Systematic and phylogeographic implications of molecular variation in the western North American roseroot, Rhodiola integrifolia (Crassulaceae)

(1)

Systematic and phylogeographic implications of molecular

variation in the western North American roseroot,

Rhodiola integrifolia (Crassulaceae).

by

Heidi J. Guest

B.Sc. University of Victoria, 2001

Thesis submitted in Partial Fulfillment of the Requirements for the Degree of

MASTER OF SCIENCE

in the Department of Biology

(2)

Systematic and phylogeographic implications of molecular

variation in the western North American roseroot,

Rhodiola integrifolia (Crassulaceae).

by

Heidi J. Guest

B.Sc. University of Victoria, 2001

Supervisory Committee

Dr. Geraldine A. Allen, Supervisor (Department of Biology)

Dr. Barbara Hawkins, Departmental Member (Department of Biology)

Dr. John Taylor, Departmental Member (Department of Biology)

Dr. Ken Marr, Additional Member (Royal British Columbia Museum)

(3)

Supervisory Committee

Dr. Geraldine A. Allen, Supervisor (Department of Biology)

Dr. Barbara Hawkins, Departmental Member (Department of Biology)

Dr. John Taylor, Departmental Member (Department of Biology)

Dr. Ken Marr, Additional Member (Royal British Columbia Museum)

Abstract

The roseroot genus Rhodiola is widely distributed in arctic and alpine areas of the Northern Hemisphere. It is most speciose in the high mountain ranges of central Asia.

Rhodiola integrifolia occurs at high altitudes and high latitudes in western North America

and northeastern Asia. During the Pleistocene glaciations the region between Asia and North America known as Beringia was ice free and acted as a glacial refugium for cold-adapted taxa. I surveyed variation in a nuclear (ITS) and chloroplast (psbA-trnH spacer) DNA region in R. integrifolia and its North American relatives, R. rosea and R.

rhodantha. Phylogenetic analyses based on ITS showed that (i) the western North

American species R. integrifolia and R. rhodantha are distinct but closely related sister taxa; and (ii) these two species and the eastern North American R. rosea belong to separate clades within Rhodiola. Analyses of the plastid region showed that although the

(4)

sister species R. integrifolia and R. rhodantha are distinct, some populations sampled in the southern Rocky Mountains (where the two species overlap) share psbA-trnH

haplotypes, suggesting that they hybridized at some time in the past. Within R.

integrifolia, both nuclear and plastid DNA regions showed strong north-south patterns of

differentiation, a pattern consistent with western North America’s glacial history. Restriction site analysis and sequencing of the plastid psbA-trnH spacer region from samples from 66 populations of R. integrifolia revealed 12 restriction-site haplotypes and 28 sequence haplotypes. A few of the sequence haplotypes were widely distributed, but most were relatively localized. Of the localized haplotypes, 10 were exclusively

Beringian and an additional four were found along the northern boundary of glaciation (at the last glacial maximum) in the Yukon and Alaska; two haplotypes were found in

northern coastal BC (Queen Charlotte Islands and adjacent mainland), in the vicinity of possible glacial refugia on the Queen Charlotte Islands. Only five haplotypes occurred exclusively south of the glacial maximum. Haplotype diversity in R. integrifolia decreased toward the south. Populations north of 60 N contained 21 (75%) of the 28 sequence haplotypes, and often contained multiple restriction-site haplotypes.

Populations south of that latitude contained a total of only 13 restriction haplotypes, and were usually monomorphic for restriction-site haplotypes. Phylogenetic analyses of R.

integrifolia plastid DNA sequences supported a hypothesis of southward spread from

Alaska, and suggested that two to three clades of R. integrifolia independently migrated southward in western North America.

(5)

List of Tables

Table 1 Ecological and morphological characteristics of the four subspecies of Rhodiola integrifolia... 11 Table 2 All Rhodiola Collections by Latitude within Provincial, State or Federal

Jurisdiction. ... ... 14 Table 3 Details of PCR (polymerase chain reaction) conditions, including primers,

parameters and references for eight cpDNA regions screened. ... 23 Table 4 Numbers of individuals per population amplified and used in RFLP and .

sequencing for all Rhodiola integrifolia, R. rhodantha and R. rosea... 24 Table 5 Restriction enzyme recognition sequences and numbers of restriction

sites for the psbA-trnH region in Rhodiola integrifolia. ... 26 Table 6 Restriction digest cut site positions and patterns of cut sites for each

haplotype found in psbA-trnH sequences of Rhodiola integrifolia. ... 27 Table 7 Characters varying in ITS sequences of Rhodiola integrifolia and R.

rhodantha as compared with 16 other species of Rhodiola. ... 39

Table 8 Summary of cpDNA and ITS variation in western North American

Rhodiola (including RFLP and sequence haplotypes). ... 40 Table 9 Characters varying in psbA-trnH sequences of Rhodiola integrifolia and

R. rhodantha. ... 44 Table 10 Distribution of psbA-trnH restriction site haplotypes among sampled

populations of Rhodiola integrifolia and R. rhodantha... 50 Table 11 PsbA-trnH sequence haplotypes of Rhodiola integrifolia and R.

rhodantha. ... 64

Table 12 Pairwise difference matrix of Rhodiola integrifolia psbA-trnH sequence haplotypes. . ... 68 Table 13 Pairwise difference matrix comparing Rhodiola integrifolia psbA-trnH

restriction site haplotypes to themselves. ... 70 Table 14 Inferred population demographic events from Nested Clade

Phylogeographic Analysis (NCPA) of Rhodiola integrifolia psbA-trnH restriction site haplotypes psbA-trnH restriction site haplotypes. ... 72

(9)

List of Figures

Figure 1 Western North America at the maximal extent of the Wisconsinan

glaciation ~18ky before present. ... 3 Figure 2 Approximate extent of Beringian refugium during the Pleistocene epoch

showing the extent of seabed of the Bering Strait that was exposed during glacial cycles. ... ... ... 5 Figure 3 Ranges of the three species of Rhodiola native to North America. ... 10 Figure 4 All Rhodiola integrifolia, R. rhodantha and R. rosea (North American)

collection sites. ... 19 Figure 5 Gel images of restriction enzyme banding patterns... 28 Figure 6 Phylogenetic trees for 18 species of Rhodiola based on ITS. ... 42 Figure 7 Phylogenetic trees for populations of R. integrifolia and R. rhodantha

based on the chloroplast PsbA-trnH region, with R. rosea as outgroup. ... 47 Figure 8 A comparison of bootstrap consensus trees from nuclear (ITS) and

chloroplast (psbA-trnH) DNA regions in R. integrifolia and

R. rhodantha... 48

Figure 9 Geographic distribution of cpDNA psbA-trnH restriction site haplotypes and associated sequence haplotypes in western North American

