis x dentata 30. 2 dentata" 31. ifltga£ifclla x IsJio^gtala 32. integrifolia x leptopetala ; leptcpetala x dentata 33. rufidula x "(EP605) Chehalis fi. 34. rufidula 35. dentata 36. rufidula x dentata 37. (EP605) Chehalis R, , 5a. occidentalis x S. integrifolia hexaploid population 1500 — 500— oo VO 90 S. ^gujijula and S. occidentalis groups. Results of the hierarchical grouping analysis of S. occidentalis- var. dentata and S. occidentalis var., latipetiolata clearly indicate that these two groups deserve specific status. Saxif raga initge,gr.j.foijLia -var. leptopetala failed to separate frcm S. inteqrifolia var. inteqrifolia. probably because of the extensive numbers of intermediate Columbia River Gorge plants in the sample.. In contrast, S..inteqrifolia var. claytoniifolia populations form a fairly distinct group from S. integrifolia in spite of close contacts in the Columbia River Gorge (Fig. 26). The population from Chehalis, Washington (EP605) clustered independently with some affinities for the group of mixed S. integrifolia, S. integrifolia var., leptopetala. S. integrifolia var. claytoniifolia, and hybrids. Its morphology and chromosome number (n=29) suggest an allohexaploid origin between S. occj,dentaj.jsand S. integrifolia. The relationships of groups in the hierarchical clustering analysis compare only roughly with present taxonomic treatments (Krause and Beamish 1972,1973, Hitchcock et al, 1961, Hitchcock and Cronguist 1973, Elvander 1975) .Saxif raga occidentalis var. latipetiolata and S. Occident alls var. , dentata do not appear to cluster as good subspecific entities of S. occidentalis var. occidentalis. The allohexaploid population from near Chehalis, Washington does not group with the S. occidentalis or •S. Ofidula individuals although it is considered within S. occidentalis by Krause and Beamish (1972) and is probably a hybrid involving S. occidentalis. The clustering of this 91 population as well as that of a number of artificial hybrid individuals is similar to the clustering pattern described by Smith (1969) in Vaccinium where natural hybrids formed clusters that were distinct from clusters of their presumed parents. The usefulness of a clustering technique in the present study appears limited tc the definition of larger natural groups on the basis of multiple characteristics. Determination of biological relationships, phenetic distance among the groups, hybrid affinities, and correlations to chromosome number are all guestionable using this technigue. However, the surprisingly high grouping error level between S. occidentalis var. dentata and S. occidentalis var., latipetiolata strongly suqqests that they are morphologically distinct from S. occidentalis var. gccidentalis and S. rufidula. The present study confirmed reports by others (Eyles and Blackith 1965, Rising 1968, Hhitehouse 1969, Schilling and Heiser 1976) that Principal Components Analysis (PCA) treats hybrid relationships and other factors including correlations with cytological data in a more precise manner than standard hierarchical clustering methods. The recognized taxonomic groups present spatial overlaps with intermediate individuals among several groups (Fig. 27-29).; In contrast to a repeated multivariate study of introgression which involves crossovers in hybrids and backcrossing to one parent at the diploid level (Bloom 1976) the introgression within the present group appears to involve processes of polyploidization and consequently increased variability within the parental entities. Considering the probable hybrid origins of such polyploid taxa as 92 S. occidentalis, S. jjite^Kif alia, S. occidentalis var. latipetiolata. and certain populations of S. rufidula and S« .occidentalis var. dentata. interaediacies represent reasonable assessments of the ccmplex relationships involved. Saxifrajga ruf idula populations separated into two clouds with northern diploids grading into a mixed group of diploid, tetraploid, and hexaploid individuals from the Columbia Biver Gorge (Fig. 27,28) . Tetraploid S. .occidentalis var. dentata populations showed greater variablility and tended to group in intermediate hyperspaces close to S. occidentalis, and S. occidentalis var., J=ati£etiolata, S. integrifolia and the allohexaploid plants from Chehalis, Washington., When artificial hybrids are included in the PCA, their intermediacy and variability is clearly represented (see Fig. 19-21). Occasional segregates (Fig. 19-21, plant 3*) appear different and are probably extremely unbalanced monstrosities which would not survive in nature. Most hybrid individuals show closer affinities to the Columbia River Gorge S. rufidula. S. occidentalis or S. integrifolia parent than to the other parental group. The raw principal component (eigenvector) solutions (Table V, hybrids excluded) are similar whether or not artificial hybrids are included in the analysis. Ten parental and hybrid groups defined by hierarchical grouping and principal components analyses and standard taxonomic treatments were recognized for use in the stepwise discriminant function (SDF) analysis (Fig., 30-35). The six discriminating variables and F values for inclusion (critical F probability 0.05), in the order used, were: plant height (51.9), 93 Figure 27: Principal components qraph of axis one vs. two showing individuals frcm several natural taxa and populations. Artificial hybrids are not included. Symbols are as follows: B=diploid Vancouver Island and Olympic Mountains S. rufjdula r=polyploid Columbia Biver Gorge S. uiiiLaJLa D= S. ocsidsntalls var. deflt&ta 0= S. occj,<Jgntaiis var. sssiijgfltalis L= S. ocgiigntalis var. laJijaetloJaia X=hexaploid hybrid population from Chehalis Biver EP6C5 1= S. integrifolia var. integrifolia c= s. iatsatiiciia var. cj^itsniUeiia. M= S. ifiteg£ifclia var. l& The percentage of the total variance accounted for by the first three axes is 19.95, 12.17, and 6,87 percent respectively. Eigenvector 2 to o LO 33 33 33 SL33^ -i 33 ^V3 50 33 S3 S3 S3 S3 S3 33 S3 S3 *S3^ -, siV _S3 S3 ~> -J S3. -J -J m CD < , CD o-f-o o o o o o o D O O M M 7 H D M M 0DH° oo a ox o o o o M M MM M O O O O O o o on o n O H o ^ o o n °o o n o o o o o o o^ o o_c9 o X o ^ r- O X X o o o o o c o c o o n o 95 Figure 28; Principal components graph of axis one vs. three showing individuals from several natural taxa and populations. Artificial hybrids are not included. Symbols are as in Figure 26. 96 .3-00 £_ o +-» u (D 0 c CD Lu -.3-O I I M °I I I C M M M M C C IC M C M R „ r r 1 r r r r r R r r r r P R p BRR RR R r R r r R R K R r R R R _ r R r R r R r r R r r r x I 11 I x ° .o ix. c 0 c X X c L D C o ib D- oi O OO 0 L C • cc D ^ C D CI n C r D D  u L C D D L D M C DC C c C C D D D D D D _ D D D D D C CC D H_ 0 Eigenvector 1. .2 97 Figure 29: Principal components graph showing individuals frcm several populations. Artificial hybrids are are as in Figure 26, Plant *C is the right in this representation. of axis two vs. three natural taxa and not included. Symbols displaced slightly to 98 00 L. o •+-> u 0) o > c <D D) Lu c*c c c c CC c c M 1 i1 M M M M M M M I L .M r r O r X 0_ X IC I D ic R OR R Rr R & R" R IBC R °R CC ° c c c ' c c D Si r 0 rXx r Mn L r r r r L rr 1 r r r Xr Dr r -j-%r X r Oo r OO r 0°r O. L D O O O L r D D. D 1 D D r DL L D D , D D D D 1 Eigenvector 2 99 Table V.. The raw principal component solution.* Variable No. Jst 2nd 3rd 4 th 1. -0.1174 0.2560 0.0 362 0.0733 2. 0.0941 0.1533 0.0639 0.0 938 3. -0.0110 0.1943 0.1528 0.0401 4. 0.1087 0.123 8 -0.0 270 0.2 052 5. 0.2609 -0.0537 0.1897 -0.0 963 6. -0.0485 -0.0380 -0.0228 -0.1*23 7. -0.1232 0.2034 0.0256 0.2571 8. 0.1223 0.0767 -0.0233 0.0258 9. 0. 1798 -0.0644 0.3628 -0.0733 10. -0.0812 0.2692 -0.0656 0.1438 11. -0.0397 0.0204 0.1641 -0.2045 12. 0.0768 0.1146 -0.3414 0.1300 13. 0.0114 0.3009 0.0120 -0.0150 14. -0.0368 0.1845 0.0738 0.0048 15. 0.1485 0.2158 0.1051 -0.0034 16. 0.1247 0.1240 C, 10 93 0.0663 17. -0.0268 0.0197 -0.0664 -0.0991 18. 0.0922 0.1000 0.0863 -0.1326 19. 0.1828 0.1409 0.1536 0.1225 20. 0.1451 0,1982 0.1312 0.0790 21. -0.1307 0.2130 0.0324 -0. 1724 22. 0.1143 0.0221 -0.2373 -0.1960 23. 0.0214 -0.0568 -0.1147 -0.0713 24. 0.1383 0.1932 -0.2079 -0.OCC1 25. 0.2367 0.0714 -0.1214 -0.2404 26. -0.2871 0.0182 -0.0S38 -0.1099 27. 0.2956 -0.0747 -0.0520 -0.0546 28. -0.0455 0.2652 0.1701 0.1720 29. 0.0639 0.1619 -0.1869 -0.0798 30. 0.2615 -0.0067 -0.2043 -0.0916 31. 0.2743 -0.0533 -0,1184 0.1539 32. 0.1939 -0.0791 0.0 717 0.3015 33. 0.1864 0.2205 -0.1632 -0.0470 34. -0. 1018 0.2619 0.0607 -0.0 926 35. 0.G595 0.0153 -0.2388 0.1640 36. 0.2190 -0.002 5 0.2913 0.0396 37. 0.0748 0.1716 0.1181 -0.3577 38. 0. 1621 -0.1472 -0.1351 0.3517 39. -0.1733 0.2272 -0.2316 0.0896 40. 0.2593 0.0451 -0.1651 -0.1621 41. 0.1418 0.0209 0.1137 0.0500 42. -0.0074 0.0292 C.C793 0.2037 43. 0.0821 0.1830 -0.0541 -0.1629 44. 