Rhodiola populations. ... 53

Figure 10 Geographic distribution of all R. integrifolia populations illustrating

within-population haplotype diversity... 66 Figure 11 TCS-generated network of all PsbA-trnH sequence haplotypes of

R. integrifolia and R. rhodantha ... 70

Figure 12 Network of restriction site haplotypes produced by hand. ... 72 Figure 13 Nested cladogram of restriction site haplotypes ... 72

(10)

Acknowledgements

Thanks are owed to so many! I thank Gerry Allen and Joe Antos for support and guidance, and John Taylor, Ken Marr and Barbara Hawkins for agreeing to be on my committee. Special thanks to Erica Wheeler for her undying support and positivity throughout and for talking me into doing this (I think!). Special thanks also to Laurie McCormick for her constant support and encouragement, and for tolerating me in the lab and helping me with the last chunk of my lab work! Thanks to my funding sources, the Natural Sciences and Engineering Research Council (NSERC), the Northern Scientific Training Program (NSTP), the University of Victoria, and the Lewis Clark Memorial Fellowship.

Many people helped with my collecting and collected samples for me. I want to give special thanks to Bruce Bennett (and his team) for the many collections he made for me in the Yukon, and to his family for their hospitality in Whitehorse, particularly when the Grunner broke down. I also thank: Ken Marr and Richard Hebda who also collected samples from many sites in Northern BC for me; Mike Miller, Emily Beinhauer, Jim Benedict, Alan Batten, David Lowry and Mike Cheney, who collected samples for me in Western North America; and Mariannick Archambault and Christine Westergaard who sent me samples from Eastern Canada and Europe. Phil Caswell and Sylvia Fischer both accompanied me on collecting missions in the Yukon. David Player and Janet Lawson not only supported me with their friendship, but also accompanied me on a couple of my collecting trips.

Many others were there with encouragement and help, and I hope you all know who you are (Charlie, Ryan, Sherri, Vicki, Steve, Victoria, the ladies from the club and of course My Family!!!).

(11)

Dedication

To my Parents -

Florence Catherine Guest (September 1, 1916 – September 28, 2002)

Eric Tufts Guest (February 18, 1923 – January 31, 2005)

With Love,

(12)

Chapter 1: INTRODUCTION

Determination of the factors that shape the current distribution of plant species is a central question of plant biogeography. The range of a species is the result not only of present day features of the landscape (e.g., ecology, topography, surficial geology, elevation) and climate (e.g., moisture and temperature regimes, growing degree days), but also of historical events and processes. The dispersal of a species into its present range may be influenced by a combination of factors, many of which can be surmised from our understanding of Earth’s history. Because of changes in past connections between the continents, the Pleistocene glaciations, and climatic change associated with each of these, suitable habitats for species have often been in flux, expanding, shifting and/or contracting over time.

Studies into the historical ranges of plants have traditionally relied on paleobotanical evidence in the form of macrofossils and pollen records. More recently, the advancement of molecular tools has made it possible to evaluate the degree and distribution of genetic variation within a species, which can lead to an enhanced understanding of the historical processes affecting present-day ranges and their past development and evolution.

1.1 Climatic and Glacial History of Western North America

During the Quaternary period (Pleistocene and Holocene Epochs; 2.6 million years ago (mya) to present), Earth has seen several rapid and dramatic changes in climate. Periodic drops in global temperature, most likely brought about by Milankovitch climate

(13)

cycles, facilitated the formation and expansion of continental ice sheets that at times covered a large percentage of the land mass of the Northern Hemisphere (Hays et al, 1976;

Mascarelli, 2009; Andersen and Borns, 1997; Ehlers and Gibbard, 2004). Although

researchers once recognized four main glacial advances punctuated by interglacial intervals (Ogilvie 1998), fluctuations in oxygen-isotope concentrations in marine sediment and ice cores show a record of cooling and warming over the Pleistocene period accounting for as many as 24 glacial-interglacial intervals (Booth et al. 2003; Hewitt, 2000; Andersen and Borns, 1997; Matthews, 1978).

In North America, the last major Pleistocene glaciation was the Wisconsinan approximately 115 – 10 thousand years ago (kya) (Figure 1). The last major advance of the Wisconsinan glaciation, during which most of Canada was covered with ice, occurred between approximately 28 kya and 10 kya (Clague et al, 2004, Ehlers and Gibbard, 2009). East of the Rocky Mountains, the Laurentide ice sheet reached its maximum extent between 21 kya and 18 kya, whereas in the west, the Cordilleran ice sheet reached its maximum extent later, approximately 15 kya to 14 kya (Booth et al, 2003; Andersen and Borns, 1997; Clague et al, 1980; Clague and James, 2002).

In western Canada, expanding alpine glaciers and ice fields from mountain ranges along the Pacific coast (including the St. Elias Range of Alaska and the Yukon and the Coast Ranges of British Columbia) coalesced with valley/piedmont glaciers and alpine glaciers in the Rocky Mountains to form a sheet of ice that surpassed 2000m in thickness in some areas (Clague et al, 2004). Mountain peaks of the Rockies higher than 2500m may have protruded above the ice (Ogilvie, 1998), and many show evidence of having been ice-free (Pielou,

(14)

1991; Hultén, 1937). At its maximum extent, lobes of the Cordilleran ice sheet reached the Pacific Ocean and the Interior Plains east of the Rocky

Figure 1: Western North America at the maximal extent of the Wisconsinan glaciation

~18kya. Shading represents the approximate extent of continental glaciers. Putative high arctic glacial refugia not shown. (Ice sheet coverage based on figures from Williams et al, 2004, and Clague and James, 2002)

Mountains, where the Cordilleran and Laurentide ice sheets may have converged (Booth et al, 2003; Hetherington et al., 2003; Pielou, 1991).