0.0821 0.1830 -0.0541 -0.1629 * These expressed as eigenvectors; results of the principle components without artificial hybrids are similar to those with hybrids (above). **• Variables are as coded in Table II. 100 leaf blade angle at petiole (23.16), gland height/width (21.09), petiole length/width (17.81), gynoecium height (15.52), and leaf length (12.76). Group classification functions show that plant height (Variable 27) effectively discriminates S. occidentalis and diploid, polyploid, and hybrid S. rufi(|ula individuals from the remaining taxa and S. integrif o'\ja. var. claytoniif olia from all the others.. Leaf shoulder angle (Variable 37) separates S. occidentalis var. latipetiolata and S. inteqrifolia var. IgPjopetala from the other groups while gland height/width ratio (Variable 6) discriminates diploid S. rufidula, S. occidentalis var. latipetiolatat and S. occidentalis var. dentata groups. Petiole length/width ratio (Variable 39) distinguishes S. inteqrifolia var. claytoniifolia and S« integrifolia var. leptopetala and gynoecium height (Variable 10) separates S. occidentalis and the S. occidentalis hexaploid hybrid population. Leaf length (Variable 32) separates both groups of S. rufidula from the remaining group. Of the total cases, 65.79% were correctly classified by SDF analysis including over 65% of S. integrifolia var. integrifolia, J. occidentalis var. occidentalis. S. occidentalis var. 2aii£§M2l3iS 31) , S. integrifolia var. claytoniifolia, and Vancouver Island and Olympic Mountains S. rufidula individuals. Some 63.7% of the S. jlntegrifolia var. leptopetala plants were grouped with S. integrifolia (Fig. 31,B), while individuals of the allohexaploid population (FP605) were scattered among S. occidentalis (20%), tetraploid or hybrid S. rufidula (30%), and diploid S. rufidjila (20%) groups (Fig, 101 Figure 30: Stepwise discriminant functions plot of the first three canonical variables. The first two canonical variables are stratified on the third variable to produce a stratified three-dimensional plot which clarifies hyperspatial relationships. This representation depicts relationships of group means as spheres on supporting lines. Group means symbols are as follows: A= S. rufidula plants from the Columbia River Gorge and hybrids B- S. integrifolia Columbia River Gorge O S. inteqrifclia D= S. occidentalis L= S. occidentalis M = S. integrifolia 0= S. occidentalis B= J« rufidula plants from Vancouver Island and the Olympic Mountains I= S. integrifolia var. integrifolia plants frcm Vancouver Island X=Hexaploid S. occidentalis x S. inteqrifolia plants from Chehalis, Hash. var. integrifolia plants from the var. clajftoniifolia var. dentata v ar. l3ti£etiolata var. lentopetala var. occidentalis 103 32,A). Of S. occidentalis var. dentata individuals, 28% were placed into the S. inteqrifolia group (Fig. 33,A). Columbia Biver Gorge S. integrifolia plants (Fiq. 34,A) demccstrate intermediacies with S. inteqrifolia var. clavtoniifolia (19%) (Fiq. 34,B) but probably they have insiqnificant differences frcm the Vancouver Island S. inteqrifolia var. inteqrifolia qroup with which 12% are classified. Several Columbia Biver Gorge S. rufidula individuals including artificial hybrids (Fig. 35,B) are intermediate between S. inteqrifolia var. integrifolia (5%), S. occidentalis (12.5%). S. rufidula (10%) and S. occidentalis var. dentata (5%). Such intermediacies and increased variability in SDF analysis relationships have been used as evidence for the occurrence of introqression in natural populations (Schueler and Rising 1976). It has been suggested that SDF analysis is a preferred method for separation of parental species groups, especially where they are difficult to separate by other means (Danick and Burns 1975, Namkoong 1966). The probable hybrid nature of the tetraploid S. cccidentalis var. dentata plants, Columbia Biver Gorge S. rufidula groups (Fig. 35,B), and a hexaploid population from Chehalis, Washington is documented by SDF analysis as is the uniqueness of the S. occidentalis var. latipetiolata morphology. Taken in conjunction with results of the hierarchical clustering analysis program, the SDF program results agree favorably with a treatment which classifies S. rufidula. S. o c ci dent a11s var. lsii£§ij=2JsSta, and S. occidentalis var, dentata as distinct species from S. occidentalis. According to the rules of botanical nomenclature 104 Figure 31: Stepwise discriminant functions plots of the first three canonical variables. The first two canonical variables are stratified on the third variable to produce a stratified three-dimensional plot which clarifies hyperspatial relationships. Group means are shown as in the Figure 29. Part A shows individual plant points for the S. integrifolia var. claytonjifglja cloud as triangular symbols and Part B shows similar hyperspace points for the S. integrifolia var. leptopetala cloud. Symbols are as in Figure 29. 105 106 Figure 32: Stepwise discriminant functions plots of the first three canonical variables. The first two canonical variables are stratified on the third variable to produce a stratified three-dimensional plot which clarifies hyperspatial relationships. Group means are shown as in the Figure 29. Part A shows individual plant points for the S. occidentalis hexaploid hybrid (Chehalis River, EP605) cloud as triangular symbols and Part E shows similar hyperspace points for the S. •,Occidentalis- var, latipetiolata cloud. .Symbols are as in Figure 29. 5 G tr Canonical Variable 2 108 re 33: Stepwise discriminant functions plots of the first three canonical variables. The first two canonical variables are stratified on the third variable to produce a stratified three-dimensional plot which clarifies hyperspatial relaticnships. Group means are shown as in the Figure 29. Part A shows individual plant points for the S. occidentalis var. dentata cloud as triangular symbols and Part B shows similar hyperspace points for the S, occidentalis var. occidentalis clcud, Symbcls are as in Figure 29. Canonical Variable 2 6br 110 Figure 34: stepwise discriminant functions plots of the first three canonical variables. The first two canonical variables are stratified on the third variable to produce a stratified three-ditiens icnal plot which clarifies hyperspatial relationships. Group means are shown as in the Figure 29. Part A shows individual plant points for the §. integrtfglia var. integrifolia cloud frcm the Columbia River Gorge as triangular symbols and Part B shows similar hyperspace points for the S. integrifolj.a var. integrifolia cloud from Vancouver Island and the B.C. Mainland. Symbols are as in Figure 29. in 112 Figure 35: Stepwise discriminant functions plots of the first three canonical variables. The first two canonical variables are stratified on the third variable to produce a stratified three-dimensional plot which clarifies hyperspatial relationships. Group means are shown as in the Figure 29. Part A shows individual plant points for the S. rufidula cloud from Vancouver Island and the Olympic Mountains as triangular symbols and Part B shows similar hyperspace points for the S. rufidula cloud from the Columbia Biver Gorge and the S. rufidula hybrid cloud. Symbols are as in Figure 29. Canonical Variable 2 114 S. rufidula becomes S. aeguidentata, S. occidentalis var. latipetiolata becomes S. latipetiolata. S. occidentalis var. dentata becomes S. gormanii. and S. ... occidentalis becomes S. occidentalis (see descriptions in Taxonomy section). Breeding System Observations Baacfing, Tests The results of tests for automatic selfing and self incompatibility are presented in Table VI, Seed set from automatic selfing tests in the bagged, untreated condition is frequently higher than seed set in the hand-selfed condition. This probably resulted from damage in handling especially during emasculation and difficulty in timing pollinations to coincide with stigmatic receptivity. All taxa are clearly self-compatible and the relatives of S. rufidula, e.g. S. occidentalis var. dentata and S. occidentalis as well as S. inteqrifolia var. leptopetala may have a higher degree of potential self-fertilization than S, intejriJalia var, integrifolia. S, .integrifolia• var. claytoniifolia. or S. occidental's var, latipetiolata. Elvander (personal communication) reports that he obtained no seed set for members of the S. inteqrifolja-S. oregana complex in bagged, untreated conditions but similar seed set in bagged hand-selfed tests to those reported here for S. integrifolia and its relatives. Differences in the results cited here are possibly due to the bagging method used in this 115 Table VIj, •, Bagging tests for selfing. Selfed by. Hand Ro. Total Bruits Seed (Plants) Abbreviated Taxa ruf. (dipl.) ruf. (poly.) dent, (dipl.) dent, {tetr.) occ. (tetr.) 57 (6) 33 (5) 82(4) 35(3) 40(3) lati. 112(3) (n=38) int. (tetr.) clay, (dipl.) lept. (tetr.) 13(2) 67(3) 8(1) % Good* Seed 1402 646 1313 1312 1719 1087 646 603 37 55.2 25. 1 55.9 59.1 27.9 6.3 13.9 16.9 21.6 Bagged Untreated No. Total Fruits Seed (Plants) 19(3) 4(1) 7(1) 11(1) 27(3) 59(2) 5(2) 20(1) 37(3) % Good Seed 990 230 228 254 1516 530 407 179 470 48. 7.8 0.0 76.4 48.9 5.8 0.4 15. 1 35.5 * "Good" seed defined as large, swollen seed versus small, shriveled seed which was considered unfertilized potential seed. 116 Table VII. Bagging tests for outcrossing and apomixis. Outcrossed Ml land Outcrossing C cntrcl.Ontreated Emasculated Bagged No. Total % Good* No., Total % Good* No. Total % Good Fruits Seed Seed Fruits Seed Seed Fruits Seed Seed (Plants) (Plants) (Plants) abbreviated Taxa ruf. 10(2) 505 59.6 315(8) 7698 66.3 (dipl. ) ruf., 20(3) 932 56. 1 103(5) 4054 38.9 (pcly.) dent. 17(1) 206 33.5 355(5) 3304 69.4 (dipl. ) dent. 17(1) 675 71.0 61 (4) 2253 63.3 (tetr.) occ, 21(2) 1210 .6-2.5 41 (3) 2767 45.5 (tetr.) lati. 52(2) 707 39.6 (n=38) int. ./ (tetr.) 6(1) 348 0.0 35(5) 1852 75.1 401 0.0 clay. 22 (2) 1271 74.3 41 (1) (dipl.) lept. 8(4) 681 82. 5 12(1) 544 1.3 (tetr.) 10 (2) 608 0.0 21(4) 621 0.5 6 (1) 149 0.0 47(7) 2554 0.2 15(2) 676 0.0 30(6) 1691 0.0 32(2) 410 O.O 44(6) 1 175 0.3 * Gocd seed defined in Table VI. 117 study which may have allowed autodeposition as a result of mechanical jarring of the bagged, untreated inflorescences. Seed set in outcrossing experiments (Table VII) was usually higher in J3. integrifolia relatives, or nearly egual in S. occidentalis relatives, to seed set in the selfing tests. In a few cases the outcrossed seed set was lower than in selfing tests. Ter-Avanesian (1978) has reported that in several different flowering plant groups, low numbers of pollen grains deposited on receptive stigmas result in a general failure of fertilization. This may account for failure of some individuals to set seed where efforts to deposit large amounts of pollen were unsuccessful. Single individuals of S. occidentalis var. ia t i£ e t i o 1 a t a , S. integrif olia var., cla ytoniif olia . and S. inteqrifolia var. leptopetala failed to set large amounts of seed when exposed tc pollinators, perhaps as a result of premature harvesting or infertility in those particular individuals. In all cases seed set was absent or at low background levels (less than 1.0%) in emasculated, bagged inflorescences (Table VII). Although apomixis which reguires pollination is possible, the wide range of variability of offspring from individual plants and the fact that hybridizations produce offspring which differ markedly from the female parent argues against its occurrence at least as a major means of seed production. 