The climatic oscillations of the Pleistocene led to several cycles of expansion and contraction in the geographical ranges of alpine and arctic plants (Comes and Kadereit, 1998; Waltari et al, 2004). With large amounts of water locked in glaciers, sea levels dropped as

Cordilleran Ice Sheet

Beringia

Laurentide Ice Sheet

(15)

much as 200 m below present sea level, exposing large areas of continental shelf along coastlines worldwide (Lambeck and Chappell, 2001; Andersen & Borns, 1997; Pielou, 1991; Hultén, 1937). Exposed coastlines, ice-free mountain tops protruding through the ice sheets (nunataks), and larger ice-free regions both south and north of the main Cordilleran and Laurentide ice sheets were habitable refuges for plants and animals within or near the uninhabitable glacial landscape (Clague et al, 2004; Stehlik et al, 2002; Pielou, 1991; Matthews, 1978; Ives, 1974). All of these areas could serve as refugia i.e., geographical areas of varying size where taxa are able to persist during a period of range contraction (Bennett and Provan, 2008; Bennett et al, 1991 in Ferris et al, 1999; Pruett and Winker, 2007). During glaciations, refugia were important reservoirs of genetic diversity for many plant species. In North America, the area south of the ice sheets that was never glaciated harboured the majority of temperate plant taxa (Soltis et al, 1997; Hewitt, 2004; Ives, 1974). Beringia, the region northwest of the ice-sheets on either side of, and including, the exposed continental shelf of the Bering Strait (Figure 2), was also a significant refugium, especially for arctic and alpine plants (Eidesen et al, 2007; Abbott et al, 2000).

The concept of Beringia as a ‘land bridge’ linking Asia and North America had occurred to several explorers and scientists before the Swedish botanist Eric Hultén

published his analysis (Hultén, 1937) of the distribution of boreal and arctic plants (Hopkins, 1967). In his “Outline of the History of Arctic and Boreal Biota during the Quaternary Period” (1937), Hultén coined the term Beringia and concluded that this region had remained an ice-free refugium for at least the last two major glaciations. Because of the shallow waters of the Bering Strait, “a lowering of the sea-level of only 50m would result in a connection between Asia and America 300 km broad” (Hultén, 1937). The lowering of sea

(16)

levels during glaciations, coupled with the effects of isostasy as the Earth’s crust responded to the removal of the mass of water in some regions and the addition of the weight of ice in others, indicated that the land bridge was probably exposed during much of the past two

Figure 2: Approximate extent of Beringian refugium (shaded area) during the Pleistocene

epoch showing the extent of seabed in the Bering Strait that was exposed during glacial cycles. Outlines of Beringia based on image at: http://www.beringia.com.

million years (Morlan, 1996). Hultén proposed that Beringia was the initial centre from which most arctic plants radiated eastwards and westwards across circumpolar regions during the Late Tertiary, before the first ice age of the Pleistocene. He also hypothesized that the majority of arctic plants, although their larger ranges shifted and became fragmented

Beringia

Siberia

(17)

with each glacial cycle, were able to persist in Beringia and re-populate suitable habitats when they became available (Eidesen et al., 2007, Hultén, 1937).

1.2 Phylogeography and Plant Molecular Markers

Populations of organisms in recently colonized regions often have low genetic diversity compared with those found in long-term reservoirs of that taxon -- so-called “leading edge colonization” (Hewitt, 1996). Taxa recolonizing formerly glaciated regions typically experience repeated genetic bottlenecks that occur when only a few individuals contribute their genetic material to new populations (Alsos et al., 2005; Ehrich et al., 2007). In addition, once these new populations are established in a region, it may be more difficult for later long-distance immigrants to become established and contribute their genetics because the available habitat has already been taken, thus perpetuating low genetic diversity in the new range (Hewitt, 1999).

The term phylogeography was introduced by John Avise and coworkers in 1987, to describe a field of study that investigates the geographic distributions of genealogical lineages within or among species, or among other taxonomic groupings. The study of molecular variation within the context of geography can reveal the historical and evolutionary events and processes that led to the present evolutionary and geographical circumstances of a taxon (Koch and Kiefer, 2006, Avise et al., 1987). Although a majority of phylogeographic studies to date have dealt with animal species (Kuchta and Meyer, 2001; Schaal et al, 1998; Brunsfeld et al, 2001), the number of plant studies has grown

exponentially in the last decade (Smith, 2007; Schaal et al, 1998; Dobes et al, 2004). Most of these, however, have focused on European plant species, or on post-glacial expansions of

(18)

circum-arctic species within Europe (Hewitt, 2004). Phylogeographic analyses of North American plant species, and those of northwestern North America in particular, are still relatively uncommon. Thus, significant gaps exist in our understanding of the

phylogeographic history of the western North American flora.

Phylogeographical methods employ molecular techniques to uncover the genetic structure of a taxon across part or all of its range. Molecular markers commonly used for phylogenetic analysis in plants occur in both organelle and nuclear DNA regions. Of these, non-coding regions of the chloroplast DNA (cpDNA) are especially useful in the study of plant phylogeography. The chloroplast DNA genome in flowering plants is highly conserved in terms of size, structure and gene content (Olmsted and Palmer, 1994; Doyle 1993), and is one of the most attractive sources of molecular markers for the phylogenetic study of plants. In angiosperms, cpDNA is maternally inherited and thus is only dispersed through seed. Seeds are typically dispersed less widely than pollen, thus, the geographical migration of the species due to seed dispersal can be tracked separately from any gene flow that may be caused by pollen exchange (Petit et al 2003; Latta 2006). Since the accumulated changes in the maternal line are not obscured by the elevated gene flow associated with pollen

exchange, geographic patterns are likely to be retained over longer periods than would be the case with biparentally inherited nuclear markers (Comes and Kadereit, 1998; Ferris et al, 1999; Doyle 1993; Harris and Ingram 1991, Hewitt, 2004; Tremblay and Schoen, 1999; Soltis et al, 1997). Evidence of glacial refugia, migration routes, and range expansions that occurred in the distant past can now be inferred through examining the geographical

(19)

The chloroplast genome harbours considerable intraspecific variation, especially in introns or non-coding spacer regions; this variation is geographically structured in many species (Soltis et al, 1997; Harris and Ingram, 1991). One widely used approach is to amplify selected DNA regions and survey these for restriction fragment length

polymorphisms (RFLPs), (Wyman and White, 1980; Holderegger and Abbott, 2003).

Alternatively, DNA regions of interest such as non-coding spacers can be sequenced directly.