118 flojEili Observations Nectar production begins with anther dehiscence and continues for several days until the apparently receptive, papillate, stigmatic surface loses its characteristic wetness. Some UV-absorbing and re-emitting substances are present in Saxifraga nectar which may be visual stimuli to foraging bees or flies. Saxifraga flowers are characteristically protandrcus but there is an overlap in final anther dehiscence and onset of stigmatic receptivity., Although some spatial separation exists between dehisced anthers and stigmas (Fig., 2,3), filament and style elongation and curvature may bring the anthers and stigmas into close proximity so that auto-depcsition is possible provided that pollen has not been previously removed by foraging insects and that slight external forces bring the anther and stigma into contact. There is also the possibility that pollen may float across a film of water from anther to stigma, as observed for other plants by Hagerup (1950, 1951) or be blown about by wind (Hyde 1950,1969, Hyde and Williams 1961, Proctor and Yeo 1973) , especially since pollen was observed to lose its self-adherent properties after prolonged contact with air. No tests for wind-pollination or repeated tests for water-mediated selfing were carried out in the present study. However, casual observations demonstrate that fresh Saxifraga pollen is hydrophobic and droplets from a light misting can accumulate in Saxifraga flowers producing a continuous water surface between anthers and stigma on which rafts of pollen were observed to float and apparently contact the stigma. Also a reduction was 119 noted in the adherence of pollen grains to each other and to the tapetal wall after prolonged exposure to the atmosphere. Such a reduction in oily adherence may enhance the chances of limited anemophily for these open, dish-type flowers in the absence of animal pollen vectors. Temporal separation of flowering times is evident between sympatric taxa. However, in most cases there is a short period of phenological overlap during which interspecific pollinations are possible. No such pollinations were recorded during periods of insect observation but the presence of hybrid individuals testifies to the occasional occurrence of hybridizing events. Whether there are phenological differences correlated with different levels of polyploidy within populations as Lewis and Suda (1976) have described for Claytonia virqinica is unclear from the present samples. Determination of such differences would require more extensive and temporally separated sanplinq of individual populations which exhibit two or more levels of polyploidy. It is interestinq to note that where different diploid and polyploid taxa occur sympatrically, the polyploids tend to flower later in the season than the diploids. Seasonal isolation also appears to occur to some extent within taxa over broad qeographic or topographic ranges.t Southern lowland populations tend to bloom earlier than those at higher latitudes or altitudes, even under uniform transplant conditions. The number of ovules per carpel varies considerably within an inflorescence. The earlier-maturing flowers which occupy positions closer to the central branches produce more ovules than the later maturing ones which are at progressively more 120 terminal positions on the inflorescence branches. Some differences in ovule/carpel number among taxa exist. When the earliest, most central flowers are compared, S. occidentalis var. dentata. 5. rufidula. S. integrifolia var. inteqrifolia and S. occidentalis var. latipetiolata plants form a series frcm lowest to higher numbers of ovules per carpel with the highest numbers appearing in S. integrifoli a var. leptopetala. S. integrifolia var. claytoniifoliaand S. occidentalis plants in that order. ,• These trends do not seem to be correlated with infloresence size, extent of branching, or flower number. Although detailed observations are lacking, other floral features do not seem to show such a marked reduction in numbers or size with position in the inflorescence. Further studies are needed to determine the systematic and biological significance of these observed differences in ovule/carpel number. Elvander C1978) has observed that some populations of S. integrifolia var. inteqrifolia are gynodioecious, an observation which is supported by the finding of male-sterile plants, apparently seed-fertile on Vancouver Island. He has noted a positive correlation between the degree of compaction of the inflorescence and the degree of flowering synchrony among varieties of 5. integrifolia. He concluded that inflorescence compaction and associated flowering synchrony promotes outcrossing in populations with compacted or congested inflorescences. Evidently S. integrifslia var., integrifolia populations with less compacted inflorescences are the ones which are gynodioecious. Evidence of male-sterile individuals and evidence for similar correlations involving degree of 121 compaction and gynodioecy in the taxa other than S. integrifolia is lacking from the present study. The reasons for complete seed and pollen sterility in some individuals are unclear, but one possibility is that they represent vegetatively vigorous segregant hybrid or mutant forms which closely resemble one or the other parent in morphology and habitat preference, Another explanation may involve the degree of instability inherent in a polyploid genome with geologically recent origins. In certain sterile individuals what appear to be mycelial fragments may be seen among the abortive pollen or pollen mother cells. Fungus-induced sterility may indeed be. a common feature of certain populations. Whatever the explanation, such vegetatively vigorous, sterile individuals may compose a large portion of certain populations (Beamish 1961). Although the bagging tests and floral observations reveal interesting biological and evolutionary tendencies, they are not useful in uncovering characteristics to separate the closely related taxonomic entities involved in this study, . Geoggaghy• Habitat And Pollination Ecology Observations Geographically S« occidentalis var. occidental's is a widespread entity which is found on mountaintops and cliffsides in the Bocky Mountains and Cascades from southern Alaska and the Yukon to Montana and Idaho. It grows on exposed mossy or bare grcund in rock crevices and cn rocky peaks in vernally mesic habitats which become rapidly xeric as the season progresses. It is not found on Vancouver Island, the Olympic Mountains or in 122 the Columbia Biver Gorge. Saxif raga rufjulula grows commonly on losland cliffs and other sites where early spring runoff is abundant. It is found from Lane County in Oregon north into the Columbia Biver Gorge and then in the Olympic Mountains and northward on the east side of Vancouver Island. It occurs on mountaintop areas in the Olympic Mountains and on Vancouver Island as well as at lower elevations. Local population extinction caused by increased grazing pressure with the recent introduction of mountain goats may explain failure to find S. rufidula plants which have been reported on Hurricane Ridge and in the Obstruction Pass Area of the Olympic Mountains. Verbal reports that S. rufidula occurs on the British Columbia mainland were also investigated but only one occurrence has been documented. Literature reports (Welsh 1974) of S. rufidula in Alaska may be the result of confusion with S. occidentalis var. occidentalis or S. nivalis since there are no specimens of §• rufidula or S. occidentalis var. rufidula in the collections at the University of Alaska, the State Museum at Juneau, in the Anderson collection at Ames, Iowa, or among Welsh's specimens from BYU. The specimen reported for southern Alaska by Engler and Irmscher (1916) is also probably a small S. occidentalis var. occidental's plant since it was first labeled as S. yj.ralliensis, although it was not seen during the present study. The habitats of S. occidentalis var. dentata are similar tc those of S. rufidula but S. occidentalis var. dentata is found west of the Columbia River Gorge in the Columbia River and Willamette River lowlands and higher into the Coast Range of 123 Oregon. It is commonly found on shallow soil of volcanic rock faces in open or partly shaded seepage areas where early spring moisture is abundant. Its extensive rhizome system apparently allows it to obtain footholds on loose or crumbling substrate and propagate by vegetative means. It is interesting that Chambers (personal communication) has collected material from a sunny location on Saddle Mountain which closely resembles §* rufidula while other plants in shady areas from the same locality are nearer to S. occidentalis var. dentata., Examination of plants from open, sunny areas at the same locality in the present study indicates that the snaller, heliophilic forms are diploid S. occidentalis var. dentata plants but further cytological studies might reveal tetraploid or hexaploid S. rufidula plants in the vicinity. Saxifraga occidentalis var. latipetiolata is endemic tc a few isolated volcanic plugs and mcuntaintop balds of Clatsop and Tillamook Counties, Oregon. It grows in open, rather grassy areas in somewhat deeper and perhaps drier soil. It blooms later in the season than S. occidentalis var, dentata which is sympatric with it on Saddle Mountain and probably at other sites as well., The varieties of S. inteqrifolia are geographically segregated over much of their range, but in the Columbia Siver Gorge they often are found growing together in mixed populations. Seasonal differences in flowering times and possible microhabitat differences apparently isolate the varieties which are sympatric although there is evidence that hybridization and introgression are common in the Columbia Eiver 124 Gorge. Saxifraga integrifolia var. integrifoliaoccurs on open grassy soil pockets and ledges of rocky headlands and outcrops from the Upper fraser Valley and Central Vancouver Island south through the Puget Sound and then east into the Columbia Biver Gorge and southward as far as south central Oregon. S. integrifolia var. leptcpetala is commonly found in open rocky places or grassy savannah beneath Pinus ccnderosa in the British Columbia interior southward to Montana, Idaho, and Oregon and west as far as the eastern portion of the Columbia Biver Gorge. S. .integrifolia var. clavtoniifolia occurs in shallow and often gravelly or disturbed soil along roadcuts and strearabanks as well as on ledges of rocky outcrops and cliffsides from the Washington Cascades south to the Columbia Biver Gorge and from northeast Oregon and western Idaho west and south to Northern California. Observations on soil depth indicate a significant difference between the S. rufidula or S, occidentalis plants and the locally sympatric varieties of S. integrjlfolja (Table VIII). Differences in soil depth preferences for one larger sample of sympatric S. integrifolia var. claytoniifolia and S. /integrifolia•• var. integrifolia plants from the Columbia Biver Gorge were less obvious. Similarly no clear differences in depth of soil to rocky substrate emerge between northern and southern S. rufidula populations, S. occidentalis var. occidentalis •• may have a deeper soil depth preference than does S. rufidula but further observations are necessary. Naturally occurring hybrids between S. rufidula and S. integrifolia were sampled at one locality and average soil depth is intermediate 125 Table Villi Average soil depth (cm) beneath plants representing two sympatric taxa from localities in the Columbia Biver Gorge and British Columbia. rufidula i£tejrilglia No. . Ave. No. , Ave. IfOca 1 i ties Hants Depth Plants Depth Columbia Biver Gorge Troutdale 10 4.3 10 4.9 Mt.Pleasant 9 3.6 9 6.7 Clark Co. Line 9 3.3 10 9.6 Hosier 10 4.4 10 8.6 Mayer Park 10 4.9 10 14.0 The Dalles 0 4.5 20 11.6* British Columbia Ht.Finlayson 15 6.2 15 10.7 Sooke 15 3.3 15 6.5 Nanaimo** 15 3.6 15 6.5 Nanoose 5 3.1 15 5.7 * A subsample of integrifolia var. integrifolia (n=20) from this locality had an average depth of 14.9cm. Other Columbia Biver Gorge plants are var. ciaytonijfolja and B.C. plants are var. integrifolia. ** An intermediate hybrid subsample (n=15) from this site had an average soil depth of 5.5 cm. 126 Table IJU Soil moisture in g. H20(g. dry weight)-1 under plants at five sympatric localities. Site Date rufidula integrifolia Nanoose March 1.039* .565 April .801 .675* May .514 .216 Nanaimo March 1.117 .487 May 1.014* .517* Ht. Finlayson March .995* .638 Sooke March .752* .725 occidentalis integrifolia lale April .614* ~* .426 * These dates closest tc height of bloom for that plant taxon at each locality., 127 for the hybrid plants. A few sympatric locations were sampled for soil moisture to document differences in water relations between S. rufidula and S. inteqrifolia and between S. occidentalis and S. integrifolia (Table IX). Especially during peak flowering times the soil beneath S. rufidula plants contains more moisture than soil beneath S. integrifolia plants. If field germination occurs in about April or May, as it does for seedlings in cultivation under natural temperature and light conditions, similar differences in soil moisture levels may be important factors in habitat selection at the seedling establishment phase. Such differences in water relations help to explain higher observed mortality rates in transplanted S. rufidula specimens, assuming that the artificial watering regime duplicated S. inteqrifolia soil moisture parameters but failed to maintain adequate or sustained soil water levels for S. rujidala. Insect spectra are summarized in Table X* , Flowers of all taxa studied were visited predominantly by nectar-loving flies although bees and occasionally wasps are also attracted to the flowers and collect both pollen and nectar. The flies are divisible by their behavior into motile and more sedentary categories with the larger anthophilous Syrphids, Bombylids, and deerflies much more likely to effect cross-pollination than their smaller^ ground-loving cohorts. One possible pollinator difference among the taxa observed could be the greater dependence of S. inteqrifolia var. integrifolia upon flies as pollinating agents (Table X). " Numerous authors have suggested that bees in particular show 128 Table XA Summary of major groups of insect visitors observed in 4 populations. Numbers indicate the total number of insect visitors recorded. No. Bees, % Bees, No. Total gasps Wa s£s pjpt era Visitors §jt Ofidula 14 32.53 29 43 (NOOS-B) 10.5 hrs.;4,5 April Sj, rufidula 10 19.0% 43 53 (NMO-B) 5.25 hrs.;21,22 April Sj. occidentalis 13 54.2% 11 24 (Yale-626) 2 hrs.; 19 April SA integrifolia 2 5.8% 3 2 34 (NHO-I) 4 hrs.; 10,11 May 129 less interest and constancy in a species as its freguency in the immediate flora declines (Brittain and Newton 1933, Stephens 1956, Simpson and Duncan 1956, Lewis 1961, Free 1963,1968, Levin and Anderson 1970) . Competition for Hymenopteran pollinators in the later flowering S. integrifolia var. integrifoliacculd be more intense judging from the greater variety and abundance of concurrently flowering species which are perhaps more attractive to bees than Saxifraga flowers. Slight selection pressure due to competition for pollinators may favor more pronounced seasonal differences in flowering time (WcNeilly and Antovonics 1967, Faterniani 1969) and greater self-compatibility within sympatric S. rufidula populations (Antonovics 1968). A shift in visitor spectra between sympatric S. integrifolia and 5. rufidula type plants, if further, more detailed observations confirm this speculation, may have been an important evolutionary influence in the development of such possibly fly-adapted structures as the enlarged nectar gland which presents diffuse, glistening, ultra-violet absorbing nectar {Percival 1965) as well as reduced petal size, papillae, and showiness in S. integrifolia var. leptopetala. The evolution of gynodioecy in S. integrifolia may also be correlated to such a pollinator shift if selection favors proportionately mere outcrossing events to compensate for irregular and unreliable fly visitors (Percival 1965, Faegri and van der Pijl 1966). 130 TAXQNGJ3Y SJBfcies Definition In This Complex The definition of a species in this complex group cf taxa must allow for overlaps in variability and the morphological intermediacy of many specimens. Hence, it must focus on the clusters of variability which exist around certain morphological characteristics. Isolating mechanisms which restrict gene flow are considered of secondary importance to morphological trends in arriving at a workable classification in this particular group., If the presence of intermediates and evidence of gene flow are used as main criteria for treating the taxa at a subspecific level, the resulting species are extremely large and variable with unmanageable ranges in morphology, ecology, cytology, etc, When these entities are included in regional floras, perhaps a functional approach would be to treat them as species aggregates in the main key with a separate, more precise key to the aggregate taxa. 131 Key To The Species The following key does not discriminate varietal or sufcspecific taxa within S.;inteqrifolia or S. oregana sensu Hitchcock and Cronquist (1973) or the taxa peripherally related to 5. occidentalis such as subspecies of S. marshallii, S. reflexa. or the S. nivalis-tenuis complex • sensu Krause and Beamish (1972,1973). Saxifraga integrifolia and S. oregana are not included in the following descriptions. A shorter, simplified key is presented in the Appendix. It is designed for identification of the majority of specimens, but many intermediate or hybrid specimens will not identify easily using that key. , A. Ovary less than half inferior at anther dehiscence and scapes ,5m or less tall (nearly to about half inferior in seme hybrid entities of S. occidentalis and S. gormanii: petiole distinct, usually more than 2x as long as wide; leaf nargins from shallowly sinuate-dent ate to, more commonly, rounded or square-dentate (if teeth markedly reduced or near absent see lead AA); petals usually 2.5mm or greater (sometimes small,greenish, reddish or lacking in alpine forms of S. ,occidentalis.C and S. aeguidentata.CC). B. Nectar qland at anthesis a swollen, cylindric or douqhnut-shaped rinq that almost covers the ovary; rhizomes forminq a fine network rarely present or as a few small reddish-brown fraqments in pressed material; leaves tapering abruptly into an elenqate petiole; ovary usually 132 from 1/3 to about 1/2 inferior at anther dehiscence, •,...............>,«.•••• •v..,.. Saxifraga gormanii 1. BB. Nectar gland reduced to a narrow band, ringing the ovary wall; rhizomes short, stout, with few branches, horizontal; leaves tapering more or less gradually from the blade into a petiole; ovary 1/3 or less inferior at anther dehiscence. C. Infloresence usually from conical to a tightly clustered headlike panicle; filaments usually clavate to sometimes narrowly oblanceolate; teeth on leaf margins sguare-dentate; petals usually tapered into a somewhat narrowed base; gland from inconspicuous to a somewhat broadened band. ................... ................... S. Occidentalis 2. CC., Infloresence usually flat topped to obtuse conical, not tightly clustered into a dense headlike panicle; filaments linear or subulate, rarely slightly clavate or oblanceolate in montane forms; teeth on leaf margins sinuate-dentate to sguarish-dentate; petals with a rather broad base; gland an inconspicous, narrow band. .... .. S. aegnidentata 3, ,• Ah. Ovary half or more inferior at anther dehiscence (becoming superior in fruit in some cases) or, if less than half inferior, then scapes exceeding .5 meters; petiole from almost lacking, to elongate and narrow; leaf margins entire or minutely denticulate to shallowly or unevenly sinuate-dentate; petals about 2.5mm or less (greater in 133 S. integrifolia* see DD and S. latipetiolata. £). D, Petioles almost lacking, short and broad usually less than 2x as broad as long; leaf blades tapering gradually into petiolar region; margins uneven undulate dentate to shallcwly and unevenly sinuate dentate; vestiture in inflorescence often with long, clear or merely pinkish glandular hairs., E, Scapes usually less than 0.5m; plants in moist grassy soil of Northwest Oregon Coastal Range mountaintop «balds". ..................................... S. latifietislata 4. EE. Scapes often 0.5m or greater; plants in boggy or swampy places from the Sierra Nevada north to the Cascades of central Hashington and west throughout the Rockies Mountains., ..........