1.3 The Study Species, Rhodiola integrifolia

Rhodiola integrifolia Raf. is a distinctive, succulent-leaved alpine plant with reddish

flowers that is widespread in western North America. It is one of 60-90 recognized species of Rhodiola (Mayuzumi and Ohba 2004; Lei et al, 2004). All of the species of this genus are native tohigh latitude and/or elevation, cold regions of the Northern Hemisphere with the majority concentrated in and around the central Asiatic highland (Lei et al, 2004, 2006; Ohba, 1988; Moran, 2005 (unpublished manuscript). Several species of Rhodiola (e.g. R.

rosea L., R. alsia (Fröderström) S. H. Fu, and R. crenulata (Hooker & Thomson) H. Ohba)

have been in use for centuries as traditional medicines in Asia and other parts of the Old World, and are now becoming popular in alternative medicine in the West (Lei et al, 2004, 2006, Xia et al, 2005). As a result of this recent popularity, many more species of Rhodiola are being studied and exploited for their potential medicinal properties. This, along with deforestation and grazing pressure in their historical range, may threaten several members of this genus (Lei et al, 2004, 2006; Xia et al, 2005; Wang et al, 2005).

The geographic range of Rhodiola integrifolia (Figure 3), spans previously glaciated areas as well as former refugia. In Asia, it extends from the Himalayan Mountains through

(20)

China to eastern Siberia, and in North America it ranges from northern and western Alaska to the western Northwest Territories, BC, Alberta, and the western continental USA. In the southernmost part of its range it is limited to high elevation areas in Colorado, New Mexico and California.

Besides R. integrifolia’s distinctive morphology and widespread (although patchy) distribution, it is also representative of Northern Hemisphere arctic/alpine plant species that were reduced in their ranges to one or a few glacial refugia during the last glaciation. These features make it a good study species to use in locating those refugia and in determining the post-glacial history of arctic/alpine plant species’ geographical expansion.

In addition to R. integrifolia, two other species of Rhodiola (R. rosea (a.k.a. Sedum

rosea (L.) Scop., and R. rhodantha (A. Gray) H. Jacobsen (a.k.a. Clementsia rhodantha

(Gray) Rose, or Sedum rhodanthum Gray), are native to North America. The yellow flowered R. rosea occurs in eastern North America and east to the higher elevations and latitudes of Eurasia (Moran, 2005 (unpublished manuscript); Clausen, 1975). Rhodiola

rhodantha is found in alpine habitats in the Rocky Mountains south of the glacial maximum

(Figure 3), where its range overlaps with the southern range of R. integrifolia. All three species are perennial, with thick rhizomes, succulent leaves, and annual floral stems that arise from the leaf axils. Rhodiola rhodantha has pinkish hermaphrodite flowers in an elongate, raceme-like inflorescence; R. integrifolia and R. rosea have smaller, unisexual flowers in corymbose cymes. Both R. integrifolia and R. rosea are dioecious (Clausen 1975).

The taxonomic relationship of R. integrifolia and R. rosea has been subject to debate. Perhaps due to its similar morphology Rhodiola integrifolia has been considered to be a

(21)

Figure 3: Ranges of the three species of Rhodiola native to North America (red, R.

integrifolia; yellow, R. rosea; and pink, R. rhodantha). Ranges adapted from Clausen 1975.

subspecies of R. rosea by many authors (e.g. R. rosea L. subsp. integrifolium (Raf.) Hult. (Hultén, 1968) (1), or subsp. integrifolia (Raf.) Hara, (Uhl, 1952; Cody, 2000) (2); and,

Sedum rosea (L.) Scop. subsp. integrifolia (Welsh, 1974; Clausen, 1975) (3). More recently

the two taxa have been treated as separate species within Rhodiola: R. integrifolia Raf. and

R. rosea L. (Moran, 2005 (unpublished manuscript), 2000; Amano et al, 1995; Ohba, 1999;

Lei et al, 2004). This separation of the complex into two species is based on differences in

Rhodiola rosea

Rhodiola rhodantha

(22)

chromosome numbers (R. integrifolia: n=18; R. rosea: n=11), and floral characteristics (Clausen, 1975; Moran, 2005 unpublished manuscript). Here, botanical nomenclature follows Reid Moran’s (2005, unpublished) treatment prepared for the Flora of North America. Currently, R. integrifolia is divided into four recognized subspecies in North America (Ohba, 1999; Olfelt et al, 1998, 2001) (Table 1). Most R. integrifolia specimens used in this study were of subsp. integrifolia.

Table 1: Ecological and morphological characteristics of the four subspecies of Rhodiola integrifolia. (Sources: Moran, 1996 (unpublished manuscript); Clausen, 1975)

Rhodiola integrifolia subsp. integrifolia subsp. procera subsp. neomexicana subsp. leedyii Range Widespread Alaska to California, E to Rockies Southern Rockies of Colorado and New Mexico

One site only in the Sierra Blanca Mtns. of Lincoln County, New Mexico

Locally endemic in two disjunct areas: Minnesota (two counties), and New York (two counties)

Habitat

Cliffs & rocky slopes, alpine meadows or tundra Montane to alpine slopes among rocks, gravel or stony loam Exposed sites on porphyritic rock from 3300 to 3600 m NW facing limestone cliff ledges, 100 to 400 m Petal

colour Dark red Dark red

Yellow, red at apex and on keel

Dark red or yellow at base Leaves Bright green, ovate to elliptic, oblanceolate Glaucous to bright green, oblong, oblanceolate

Bright green, linear-oblanceolate Blue-green, long, narrow, oblanceolate Plant ht. 3-15 cm up to 50 cm 15-25 cm 15-45 cm Status Not threatened Not threatened Of conservation

concern

Of conservation concern

Although arctic and alpine ecosystems have undergone dramatic fluctuations in environmental conditions since the Pleistocene, researchers are unsure about how resilient these systems will be to rapid warming (Weider and Hobaek, 2000). It is already clear that

(23)

arctic and alpine habitats are at an elevated risk from climate change (Weider and Hobaek, 2000, Waltari et al., 2004; Callaghan et al., 2004), and it is expected that as warming continues the available habitat for arctic and alpine species will decrease as other plant species encroach into their ranges (Hewitt, 1996; Lesica and McCune 2004; Callaghan et al., 2004). Understanding the local and regional variability within the genome of R. integrifolia will lead to a better understanding of its origins and evolution, and may also allow us to predict its potential future range.

1.4 Objectives:

My objectives in this study are as follows: i) to clarify R. integrifolia’s position within the genus Rhodiola and its taxonomical relationships with the other North American species; ii) to determine the origins of extant populations of R. integrifolia in western North America; iii) to estimate macro scale dispersal patterns followed by R. integrifolia in

reaching its current range, and if possible obtain information on the evolutionary processes involved; and iv) to shed some light on this species’ ability to respond to changing climates in the future. To achieve these goals, I used data from the chloroplast and nuclear genomes to investigate local and regional patterns of genetic variation in R. integrifolia, and to make comparisons with the other North American species, R. rhodantha and R. rosea.