•................... ............. S. oregana. CD. Petioles elongate even if reduced in length, distinctly narrowed and evident; leaf blades tapering gradually or abruptly into petiolar region; margins various, commonly nearly entire; vestiture in inflorescence usually with dark reddish-tipped glands. 134 1» Saxifraga gormanii Suksdorf Torreya 23:106. 1923. (8.W. Gorman 4081, Elk Bock, Multnomah Co., Oregon, June 2,1917 HSI) S. marshallii f. dentata Engl. & Irmsch. Pflanzenr. IV,117,1:36. 1916. {Heller 10059, Elk Bock, near Oswego, Clackamas Co., Oregon) S. occidentalis Wats. var. dentata (Engl. & Irmsch.) CL. Hitchc. , Vase. PI. Pac. Borthw., Univ. Wash. Press, Hitchcock et al. Ill: 149. 1961. , Perennial, rosette-forming herb with dark, reddish, delicate, deeply-growing and branching rhizomes. Leaves simple, exstipulate; blades elliptic to ovate, tapering abruptly or somewhat gradually into a distinct petiole usually mere than 2x as long as broad, glabrous above, from near glabrous to sparsely rnsty villous below; margins from subentire (in some tetraploid introgressants with S. 'ntegrifolia var. claytoniifolia) to sinuate dentate; teeth nearest apex about 0.5 mm long. Inflorescence several-flowered (20-81), open, conical paniculate; rachis indefinite or evident. Flowers perfect, regular 5-merous; sepals spreading or ascending; petals about 2.5 mm long or longer, white, tapering gradually to a broad base, mostly deciduous in fruit; anthers orange to yellow; filaments linear to subulate (rarely clavate in intermediates with S. marshallii subsp. marshallii : carpels broadly pyriform at anthesis; gland swollen, doughnut-shaped, surrounding the upper portion of the ovary; ovary usually less than 1/2 inferior at anther dehiscence; gland remnant often a linear ridge encircling the fruit., (n=10,19,20) March to April., Bocky cliffsides in vernally wet seeps or moist places from Clatsop 135 County, Oregon east to Cowlitz Co, Washington and south to Lincoln and Marion Counties in Oregon. £S£E£§§J3iative Specimens OREGON: Clatsop Co.: exposed bare slope of main westernmost summit peak. Saddle Mt. State Park, on Saddle Mt., ca. 3225*, 3 June 1973, K.L. Chambers 3752 • (WT0,0BC); rock crevices and bluffs, near Astoria, 6 May 1933, G.P. Baker s.n.(QBE); Lincoln Co.: Otter Crest, 26 March 1930. anonymous s.n. I QBE): Tillamook Co.: exposed knobs, ridges above burned forest, gravelly slopes, Tillamook Burn area, N. Wilson Biver Hwy. , exactly 3 mi se. of Blue Lake, 3000*, 6 June 1S75, K.L. Chambers 4065 (OSC). The following specimens tend to r esemble S. inteqrif pli,a var. claytoniifolia in their qreater plant size {ca. 20 cm), more nearly half inferior ovaries, and reduced teeth: OREGON: Columbia Co.: damp rocky bluffs, about waterfalls, s. fork of Clatskanie Creek, 10 mi above Clatskanie, 15 May 1927, J..W. IMlJson 24j47 (STD) ; Lane Co.: Hill's Creek, 300', 29 May 1938, Dgtlina 2783 (ORE). This specimen is probably a hybrid between qormanii and •S. integrifolia var. claytonij.folia (pollen is 100% sterile): OREGON: Clackamas Co.; near Milwaukee, Elk Rock, 28 March 1885, anonymous 239 (OBE). 2. Saxifraga occidenta1is S. Wats. Proc. Am. Acad. 23:264 ,1888. §•,saximontana E. Nels. Erythea 7:168. 1899. Micranthes occidentalis Small, N. Am. Fl. 22 (2):14 4. ; 1905. 136 (J. Macoun. N.c. Yale Mt., B.C. May 17,1875. CAN) Micranthes saximcntana Small, N. Am. Fl. 22 (2) :145. 19C5. (A. & E. Nelson, 5917, Yancey«s, Yellowstone Natl. Park July 17, 1889 MINN!) Mer anthes la£a Small, N. Am. Fl. 22 (2): 145. 1905. (John Macoun, lytton, B.C. April 16,1889) Micranthes allenii Small, N. Am. Fl. 22(2):144.,1905. (O.D. Allen 242, Goat Mountains, Sash. June 27, 1896) S. lata Fedde, Just Bot. Jahresb. 31(1):613. 1906. S. allenii Fedde, Just Bot. Jahresb. 31(1):613. 19C6. S. microcarpa Johnson, Minn. St. PI. Sci. 4:25. 1923. (M.F• Elrod 98a, Missoula, Mont. MINN!) S. occidentalis var. wallowensis Peck, Leaf1. West. Bot. , 5:60. 1947. (Peck 18542, above Ice Lake, Wallowa Mts., Wallowa Co., Oregon July 4, 1894) S. occidentalis var. allenii C.L. Hitchc., Vase. Pi. Pac. Northw., Univ. Wash. Press, Hitchc. £l al., 111:49. 1961. Perennial, rosette-forming herb with short, stout, seldom-branching rhizome. Leaves simple, exstipulate; blades ovate to obovate tapering gradually or sometimes abruptly into a distinct petiole, usually glabrous above, or rarely puberulent, sparsely to distinctly rusty villous below; margins usually sguare dentate; teeth nearest apex usually ca. 5mm long or longer. Infloresence several-flowered (13-50), paniculate to densely clustered headlike paniculate, conical to spherical in more headlike individuals; rachis indefinite or distinct. Flowers perfect, regular, 5-merous; sepals spreading to souewhat reflexed; petals about 2.5 mm long, white, usually tapering 137 abruptly or gradually to a narrow, clawlike base; anthers orange to yellow; filaments usually at least slightly clavate to definitely clavate (although not petaloid); carpels elongate, bottle shaped at anthesis; gland a band-like ring encircling the ovary wall; ovary 1/3 or less (sometimes almost 1/2) inferior at anther dehiscence, Stylar beaks recurved in fruit; gland remnants usually inconspicuous, , (n= 10, 19, 28,29) April to August, Bocky outcrops, cliffsides and mountaintops adjacent to vernal streamlets, seeps, wet rock faces or moist areas in shallow soil (about 7 cm deep), widespread in the Bocky Mountain region and north Cascades from Washington, Idaho, and Montana north to Scutwestern Alaska and adjacent Yukon, ISlU§e£iaili§ Specimens ALEEBTA: rocky slope above Bertha Lake, Waterton Lakes Natl. Park, 21 June 1930, jB.-C. McCalla 3618 (MIN); ERITISH COLUMBIA: among rocks, Atlin Hot Springs, 2250', 5 July 1914, I.J. Anderson 2420 (V) ; Lillooet, 13 Hay 1916, E.JH. Andersfin 2419 (V); rock outcrops, Manning Park, Blackwall, 14 July 1960, K.I. Beamish. F. Vrugtman 60805 (V,UBC); calcareous foot of glacier, Cougar Valley, Selkirk Mts., 1600», 18 July 1908, F.K, Butters and E.W.D. Holway ,340 (MIN) : meadow, Mt. Bobson Park, Snowbird Pass, 9 Aug. 1975, C.C, Chuanjg 75J[399 (V) ; glacial moraines. Sphinx Glacier foreland. Garibaldi Park, 500 0*, 26 Aug, 1965, B, Eraser s.n. (UBC); rockslide above river on e. wall, Alexandria Bridge, 3500», 18 April 1934, E.T. HcCabe 758 (UC,WTU); Cronin Mt., 10 mi s. of Smithers, 1 July 1967, G. 138 Mendel 127 (V); wet crevices. Cathedral Lakes, Mt. Bcmford, Ashncla Distr., 7000', 12 July 1951, T.M.C. Taylor 1346 (8S.PC): Marble Mts., Lake Bootahnie, 5000', -20,25 June 1938, J.jjl. anji E.M. Thompson 70 (UC.WS.MIN); McComal Creek Quadrangle, ca. 2 mi w. of sw, corner of Thutade, Stikine Mt., 4500*, 14 June 1969, S.L. HeIsfa, K. Riqby 9109 (BB1). IDAHO; Kootenai Co.: June 1892, J.B. Hieberq s.n. (OBE); Custer Co.: among rocks in seepage from snowbank, common, Mt. ssw. of Alturus Lake, Stanley, n=10, 11 Aug. 1969* D.L. Krause 68 (UEC)MONTANA: Beaver Head Co.: top of Odell Peak, Pioneer Bange, 24 July 1946, C.L. Hitchcock and C.V. Muhlicjc 14925 (HS) ; Deerlodge Co.: ca. 18 mi s.w. of Anaconda on e. slope above trail between Storm Lake Pass and Goat Flat just on Goat Flat side of a large limestone outcropping, 9100», 7 July 1974. P. Elvander 443 (WTO). Glacier Co.: grassland on well drained soil, 2.6 mi n. of St. Mary, 4 July 1950, D. Lynch 6284 (HS) : NEVADA: Elko Co.: infreguent in crevices of rocks along stream bank, 1/2 mi above Thomas Canyon Campgrounds, La Moille Canyon, 15 June 1941, A.H. Holmgren 1122 (UC). OBEGON: Grant Co.: rocky slope, ne. of sumiit of Strawberry Mt., 8900»,1 Aug. 1953, A. 7703 (WS^HTD) : Wallowa Co.; steep exposed slopes of Pete's Point, 19 July 1962, G» lason 5428 (OSC). WASHINGTON: Snohomish Co. : under dripping cliffs, Mt. Dickerson, Cascade Mts.„ 5000», 17 July 1932, J.H. 2h£J£Son 8853 (WTO) ; Okanagan Co.: rocky outcrops, below Slate Peak at the head of the Slate Fork of the Paysatan Biver, 28 July 1940, J. Ownby., E.G. Meyer 2222 (HS,0BI,OSC,MIN,0C); Pierce Co.: rocky slope, Mt. Bainier Natl. Park, Crystal Mt. Indian Henry's, 2 July 1928, F.A. Warren 785 (WS); Skamania Co.: 139 shallow rocky soil. Sisters Bock, Columbia Mati. Forest, 4000*, 7 June 1945, D.C. Inqrahm .1877 (HS,OSC) ; Spokane Co.: damp rocky hillside, near Latah Creek, se. of Spangle, 12 flay, 27 June 1916, J.S. Suksdorf 8616 (WS); Whatcom Co.: shallow soil on rocky slopes at timberline, at. Baker, 6200', 11 July 1922, H.J..- ;>-Mason 3871 {DC). WYOMING: Grand Teton Natl. Park: moist mossy hillside. Cascade Canyon, 7500f, 19 June 1933. I. Williams J135 {CSC). 3. / Saxif raga aeguidentata (Small) Bosend. J.n Engl. Bot. Jahrb. 37, Beibl. 83:70, 1905. Saxifraga rufidula (Small) James Macoun Ottawa Nat., 20:162. 1906. JiSfantbes rjifidula Small, N. Am. Fl. 22(2) : 140. 1905. (John Macoun, Mt. Finlayscn, Vancouver Island May 17, 1887 HY) Micranthes aeguidentata Small. N. Am. Fl., 22(2):145. 1905. (Suksdorf 967, Lower Cascades, Skamania Co., Wash. WS) S. rufidula f. major Engl. 8 Irmsch,,Pflanzenr. IV,117,1:39. 1916. S. rufidula f. minor Engl. S Irmsch.,Pflanzenr. IV,117,1:39. 1916. '§.* Jtlickitatensis Johnson, Minn. Stud, PI. Sci, 4: 25. 1923. {Suksdorf, Klickitat Co., Wash. April 9 and May 1883 HS!) S. occidentalis Wats, subsp. rufidula Bacigalupi in Abrams, 111. Fl. Pac. St. 2:366.,1944. S. occidentalis var. rufidula Hitchc, Vase PI. Pac. Northw., Oniv. Wash. Press, Bitchc. E% al, , 111:49. 1961. Perennial rosette-forming herb with short, stout, few 140 branching, horizontal rhizomes. Leaves simple, exstipulate; blades elliptic to ovate, tapering somewhat gradually to abruptly into a distinct petiole, glabrous above, rusty tcmentose to rusty villous below; margins deeply sinuate dentate to somewhat shallowly sinuate dentate; teeth nearest apex usually greater than .5mm long. Inflorescence few- (4-42) to several-flowered (as high as 74), open, spreading, flat-topped, convex or obtusely conical; rachis usually indefinite. Flowers perfect, regular, 5-merous; sepals spreading or ascending; petals 2.5 mm long or longer, white, tapering gradually to a usually broad base, mostly deciduous in fruit; anthers from dark red to yellow; filaments linear or oblanceolate to slightly clavate in some montane forms; carpels narrow, bottle-shaped at anthesis; gland a narrow, inconspicous band-like ring encircling the ovary wall; ovary 1/3 or less inferior at anther dehiscence. Fruiting stylar beaks recurved; gland remnant inconspicuous in fruit. (n=10,19,28 ca, 28,29) Mid February to July. On shallow soil <ca. 3-4 cm) of rocky outcrops and cliffsides in vernally moist, often dripping seeps, washes or rivulets from the Upper Willamette Biver area north and east into the Columbia Biver Gorge, then found from the Olympic Mountains to east central Vancouver Island., Be presentative Specimens BBITISH COLOMBIA: Alberni, Vancouver Island, April 1914, Carter CI.59- (V); rocky bluffs, Cowichan Lake, Bald Mt. , Vancouver Island, 24 aarch 1940, I.M. Cowan s.n. (V); in 141 crevices of rocks, soil damp, n. end of Shawnigan Lake, Vancouver Island, (n=10), 20 Feb. 1971, D.L. Krause. 1-71 (UBC); mossy wet rocks. Hill Hill, Vancouver Island, 17 Karch 1895, J. B. Anderson 84 (V). CBEGON: Clackamas Co.; Elk Bock Cliffs 2 mi n. of Oswego Lk., (n=10) , 17 Harch 1971, D.L. Krause 9-7.1 (UBC,UBC) ; Hood Biver Co.: open basaltic knolls with Selagjnella wallacij, 12.5 mi w.of Hood Biver along portion Old Columbia Biver Hwy., 11 April 1958. CL. Hitchcock and C. V. Mul'ck 21501 (SS) ; Lane Co.: steep, gravelly, wet soil, G»Leary Htn., 2800», 28 June 1938, L.E. Detling 30.82. (OBE) ; Linn Co.: moss mats on e. facing cliff, Santiam fi., 20 mi e. of Sweethome, 7 April 1951, JU Cron^uist 6828 (HS); Marion Co.: cliff. Silver Creek Falls, 9 April 1940, M. Hrijht s.n., (OSC); Multnomah Co.: on rocks and rocky cliffs near Elk Bock, 12 April 1903. M.». Gorman s.n. (herbarium no. 22637) (WTB); along Sandy Biver at junction of e., Starke Bd. and e. Columbia Hwy., 2 2 March 1926, H.E. J?£ck 1452J[ (OSC); moist slopes, near Multnomah Falls, 18 April 1935, J.jf. Thompson 17370 (BS) ; Uasco Co.; wet cliffs, 6 mi w« of The Dalles cn Columbia Biver Hwy.,(pollen fertility, sterile-14, fertile-190, E. Perkins), 27 March 1946, H.H. Baker 274 (OSC). WASHINGTON: Clark Co.: on wet rocks beside road, 1 mi w. of Clark-Skamania Co. line, St. Hwy. 14, n=10, 14 March 1971, D.L. Krause 2-7J (DEC); Greys Harbor Co.: wet cliffs of Mt. Colonel Bob, 3500», 12 July 1930, J.JJ. Thompson 9404 (PC) : Jefferson Co.: rocky crest of Constance Bidge, 5500*, 30 May 1931, J.H., Thompson 6583 (OSC); Klickitat Co.: 3 mi e. of Bingen, 300«, 26 April 1950, L.S. Rose 50073 (OC); Mason Co.; rock outcrops where protected in narrow canyons, near summit of 142 fit. Elinor, 11 June 1940, E.G. Jejer J783 (§S) ; Skamania Co.: wet cliffs. Cape Horn, 10 April, 27 May 1920, 1. Suksdorf 10365 (WS,UC) . The following specimens, usually of higher mountain areas, show slight resemblances in clavate filaments and clustered paniculate inflorescences to S. occidentalis: BRITISH COLUMBIA: Strathcona Park, Mt. Booster Comb, Vancouver Island, July 1937, N.C. Stewart 10481 (V). WASHINGTON: Jefferson Co.: ridge, head of Dosewallips B., 6000*, 29 July 1921, l.P. Taylor s.n. (UCJ; Clallam Co.: moist rocky banks, Ht. Angeles, 5000*, 16 July 1931, J.W. Thompson 739S (UC, OSC). These specimens approach S. marshallii- in their clavate filaments, longer pedicels and often reflexed sepals: OREGON: Linn Co.: e. facing cliff along Santiam R. 20 mi e. of Sweethome, 7 April 1951. A. Cronquist 6828 (UC) : Lane Co.; steep n. slope, moist, Mt. 0»Leary, 4 800*, 28 June 1938. L.E. Petling 3082 (UC); Marion Co.: cliff, top of House Mt., 31 May 1926, M.J. Peck 14638 (OSC). These specimens are tenatively classified here as S. howellii Greene but this taxon is doubtfully distinct from S« rufidula. Further work is necessary on this problem. OREGON: Josephine Co.: dried up but lately moist bluffs. Eight Dollar Mt. near Selma, 26 March 1926, L.F. Henderson .5845 (QBE) ; Douglas Co.: rocky hillside, thin soil, 10 mi Beston Rd., 24 Feb. 1973, M. Williams s.n. (OBE) . 4. Saxifraga latipetiolata (C.L. Hitchcock) Perkins and Elvander. 143 S. occidentalis var. latipetiolata C.L. Hitchc.. Vase. PI. Pac, Northw., Univ. Hash. Press, Hitchc. et al., 111:49. 1961. (fl.H.Gorman 3561, Saddle Mt., Clatsop Co., Oregon, June 20, 1915. WTU!) Perennial, rosette-forming herb with short, stout, seldom branching rhizome., Leaves simple, exstipulate blades ciliate above, sparsely long villous below, widely ovate, tapering gradually to a short, broad, ciliate, petiolar region, puberulent above and below to faintly rusty sericeous below; margins dentate to undulate erose. Infloresence usually many-flowered (46-231), congested, conic panicle, to a somewhat congested, paniculate (occasionally corymbiform) head; central rachis evident to indistinct. Flowers perfect, regular, 5-merous; sepals reflexed; petals about 2.5 mm long or longer, white, tapering gradually to a broad base, persistent into fruit; anthers yellow; filaments linear; carpels obconic or umbonate obconic; gland a flattened disc covering the top cf the ovary at anther dehiscence, grading into the stylar tissue; ovary 1/2 or more inferior at anther dehiscence, becoming superior in fruit. Fruiting stylar beak reflexed; gland remnant a linear ridge encircling the fruit. (n=ca. 38) Late May to early July. Shallow sell of higher volcanic plugs and mountaintop "balds" in moist, grassy areas of Clatsop and northern Tillamook Counties, Oregon. fie prese nt atjve Specimens OBEGGN: Clatsop Co.: rocky slopes. Saddle Mt.,2800-3300*, 144 20 June 1915. a.l. Gorman 3561ISOTYPE (WS); moist open slopes. Saddle Mt., 28 June 1952, J.T. Howell s.n. (0C); Douglas fir-spruce forest, rock crevices nw. exposure, moderate shade, Saddle Mt., 2200', 19 June 1932, L.E. Betling 7906 ( ORE). The following are classified as S . i nt eg r i f ol i a but resemble S. latipetiolata in leaf shape, broad short petioles, and white pubescence cn the upper leaf surface: OREGON: Pclk Co.: very wet places, Monmouth, 20 May 1893 . W.J. Spil1man 78 <WS); Marion Co.: gravelly soil, common, Salem, Brooks Pasture, 5 April 1919, fl.W. Gorman 4J4JJ (WS) ; WASHINGTON: Thurstcn Co.: Rock Prairie, 12 May 1934, I.£. Q$i§ JL§93 (IS) . cojsayjsioNS The present studies indicate that a reassessment cf the S» occidentalis species complex is necessary. The varietal taxa which Hitchcock et al. (1961) and Hitchcock and Cronquist (1973) list for S. occidentalis. namely rufidula, dentata, and latipetiplata, are not merely entities which represent distinct evolutionary trends within S. occidentalis --1. Jhejsr are IS* PJipJ^SJ-SaAA X# 9S030fifei£§llX# and-- cy^plpg^ separated over much of their ranges. Althpush character inter<jr.adation does occur among some taxa in certain areas, esjpecially in .polyploid individuals and populations, the ^|.^t4]|.qtjQBS;-;b.ased-,on-certain combinations of characters are sufficient tp maintain each of these, J.anx- --.tlsej-taxenoiiic -^dif f 4euitiej§^®sn.a----these -ta xa can be attributed to hybridization and a 11 pppiyg 1oidy involving various members of the S. inteqrifolia species complex. Autopolyploid evolution in this grcup appears to be of less importance but it is possible that nearly identical diploid and tetraploid or tetraploid and hexaploid S.. aegui den tat a plants from the same location have arisen directly and without hybridization with other taxa. Saxifraga aeguidentata is confirmed as a cytologically and morphologically variable species in agreement with several previous taxonomic treatments (Small and Bydberg 1905, Hacoun 1906, Engler and Irmscher 1916, Krause and Beamish 1S73). Some individuals from the Columbia Biver Gorge area resemble S. occidentalis morphologically but the resemblance may be the result cf complex hybridization and polyploidization probably 146 involving S. integrifolia var. claytoniifcliaor its relatives in that area. There is evidence that some introgression between S. integrifolia var. claytoniifolia and S. aeguidentata is occurring in sympatric populations along the Columbia Biver Gorge. The tetraploid S. occidentalis genome is also probably the result of ancient hybridizations and polyploidy perhaps between S. occidentalis and sympatric S. j.ntegrifpjLj,a progenitors similar to S. integrifolia var. leptopeta•la. Even if resemblance is the result of past contact and introgression with S. pccidentalis it appears that S. aeguidentata is presently geographically isolated and genetically distinct with several characteristic morphological features. Artificial hybrids between S. aeguidentata and S. occidentalisare pollen and seed sterile and show no greater cytological similarities than crosses involving presumably more distantly related entities. Numerical treatments also tend to distinguish a S. aeguidentata group. Columbia Biver Gorge S. aeguidentata populations show closer similarities to S. occidental's and other polyploid or hybrid individuals than do S. aeguidentata plants from Vancouver Island and the Olympic Mountains. The relationship between S. aeguidentata and the morphologically siiilar and cytolcgically unknown entity, S. howellii. of southwestern Oregon and northwest California deserves further study. Saxifraga occidentalis is defined as a variable, montane taxon which shows evidence of hybridization in several areas of its range. Detailed examination of its relationships with S. nivalis and S. tenuis as well as with S. marshallii subsp. 147 Mishallii and subsp. idahoensis is necessary. Further work using artificial hybridizations among diploid S. occidentalis, S. ref lexa. S. nj.vaJJLs, subspecies of S. .ffiargfoaAlii- diploid S. aeguidentata. and S. gormanji would be a useful addition to the present study. , A major problem with such studies would be the strong seasonal separation in flowering times among these plants which are adapted to flowering regimes in diverse altitudes and latitudes. Saxifraga gormanii Suksdorf, formerly S. occidentalis var. is treated here as a separate species frcm S. occidentalis and S. aeguidentata. formerly S. rufidula•, with a distinguishable morphology and a geographical distribution in the Coast Range and Lower Willamette River of Oregon extending north to Clark County, Washington. The type location is Elk Rock, Oregon and specimens frcm that site closely resemble nearby tetraploid plants (n=20,19). Tetraploids group with diploid populations in the numerical studies. Artificial hybrids between tetraploid S. occidentalis and tetraploid S. germanii are sterile and consistently fail to undergo meiotic divisions in anther tissue, in contrast to crosses involving S. ajSIiajaii and S. aeguidentata where meiosis in the F1 hybrid does occur but is irregular and pollen fertility is low., As a group, S. gormanii shows closer phenetic affinities to varieties of S. integrifolia and S. latipetiolata. formerly S. gccidentaljs var. latipetiolata. than to S. ocjej, dental is or S. aeguidentata . Some specimens tend toward S. marshallji subsp. marshallii and further studies, especially in the Willamette River area are needed to clarify this problem. 