(24)

Chapter 2: METHODS

2.1. Collection of Plant Material

I collected Rhodiola at 44 localities over three field seasons (northern BC and Yukon in 2004, Yukon and Alaska in 2005, and southeast BC and western continental USA in 2006 (Table 2). Potential collecting locations were identified by extracting range and locality information from published floras, consulting with local botanists, and visiting herbaria or checking their online databases. I obtained information from herbaria at the University of Washington, University of Wyoming, University of Colorado, University of California at Berkeley, Oregon State University, Colorado State University, the Royal BC Museum, University of Victoria, University of British Columbia, University of Montana, University of Alaska, and from the private herbarium of Mr. Bruce Bennett in Whitehorse, Yukon. In total I obtained samples from 66 populations of R. integrifolia, six populations of R. rhodantha, and four populations of R. rosea (Table 2), including population samples from 22 localities contributed by other collectors (Figure 4).

At each site I collected leaves, stems and (if available) flowers from 5-20 individual plants, and one to several voucher specimens. Each population sample was collected within an area of between approximately 5m2 and 30m2, depending on plant density. I placed plant samples in individual coin envelopes, then into a resealable plastic bag containing indicating desiccant crystals for rapid drying. All voucher specimens have been deposited in either the University of Victoria or the Royal British Columbia Museum herbarium.

(25)

Table 2: All Rhodiola Collections by Latitude within Provincial, State or Federal

Jurisdiction Site

# Code Location Elevation Latitude Longitude Collector

Alaska, USA

1 PM

Pinnell Mtn. NW of Fairbanks, Twelvemile Summit trailhead, N side Hwy 6 ~km140

1126m 65° 24' 36" 145° 59' 06" Heidi Guest

2 BG

Bison Gulch, E side of Hwy 3 ~25 km N of entrance to Denali Park.

815m 63° 48' 07" 148° 58' 19" Heidi Guest 3 CA Cantwell, NE of Jct of Hwy's

3 & 8. 911m 63° 23' 58" 148° 51' 57" Heidi Guest 4 D2 Denali Hwy site 2, 60 km E of

Cantwell, S side of Hwy 8. 960m 63° 15' 03" 147° 49' 07" Heidi Guest

5 D4

Denali Hwy site 4, Tangle Lakes/ McLaren Summit, S side of Hwy 8.

1250m 63° 05' 21" 146° 24' 36" Heidi Guest

6 D3

Denali Hwy site 3, N side of Hwy 8 ~ 20 km E of Susitna R. in Clearwater Mtns.

1161m 63° 03' 06" 147° 14' 57" Heidi Guest

7 LC

Little Coal Creek, trailhead is on E side of Hwy 3 ~75 km S of Denali Hwy Jct.

902m 62° 52' 54" 149° 41' 20" Heidi Guest

8 PH

Peter's Hills, off Hwy 3 at Trapper Creek, ~10 km past Petersville NE of river camp

637m 62° 32' 10" 150° 50' 01" Heidi Guest

9 HP

Hatcher's Pass, NW off Hwy 1 at Palmer, NW above junction of road to Lodge with

Hatcher's Pass Rd.

1004m 61° 46' 32" 149° 17' 52" Heidi Guest

10 PC

Peter's Cr. Trail, NE of Anchorage in residential area E of Hwy 1 at top of trail

936m 61° 24' 33" 149° 23' 20" Heidi Guest

11 TP

Thompson Pass, Pass above Blueberry Lake SE side of Hwy 4 ~75 km from Valdez

832m 61° 07' 38" 145° 43' 31" Heidi Guest

12 BL

Blueberry Lake, Campsite #9, Chugach Range on Hwy 4 ~60 km E of Valdez

636m 61° 07' 12" 145° 42' 18" Heidi Guest

13 CP

Crow Pass, SE of Anchorage turn NE off Hwy 1 at Girdwood, on slope on E side of trail ~30 minutes hike from trailhead.

500m 61° 02' 07" 149° 07' 16" Heidi Guest

14 SM

Swetman Mine, S from Hope (Kenai Penn), past Coeur d'Alene campsite. E of rd. W of stream on E face of hump.

549m 60° 48' 07" 149° 32' 45" Heidi Guest

15 CM

Cooper's Mtn, S side of Hwy 1 on Kenai Penn, just past Cooper's Landing NW facing mini-saddle near summit.

780m 60° 28' 14" 149° 50' 30" Heidi Guest

16 CB Carbon Mtn, SE of Valdez in

(26)

Site

Yukon Territory, Canada

17 KP Kay Point, N coast of Yukon,

Beaufort Sea, on mud flats 1 m 69° 17' 31" 138° 23' 35" Bruce Bennett 18 BR Blow River Delta, N coast of

Yukon, Beaufort Sea 0.6 m 68° 55' 24" 137° 10' 26" Bruce Bennett

19 MK

Mount Klotz Camp, in meadows along creeks and in late snow melt areas

1165 m 65° 21' 48" 140° 10' 54" Bruce Bennett

21 O5

Ogilvie Mtns Site 5, Near top of ridge late snow melt area on north side.

1671m 64° 47' 58" 138° 03' 49" Heidi Guest 22 O3 Ogilvie Mtns Site 3. Ridge S

of saddle 1731m 64° 47' 28" 137° 43' 16" Heidi Guest 23 O4 Ogilvie Mountains Site 4 1527m 64° 45' 20" 137° 58' 17" Heidi Guest 24 GL Gillespie Lake, Bonnet Plume

Drainage 1378m 64° 43' 43" 133° 59' 00" Bruce Bennett 25 PL

Pinguicula Lake, Bonnet Plume Drainage, mid slope on mountain

1292m 64° 41' 25" 133° 26' 18" Bruce Bennett 26 NP North Fork Pass, Dempster

Hwy, east facing slope. 1270m 64° 35' 15" 138° 16' 53" Ken Marr 27 TV

Tombstone Valley, N of Park camp-ground, near stream, close to trail

1062m 64° 30' 47" 138° 14' 13" Heidi Guest 28 GR Grizzly Lake, Tombstone Park 1400m 64° 25' 36" 138° 27' 42" Mike Miller 29 TW

Top of the World Hwy, W. of Dawson City. Midslope N facing alpine meadow

1150m 64° 11' 38" 140° 21' 40" Ken Marr 30 KH Keno Hill, Road up hill above

town NW sloping meadow 1609m 63° 56' 13" 135° 13' 46" Heidi Guest 31 NC

North Canol Rd, in meadows 212 miles W. of Norman Wells, Dechen la'.