148 Evidence is presented that S. latipetj.olata (CL, Hitchcock) Perkins and Elvander is also a separate species from S. occidentalis., Saxifraga latipetiolata has several similarities to S. oregana and the S. inteqrifolia-complex. Morphological, cytogical <n=ca.38), and numerical analysis all indicate a hybrid origin and affinities with S.oregana (n=3 8) . Other systematic evidence from studies of the S. integrifolia group* S. rhomboidea, and Californian relatives of S. inteqrifolia (Elvander, personal communication, 1978) confirm that it is probably more closely related to S. oregana. However it is ecologically and geographically isolated from S. oregana and has a number of distinctive morphological features. Therefore it is treated here as a separate species. Sterile intermediates occur in intermediate habitats in most areas where S. aeguidentata comes in close syapatric contact with S. integrifolia or S. integrifolia var. claytoniifolia. These can be recognized morphologically and cytclogically or by studies of pollen fertility. Hybrid swarms occur in several areas, seme of which may involve S. aeguidentata and S, jntegrajglia var, integrifolia as parental entities but others are more likely the result of previous contact between S. occidentalis or S. marshallii subsp, idahoensis and S. integrifolia ?ar, , leptopetala. Cryptic natural hybrids which resemble one parental entity and undergo abnormal meiosis or exhibit reduced fertility are not uncommon, especially in the Columbia Hiver Gorge area among populations of mixed diploid and tetraploid or tetraploid and hexaploid S. aeguidentata plants, but also in ether areas and 149 for other taxa such as S, inteqrifolia var. claytoniifolia. It is difficult to tell whether these individuals, especially in the Columbia aiver Gorge, are the result of crosses between two siblinq entities with different ploidy levels or segregates of interspecific crosses which closely resemble the parent in morphology and habitat preferences., One population west of Chehalis, Washington (EP605), (n=29) is apparently an intermediate between S. occidentalis and S« inteqrifolia. The most reasonable treatment of this problematical population appears to be to classify it with §• occidentalis and further document its unigue background. The possibility cannot be excluded that it is a relictual population of a once more widespread entity. Its apparent close relationship to S. inteqrifolia plants from the Mima and Ft. Lewis, Washington, region needs further investigation. It is treated here as belonging to S, occidentalis. The general pattern of polyploid evolution in the group has most likely been one of several probably independent diploid hybridizations followed by chromosome doubling and coupled with aneuploid reduction (in the absence of an available nine-paired parental species). This cycle has apparently repeated itself in several areas to give rise to higher ploidy levels and introgressant forms. In S. aeguidentata populations which exhibit mixed ploidy levels, higher levels of polyploidy may have occurred through the combination of rare unreduced gametes from the same population but hybridization with other taxa, followed by chromosome doubling is more likely. A speculative chart of the relationships involving S. aegui.deBta,|a and its 150 allies is presented in Figure 36. The combined difficulties of limited artificial hybrid combinations, possible differences in genetic control of synapsis, and the possibility of extinct parental entities limits the precision of any such speculation. Bagging tests demonstrate that the plants investigated are most likely facultative sexual outcrossers. There is little evidence in favor of apomictic seed production. The somewhat earlier flowering members of the occidentalis group { S. S£gJdentajLj.s, S, aeguidentata. and S, gormanji) are capable of setting more autogamous seed and may be frequented by relatively higher numbers of Hymenopteran versus Dipteran pollinators than the integrifolia relatives (S. integrifolia var. istegrifolia, S. integrifolia var. claytoniifolia. §* JB£eggJf93-ia var. leptopetala. and S. latipetiolata). These observations may be correlated with the general differences in floral morphology between the two groups as well as the occurrence of gynodioecy in certain populations of S. intearifolia var., integrifolia. Although the results of bagging tests and pollination studies show interesting differences between the major species complexes studied, within the morphologically similar S. Qcc'dentalis-aeguidentata species group, they are less useful as a taxonomic tool. The post-glacial history of the group is complex and apparently varied. Bandhawa and Beamish (1972) reviewed the evidence for glacial refugia in northwestern North America. The refugial areas apparently contain relictual diploid populations in contrast tc widespread polyploid colonizers of surrounding glaciated areas, Bandhawa and Beamish (1972) used the 151 Figure 36: Speculative polyploid formations within the S. occidentalis and S. aeguidentata (the synonym- rufidula is used in the figure) lineage with emphasis on the Columbia River Gorge relationships. Several other constructions are also logically possible. Numbers in parenthesis are not supported in the literature. Joining lines indicate possible allopolyploid relationships but in certain instances autopolyploidy is probable {CF. S. reflexa). For most others, allopolyploid origins are possible but less likely. 152 ntegrifolia ca 55 SPECULATIVE RELATDNSHIPS oregana latipetiolata 38 38 occidentalis 29->28 rufidula 29^28 integrifolia (20)-> 19 leptopetala. (20) ->19 / occidentalis gormanii 19 "20-> 19 rufidula . (20)^19 (integrifolia) 10? rufidula 10 Proto integrifolia (10) Proto occidentalis (10) 153 distribution of 10 and 19 paired populations of S. ferruginea as an example of such a pattern. The narrow southern and western distribution of diploids and widespread ranges of polyploids within S. occidentalis and some members of S. integrifolia conform well to the pattern exhibited by S. ferruginea. Although diploid S. aeguidentata populations have recclonized formerly glaciated areas of the Olympic Mountains and Vancouver Island, polyploids and introgressants are abundant in areas south cf the glacial boundary presumably resulting from contacts between coastal and interior floristic elements in the Columbia River Gorge. Similar complex hybridization and polyploidization processes are also evident in the history of S. gormarii and S. latipetiolata. Zones of complex hybrid activity such as the Columbia River Gorge may provide the raw materials in the form of more variable gene pools upon which selection can operate in future glacial or interglacial epochs (Stebbins 1571). Hexaploid and octoploid populations in restricted areas may represent the remnants of a once more widespread distribution or such restriction may be the result of narrow ecogeographic sympatry between two tetraploid or tetraplcid-diploid progenitors which have since become locally extinct. The absence of S. oregana from the Coast Range of Oregon appears tc be in the latter category with respect to its probable parental relationship to S. jLatjpetiolata. A thorough examination is necessary of S, integrifolia and its relatives including S. ., oregana. S. rhomboidea, and S. californica, the latter two perhaps having rather closer ties with S. occidentalis and S. marshallii respectively. There is a 154 strong need for a systematic study of the relationships between disjunct Western and Eastern North American species pairs (Spongberg 1972) such as S. occidentalis-5. virginiensis, §• oregana-S. pensylvanica. S. ferrugjnea-S. micbauxii. S. reflexa- S. micranthidifdjia and S. mags[bap. 1ji-S. careliniana . Chromosome counts are incomplete for many taxa, especially those where the field season conflicts with institutional schedules and those which are less accessible. Attention should be given tc further systematic investigation of hybrid swarms and areas of introgression in the Opper Willamette Biver area, the Wallowa Mountains df northeastern Oregon, the Spokane area of eastern Washington, and the higher mountain areas of northern California and southern Oregon.. 155 LITERATURE C££ED Alexopoulos,C,J., and E.S.Beneke. 1952* Laboratory manual for introductory mycology. Burgess Publishing Co.,Minneapolis, Minn. Antonovics,J. 1968. Evolution in closely adjacent plant populations. V. Evolution of self-fertility. Heredity 23:219-238. Baciagalupi,B. 1944. Jn L. Abrams. Illustrated flora of the Pacific States. Vol, II. Stanford University Press, Stanford, California, Beamish, K.I. 1961. Studies cf meiosis in the genus Saxifragaof the Pacific Northwest. Can. J. Bot, 39:567-580. Beamish, K.I. 1967. A Pacific Coast Saxifrage with 10 pairs of chromosomes: meiosis, development of the female gamete, and seed production. Can, J. Bot. 45:1797-1801. Bloom, ».L. 1976., Multivariate analysis of the introgressive replacement of Clarkia nit ens by CjLajckja specicsa pel vantha (Onagraceae). Evcluticn~30:4*12-4 247" Brittain, H.H, And D.E. Newton. 1933. A study in the relative constancy of hive bees and wild bees in pollen gathering. Can. J. Bes. 9:334-349., Carr, G.R. 1975, Chromosome evolution and reduction in Calycadenia pauciflcra (Asteraceae). Evolution 29:681-699, Chambers, K. 1974. Notes on the flora of Clatsop County, Oregon. Madrono 22:278-279. Crcvelle, T.J, 1970. Analysis of character variation in ecology and systematics. Annual Bev, Ecol, Syst, 1:55-98, Damiolt, J, 1968, Zur cytotaxonomic der gattung Saxifraga L, III. Berlin Deut. Bot. Ges. 84:43-52. Dambolt, J., and D. Podlech. 1965. Zytotaxonomische untersuchungen an Saxifraga-Sippen der grex Exarato-moschatae Engl, et Irmsch.,Berlin deut. Bot. Ges. 77:332-339. Danick, B.P. And B.V. Burns. 