1706 m 63° 24' 42" 129° 36' 52" Bruce Bennett

32 RL

Rose-Lapie Pass W. side & below Hwy 6 (Canol Rd.) between road and small lake

1095m 61° 37' 49" 133° 03' 36" Heidi Guest

33 OM

Outpost Mtn, SW side of Hwy 1, trail S of Kluane Lk, at treeline W facing slope.

1447m 60° 58' 41" 138° 24' 38" Heidi Guest

34 KR

Kluane Rock Glacier, trail off Hwy 3, S of Haines Jct. lateral moraine SE side.

1080m 60° 27' 01" 137° 5' 01" Heidi Guest

35 MT

Montana Mtn. Mtn rd. S of Carcross, N side of mountain, N facing meadow.

1392m 60° 06' 45" 134° 40' 44" Heidi Guest 36 MM Montana Mtn, West above

road along E side before slide. 1485m 60° 06' 10" 134° 41' 25" Heidi Guest

British Columbia, Canada

37 CC

Chuck Creek, Haines Rd, ~120 km N of Haines, Alaska, W facing near top of trail

(27)

Site

British Columbia, Canada

38 TG

3 Guardsmen's Pass, Haines Rd ~ 90 km N of Haines Alaska, SE side of road

968m 59° 36' 10" 136° 29' 21" Heidi Guest 39 MF Mt. Fetterly, SE of Atlin,

meadow in riparian zone. 1455m 59° 31' 42" 134° 01' 39"

Emily Beinhauer 40 MO Moose Mtn, SE of Atlin, Steep

SSE slope. 1448m 59° 25' 38" 133° 28' 18"

Emily Beinhauer 41 EC

Estshi Creek, Coast Mtns, Iskut River; 6-7 km NW of Teigen Lake.

1430m 56° 45' 51" 130° 16' 15" Ken Marr

42 HA

S. end of Hanna Ridge N of Meziadin Lk, Coast Mtns. WSW facing ridge apex.

1800m 56° 13' 58" 129° 28' 57" Marr/Hebda

43 TJ

Mt. Tommy Jack, Skeena Mtns, ridge 2-3 km E of summit, S of Triangle Lake.

1770m 56° 03' 02" 127° 46' 13" Marr/Hebda

44 AT

Atna, Skeena Mtns, ridge W side of Atna Range. E facing upper slope.

1662m 56° 00' 20" 127° 40' 47" Marr/Hebda 45 IC N of Insect Creek, Nass

Ranges. W facing slope. 1718m 55° 04' 00" 128° 34' 51" Marr/Hebda 46 MC Mt. Couture, SW facing saddle

on ridge apex 1800m 54° 53' 28" 128° 41' 27" Marr/Hebda 47 HU Hudson Bay Mt, SW of S

facing slope. 1668m 54° 49' 00" 127° 17' 10" Marr/Hebda 48 HB Hudson’s Bay Mountain,

Crater Lake Trail. 1707m 54° 46' 26" 127° 14' 26" Heidi Guest 49 TM

Thornhill Mtn, E of Terrace - turn off Lakelse Rd, top of Mtn in S facing depression

1145m 54° 30' 33" 128° 27' 25" Heidi Guest 50 TH Towustasin Hill, Graham Is.

Queen Charlotte Islands 370m 53° 35' 06" 132° 26' 13" Mike Cheney 51 BB

0.75 km NW of Bonanza Beach, N end Rennell Sound, Graham Is, Queen Charlotte Islands

3.5m 53° 24' 52" 132° 32' 55" David Lowry

52 GP Gimli Peak, W of Valhalla

Park, in saddle at base of spire. 2423m 49° 45' 39" 117° 38' 56" Heidi Guest 53 FP Fisher Peak, NE of Cranbrook,

edge of pond at treeline 2154m 49° 38' 27" 115° 29' 51" Heidi Guest Montana, USA

54 RT

Reynolds Pass, Glacier National Park, steep SE facing scree slope above trail just before pass

2315m 48° 40' 38" 113° 43' 53" Heidi Guest

55 GF

Goat Flats, Anaconda-Pintler Wilderness Area, tundra ridge above Upper Seymour Lake, gentle N facing slope.

(28)

Site

Wyoming, USA

56 BP

Beartooth Pass, At top of pass on Hwy 212, Montana/ Wyoming border. N facing boulder field SW side of hwy

3352 m 44° 58' 16" 109° 28' 27" Heidi Guest

57 MB

Medicine Bow Mtn, Hwy 130, below summit, in S facing scree on trail edge (growing with R. rhodantha).

3461m 41° 21' 35" 106° 18' 36" Heidi Guest

Colorado, USA

60 AR

Arapaho Rim, Along route from Rainbow lakes to Arapaho Col. Jct btwn N-sloping meadow and frost-sorted rubble stream

3650m 40° 02' 17" 105° 15' 11" Jim Benedict

61 AA Above Arapaho, see above 3855m 40° 01' 19" 105° 38' 57" Jim Benedict 65 LP

Loveland Pass, top of Hwy 6 off I70 ~ 100 km W of Denver, trail N side flat area between rocky outcrops

3734m 39° 39' 36" 105° 53' 07" Heidi Guest

66 SP

Schofield Pass 30 km N of Crested Butte, up trail E from summit on SW slope

3368m 39° 00' 58" 107° 02' 40" Heidi Guest

68 HM

Handies Meadow, Col E of Cinnamon Pass on 4wd road btwn Lake City and Silverton, (growing with R. rhodantha).

3467m 37° 55' 52" 107° 30' 53" Heidi Guest

70 LA

La Plata Canyon, W. of Durango, Forest Rd 124 N of Hesperus, SE facing slope of road cut.

3120m 37° 25' 56" 108° 02' 07" Heidi Guest

California, USA

71 PP

Paiute Pass, W on Hwy 168 from Bishop, Inyo Co. Above treeline trail near Loch Leven in cracks of rocks

3273m 37° 13' 51" 118° 39' 10" Heidi Guest

72 MD

Mt Dana, Yosemite Nat'l Pk trail SE from Tioga Pass park entrance (E side), toe of W facing gentle slope.