1975. Multivariate analysis of hybrid populations. Natur. Can. 102:835-843. 156 Dixon, 8.J. 1970. B.M.D. biomedical computer programs. Health Sciences Computing Facilities, Department of Preventive Medicine, University of California at Los Angeles. Don, D. 1822. A monograph of the genus Saxifraga. Trans. Linn. Soc. 13:341-452. llvander, P.E. 1975. Biosystematic studies of a new species of Saxifraga and its relatives, MSc. Thesis. University of Washington, Seattle. llvander, P.E. 1978. Systematic relationships within the S. integrifolia complex (Saxif ragaceae). Abstract In Bot. Society of America, Misc. Series Publ. 156:82. Engler, A., and E. Irmscher. 1916. Das Pflanzenreich. Ed. A. Engler. Wilhela Engelmann, Leipzig IV, 117, I. Eyles, A. C. , and 8. E. Blackith. 196 5. Studies on hybridization in Scolopostethus Fieber {Heteropteraj lygaeidae). Evolution 19:465-479. Faegri, K,, and L. van der Pijl, 1966. The principles of pollination ecology. Pergamcn Press, Oxford. Free, J.B. 1963. The flower constancy of honeybees. J. Anim. Ecol. 32:119-131. Free, J.B. 1968. Dandelion as a competitor to fruit trees for bee visits. J. Appl. Ecol. 5:169-178. Hagerup, 0. 1950. Rain pollination. Kgl. Danske Vid. Selsk. Biol. Meddel. 18:1-19. Hagerup, 0. 1951. Pollination in the Faroes—in spite of rain and poverty of insects. Kgl. Danske Vid. Selsk. Biol. Meddel. 18:3-48. Henderson, D.M. 1976, A biosystematic study of Facific Northwestern blue-eyed grasses tsisyrinchium. Iridaceae). Brittonia 28:149-176. Hitchcock, C.L., A. Cronquist, M, Ownbey, and J,a. Thompson. 1961. Vascular plants of the Pacific Northwest. University of Washington Press, Seattle. Hitchcock, C.L. , and A. Cronquist. 1973. Flora of the Facific Northwest. University of Washinqton Press, Seattle. Hooker,J. 1833. Saxifraga inteqrifolia. Flora Boreal! Americana 1:249. 157 Hossain, M.G. 1977. The significance of chromosome association in an advanced population of tetraploid rye. Can. J. Genet. Cytol. 18:601-652. Hunan, C. 1974. A barefoot doctor's manual (transl. ..of Chinese). U.S. Dep. of Health, Education and Welfare, Public Health Service, Nat. Inst, of Health. Hyde, H.A. 1950. Studies in atmospheric pollen. IV.,Pollen deposition in Great Britain.,New Phytol. 49:398-420. Hyde, H.A. 1969. Aeropalynology in Britain--an outline. New Phytol.,68:579-590. Hyde, H.A., and D.A. Williams. 1961. Atmospheric pollen and spores as causes of allergic disease: hay-fever, asthma, and the aerospora. Advancement Sci. 526-533. Jackson, B.C. 1962. Interspecific hybridization in Haplopappus and its bearing on chromosome evolution in the Blepharodon section. Amer. J. Bot, 49: 119-132. Johnson, A.M. 1923. A revision of the North American species of the section Boraphila Engler of the genus Saxifraga (Tournef. }L. Minn. Stud. Biol. Sci. 4: 1-109. Krause, D.L., and K.I. Beamish. 1972. Taxonomy of Saxifraga occidentalis and S. marshallii. Can. J. Bot. 50:2131-2 141. Krause, D.L., and K.I. Beamish. 1973. Notes on Saxifraga occidental's and closely related species in British Columbia. Syesis 6:105-113.: Kyhos, D. 196 5. The independent aneuploid origin of two species cf Chaenactis (Compositae) from a common ancestor. Evolution 19:26-43. ~ Levin, D.A., and W.W. Anderson., 1970. Competition for pollinators between simultaneously flowering species. Amer, Natur., 104: 455-467. Lewis, H.; 1961. Experimental sympatric populations cf Clark'a. Amer. Natur. 95:155-168. Lewis, H., and P. Baven, 1958. Bapid evolution in Clarkia. Evolution 12:319-336. Lewis, W.H.,and Y. Suda. 1976. Diploids and polyploids from a single species population: temporal adaptations., J. Hered. 67:391-393. 158 Linnaeus, C. 1753, Species plantarum. First ed. Stockholm. Macoun, James 1906., Contributions to Canadian botany. XVIII. Ottawa Nat. 20:162-171. McNeilly, T., and J. Antonovics. 1967. Evolution in closely adjacent plant populations. IV. Barriers to gene flow. Bered. 23:205-218. Miller, J.M. 1976. Variation in populations of Claytonia perfoliate (Portulacaceae). Syst. Bot. 1:20-34. Moore, fi.J. 1959. In Calder, J.A., and D.B.C. Savile, Studies in Saxifragaceae, II. Saxifraga sect. Trachyphyllum in North America. Brittonia 11:228-249.. Namkoong, G. 1966. Statistical analysis of introgression. Biometrics 22:488-502. Crnduff, B. 1969. Beproductive biology in relation to systematics. Taxcn 18:121-133., Packer, J.G. 1968. IOPB chromosome number reports. Taxon 17:287. Paterniani, E. 1969. Selection for reproductive isolation between two populations of maize, Zea mays L. Evolution 23: 534-547. Percival, M.S. 1965. Floral biology. Pergamon Press, Oxford. Proctor, M., and P. Yeo.. 1973. The pollination of flowers, Collins, London. Bajhathy, T., and H. Thomas. 1972. Genetic control of chromosome pairing in hexaploid oats. Nature (London) 239:217-219. Bandhawa, A.S., and K.I. Beamish. 1972. The distribution of SaxjLfraga ferruginea and the problem of refugia in northwestern North America. Can. J. Bot. 50:79-87. Biley, B. 1960, Diploidization of polyploid wheat. Hered. 15:407-429. Biley, B., and C, N, Law, 1965, Genetic variation in chromosome pairing. Advances in Genet. 13:57-114. Bising, J.D. 1968. A multivariate assessment of interbreeding between the chickadees Parus atricapillus and £, carolinensi. Syst. Zool. 14: 131-132. ~ 159 Sass, J.E. 1958, Botanical microtechnique, 3rd ed. The Iowa State College Press, Ames, Schilling, E.E,, and CB. Heiser. 1976. Re-examination of a numerical taxonomic study of Solanum species and hybrids. Taxon 25:451-462. Scbueler, F.8., and J.D. Rising. 1976. Phenetic evidence of natural hybridization. Syst. Zool. 25:283-289. Sax, K. 1937. Effect of variation in temperature on nuclear and cell division in Trandescantia. Amer. J. Bot. 24:218-225. Singh, I.S. 1975. Effects de la temperature et du niveaux floral sur la fertilite' pollinique chez la petunia. Ann. Amelior. Plant. 25:365-37 0. Simpson, D.H., and E.N. Duncan. 1956. Cotton pollen dispersal by insects. Agron. J. 48:305-308, Skcvsted, A. 1934, Cytclogical studies in the tribe Saxifrageae, Dansk. Bot. Ark. 8: 1-50. Small, J.K., and P.A. Rydberg. 1905. Saxifrag aceae. In N. Amer. Flora 22:81-158. Smith, D.H, 1969, A taximetric study of Vacciniu n in northeastern Ontario. Can. J. Bot.,47:1747-1759. Snow, R. 1963. Alcoholic hydrochloric acid-carmine as a stain for chromosomes in sguash preparations. Stain Techncl. 23:9-13. Sckclovskaya, A.P. 1958. On the correlation between the number of chromosomes and the size cf pollen grains in the Arctic species of Ranunculaceae and Saxifragaceae. (In Russian). Bot. Zhur. 43:1146-1155. Spongberg, S.A. 1972. The genera of Saxifragaceae in the Southeastern Onited States. J. Arnold Arboretum 53:409-453. Stebbins, G.L. 1971. Chromosomal evolution in higher plants. Edward Arnold Ltd., Icndon, Stephens, S.G. 1956, The composition of an open pollinated segregating cotton population. Amer. Natur. 90:25-39. Subhasi, U. 1975, Interspecific hybridization and aneuploidy in the genus Nicotiana. Cytologia 40:735-741. 160 Taylor, B.L., and G. A. Mulligan. 1968. Flora of the Queen Charlotte Islands, Pt. 2. Cytological aspects of the vascular plants. Plant Research Institute, Central Experimental Farm, Ottawa, Ont. Ter-Avanesian, E.y. 1978. Siginificance of pollen amount for fertilization. Bull. Tor. Bot. Club 105:2-8. Tobgy, H.A. 1943.„ A cytological study of Crepis fuliqinosa. £• neglecta. and their F1 hybrid, and its bearinq on phyloqenetic reduction in chromosome number. J. Genet, 45:67-111. Veldman, D.J. 1967. FORTRAN proqramminq for the behavioral sciences. Holt, Rinehart, and Wilson. Ward, J.H. 1963. Hierarchical qroupinq to optimize objective function. J. Amer..Statistical Assn. 58:236-244. Watt, G. 1972. A dictionary of the economic products of India. VI. Pt. 11. Reprint. Periodical Experts, Vivek Vihar, Shandana, Delhi.„ Welsh, S.L. 1974. Andersen's flora of Alaska and adjacent parts of Canada. Briqham Young University Press, Provo, Otah. Whitehouse, B.N.H. 1969. Ah application of canonical analysis to plant breeding, Genetica Agraria 23:61-96, 161 APPENDIX Insect Specimens List of insects captured while visiting Saxifraga flowers: P=with gaxifraga pollen on body or legs; (No.)-number of visitors collected. HYMENTOPTEBA: Vespidae P(2): Polistes fuscatus (1) ; Formicidae (1); Apidae: A|is melifera P(3): Bombidae: • Bprobus spp. P (2) ; Diprionidae (1); Andrenidae Pf11). COLEOPTEBA: Elateridae (1). DIPTEBA: Anthomyidae (3): Scatophaga spp., P(6); Tachinidae P(8); Bombyliidae: Bombvlius major P(2); Empidae P(7);Calliphoridae (1); Muscidae: Musea- dpmestica (1); Ceratopogonidae (2); Mycetophilidae P<2); Chironomidae {H); Syrphidae (3): Metasyrphus spp. (5) ; Dasvsvrohus sp. (1); Sphaerophoria sp. (1) ; Agrcmyzidae (1) . 162 Figure 37: Map of distributions of S. occidentalis (in Oregcn, Washington, and southwestern British Columbia only) and S* gcrmanii. The single sguare is the hexaploid intermediate population <S. occidentalis x S. integrifolia). , 164 Figure 38: Hap of the distribution of S, aeguidentata (the synonym, S. rufidula is used in the figure) and S. latipetiolata. Dotted line indicates the western limit of S. oreqana and morphologically related !int. 166 ABBREVIETED KEY A. Ovary less than half inferior at anthesis; leaf margins shallowly sinuate-dentate to evenly dentate B. Nectar gland a doughnut-shaped ring at anthesis; rhizomes, when present, a fine, reddish brown network S_; 'gormanii BB. Nectar gland reduced to a narrow band; rhizomes short, stout, horizontal C. Infloresence flat-topped or obtuse-conical, filaments subulate S_. aeguidentata CC. Infloresence not flat-topped, usually a clustered, headlike to narrowly conical panicle, filaments clavate S_. occidentalis AA. Ovary half or more inferior at anthesis; leaf, margins entire to unevenly sinuate-dentate (distantly serrate in S_. oregana) D. Petioles almost lacking, grading into blades; vestiture in infloresence with long, clear or pinkish-glandular hairs E. :Scapes usually 0.5m or. .less-. .^S. latipetiolata EE. Scapes usually more than 0.5m S_. oregana DD. Petioles elongate, broadening more or less abruptly into blades; vestiture in infloresence with reddish glands S_. integrifolia 


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items