3460m 37° 54' 32" 119° 14' 14" Heidi Guest

Rhodiola rhodantha

Wyoming, USA 58 SR

Snowy Range, Hwy 130, below summit of Medicine Bow Mtn. in S facing scree (growing with R. integrifolia)

3461m 41° 21' 35" 106° 18' 36" Heidi Guest

Utah, USA

59 UM Uinta Mtns, Hwy 150 , N side

(29)

Site

Colorado, USA

62 AC

S. Arapaho Col. on frost patterned, south facing slope. See AR above

3855m 40° 01' 19" 105° 38' 57" Jim Benedict 63 FM Fourth of July Mine, Wetland

bench E of Mine 3390m 40° 00' 53" 105° 39' 39" Jim Benedict 64 FV

4th of July Valley, Frost patterned wetland downslope from late-lying snowbank. Forest-tundra ecotone

3400m 40° 00' 31" 105° 39' 54" Jim Benedict

69 AB

American Basin, same as Handies Meadow above, (with

R. integrifolia)

3456m 37° 55' 43" 107° 30' 52" Heidi Guest

Rhodiola rosea

73 NU

Qinngorput, Nuuk, Greenland, W coast, near town of Nuuk (Godthåb), wet S facing slope

130m 64° 10' 38" 51° 39' 39" Kristine Westergaard 74 NA

Nasarsuaq, Greenland, Signalhojen, SE of airport, moist, west facing slope

210m 61° 09' 38" 45° 24' 28" Kristine Westergaard Eastern Canada

75 NH Nunavik HH, Quebec n/a 59° 30' 39" 65° 39' 27" Mariannick Archambault 76 CE Cap Enragé, New Brunswick n/a 45° 35' 38" 64° 46' 48" Mariannick

(30)

R. integrifolia R. rhodantha R. rosea

Figure 4: All Rhodiola integrifolia, R. rhodantha and R. rosea (North American) collection sites. Shaded area represents the North

American range of R. integrifolia. Range based on map by Clausen 1975.

(31)

2.2 DNA Extraction

Using approximately 10 mg of dried leaf and/or flower tissue per sample, I extracted DNA from 10 individuals (or if <10, all individuals collected) per Rhodiola population, using a modification of the hexadectyl-trimethylammoniumbromide (CTAB) method (Doyle and Doyle, 1990) for use with dried tissue (see details in Appendix I). I extracted DNA from 555 individuals of R. integrifolia, 50 of R. rhodantha, and 14 of R. rosea. All DNA samples were diluted 1:10 with double-distilled water (ddH2O) before further use.

2.3 DNA Regions Sampled 2.3.1 Nuclear DNA: ITS

I sequenced the internal transcribed spacers 1 and 2 of the nuclear ribosomal DNA locus (ITS) from 32 individuals of R. integrifolia (from 28 populations), four of R. rhodantha (from four populations) and four of R. rosea (from four populations). I used primers ITS1 and ITS4 (White et al., 1990) to amplify the ITS1, 5.8S rDNA and-ITS2 regions of the nuclear ribosomal DNA as a single fragment.

For amplification, I used a mixture of 5L extraction product (1:10 dilutions), 5L 10x PCR buffer (New England Biolabs (NEB) or Invitrogen), 1.5L 50mM MgCl2

(Invitrogen), 5L 2mM dNTPs (Invitrogen), 2.5L each of forward and reverse primers (ITS1 and ITS4 respectively (Life Technologies/Gibco)), 28.25L ddH2O, and 0.25L Taq DNA Polymerase (5 U/L, NEB or Invitrogen). Reactions were performed on an Eppendorf Mastercycler Gradient Thermocycler (Eppendorf, Hamburg, Germany) or on a Techne TC-312 thermocycler (Techne, Duxford, Cambridge, UK).

(32)

PCR conditions for the ITS region were as follows: 3 min at 94C; 30 cycles of 30 seconds at 94C, 1 min at 55C, and 1 min at 72C; and an extension step of 10 min at 72C. Amplification products were run on 1% TBE agarose gels, stained in ethidium bromide, and observed and photographed under short-wave UV light.

Prior to sequencing I purified the PCR products using a Qiagen QIAquick® PCR Purification kit (Qiagen Inc., Mississauga ON, Canada). Sample concentrations of the purified PCR products were estimated on a 1% agarose gel by comparing band intensities with a 50 base pair standard reference ladder (New England Biolabs) or by measuring absorption intensity on a Nanodrop 1000 spectrophotometer (Thermo Scientific).

For sequencing, 5l of cleaned product were used for each sample. Sequencing of early samples was carried out by the University of Victoria’s Centre for Biomedical Research (CBR), on a CEQ 8000 (Beckman Coulter) DNA sequencer with a Dye Terminator Cycle Sequencing Quick Start Kit (Beckman Coulter). Both strands were sequenced. Later sequencing was carried out by Macrogen Inc. (Seoul, South Korea), using BigDye™

terminator conditions (Applied Biosystems, Foster City, USA). Samples were purified using ethanol precipitation and run on an ABI3730XL or BI3700 automatic sequencer (Applied Biosystems, Foster City, USA). These sequences were determined using the reverse primer (ITS4) only.

2.3.2 CpDNA markers/regions

2.3.2.1 Screening of chloroplast DNA markers/regions

I chose eight non-coding cpDNA regions for an initial survey of variation. These included six spacer regions (psbA-trnH, trnT-trnL, trnD-trnT, trnS-trnG, trnS-trnfM and

(33)

(2005) to exhibit variation in a variety of angiosperm species. Individuals of R. integrifolia from geographically widespread locations sampled in 2004 were used for this initial survey. For all DNA regions except trnL-trnF, I selected samples from six sites: two from Yukon (#27, #30) and four from B.C. (#37, #38, #48, #49). For the trnL-trnF spacer, I used samples from Alaska (#3), Yukon (#23/), Colorado (#66) and California (#71), amplifying four individuals from each population; I also amplified samples of R. rosea, including six

individuals from Quebec (#75), and four individuals from New Brunswick (#76). Details for PCR conditions for the eight cpDNA regions tested are presented in Table 3.

Two to four samples of R. integrifolia were sequenced for each cpDNA region (except the rpL16 intron, which did not amplify successfully). Of the seven regions sequenced, two (trnT-trnL and trnS-trnfM) yielded poor sequence, and one (trnL-trnF) showed little variation. Of the four remaining cpDNA regions, I selected the psbA-trnH spacer for further study because it showed variation that could be readily detected with restriction enzymes.

2.3.2.2 Amplification and Sequencing of psbA-trnH region

I amplified the psbA-trnH region in ten individuals (or as many as were extracted) from each collection site, using the PCR protocols described above. A total of 613

individuals were amplified for the psbA-trnH region, including 45 individuals of R.

rhodantha and 11 individuals of R. rosea. Samples were prepared for sequencing following

the procedures outlined for the ITS region, using primers psbA (f) and trnH (r). Sequences of the psbA-trnH spacer region were obtained for 92 individuals of R. integrifolia, 7 individuals of R. rhodantha, and 6 individuals of R. rosea (Table 4).

(34)

Table 3. Details of PCR (polymerase chain reaction) conditions, including primers, parameters

and references for eight cpDNA regions screened, and the samples sequenced for each region in preliminary trial. Samples sequenced listed by population number and site code/individual.

Region Primer Primer sequence (5’ to 3’) Amplification parameters Reference Samples sequenced psbA-trnH psbA (F) trnH (R)

GTT ATG CAT GAA CGT AAT GCT C

CGC GCA TGG TGG ATT CAC AAA TC 94ºC 3’ 94ºC 30” 57ºC 1’ 25X 72ºC 1’ 72ºC 10’ (Sang et al., 1997) #27 (TV2), #30 (KH2), #37 (CC4), #48 (HB2) trnT-L Tab A (F) Tab B (R)

CAT TAC AAA TGC GAT GCT CT

TCT ACC GAT TTC GCC ATA TC 94ºC 3’ 94ºC 30” 52ºC 1’ 30X 72ºC 1’ 72ºC 10’ (Taberlet et al., 1991) #30 (KH2), #49 (TM6) trnD-T trnD (F) trnT (R)

AAC AAT TGA ACT ACA ATC CC

CTA CAA CTG AGT TAA AAG GG 80ºC 5’ 94ºC 45” 52-58ºC 30” 30X 72ºC 1’ 72ºC 5’ (Demensure et al., 1995) #37 (CC4) #48 (HB2) trnS-G trnS (F) trnG (R)

AGA TAG GGA TTC GAA CCC TCG

GTA GCG GGA ATC GAA CCC GCA TC 80ºC 5’ 95ºC 1’ 50ºC 1’ 35X 65ºC 5’ 65ºC 10’ (Shaw et al., 2005) #27 (TV2) #38 (TG4) trnS-fM trnS (F) trnfM (R)

GAG AGA GAG GGA TTC GAA CC

CAT AAC CTT GAG GTC ACG GG 80ºC 5’ 94ºC 30” 55ºC 30” 30X 72ºC 2’ 72ºC 5’ (Demensure et al., 1995) #30 (KH2) #37 (CC4) trnL-F Tab C (F) Tab F (R)

CGA AAT CGC TAG ACG CTA CG

ATT TGA ACT GGT GAC ACG AG 94ºC 3’ 94ºC 30” 52ºC 1’ 25X 65ºC 5’ 65ºC 4’ (Taberlet et al., 1991) #3 (CA3) #23 (O43) #66 (SP3) #71 (MD3) rpL 16 rpL 16 (F) rpL 16 (R)

GCT ATG CTT AGT GTG TGA CTC GTT G CCC TTC ATT CTT CCT CTA TGT TG 80ºC 5’ 95ºC 1’ 50ºC 1’ 35X 72ºC 2’ 72ºC 5’ (Small et al., 1998) (not sequenced) rpS 16 rpS 16 (F) rpS 16 (R)

AAA CGA TGT GGT ARA AAG CAA C

AAC ATC WAT TGC AAS GAT TCG ATA 80ºC 5’ 94ºC 30” 50-55ºC 30” 30X 72ºC 1’ 72ºC 5’ (Shaw et al., 2005) #27 (TV2) #49 (TM6)

(35)

Table 4: Numbers of individuals per population amplified and used in RFLP (for psbA-trnH) and sequencing (psbA-trnH and ITS) for all Rhodiola integrifolia, R. rhodantha and R. rosea. Populations arranged by latitude within each jurisdiction (State/Province, etc.)

Site Site Number amplified Number digested Number sequenced Number sequenced

number Species Site Name Code psbA-trnH

psbA-trnH

PsbA- trnH

ITS

1 R. integrifolia Pinnell Mtn. Alaska PM 10 10 2 2

2 R. integrifolia Bison Gulch, Alaska BG 10 10 3 1

3 R. integrifolia Cantwell, Alaska CA 10 10 1

4 R. integrifolia Denali Hwy 2, Alaska D2 5 5 1

5 R. integrifolia Denali Hwy 4, Alaska D4 10 10 4 1

6 R. integrifolia Denali Hwy 3, Alaska D3 10 10 1

7 R. integrifolia Little Coal Creek, Alaska LC 10 10 2 1

8 R. integrifolia Peter's Hills, Alaska PH 10 10 1

9 R. integrifolia Hatcher's Pass, Alaska HP 10 10 1

10 R. integrifolia Peter's Cr. Trail, Alaska PC 10 10 1

11 R. integrifolia Thompson Pass, Alaska TP 10 10 2 1

12 R. integrifolia Blueberry Lake, Alaska BL 10 10 1

13 R. integrifolia Crow Pass, Alaska CP 10 10 1

14 R. integrifolia Swetman Mine, Alaska SM 11 10 1 1

15 R. integrifolia Cooper's Mtn, Alaska CM 10 10 1

16 R. integrifolia Carbon Mtn, Alaska CB 10 10 1

17 R. integrifolia Kay Point, Yukon KP 10 10 1 1

18 R. integrifolia Blow River Delta, Yukon BR 9 9 2

19 R. integrifolia Mount Klotz, Yukon MK 1 0 1 1

20 R. integrifolia Quartet Lake, Yukon QL 4 4 1

21 R. integrifolia Ogilvie Mtns 5, Yukon O5 10 10 3

22 R. integrifolia Ogilvie Mtns 3, Yukon O3 2 2 1 2

23 R. integrifolia Ogilvie Mtns 4, Yukon O4 10 10 2

24 R. integrifolia Gillespie Lake, Yukon GL 9 9 3

25 R. integrifolia Pinguicula Lake, Yukon PL 10 10 2 1

26 R. integrifolia North Fork Pass, Yukon NP 10 10 1

27 R. integrifolia Tombstone Valley, Yukon TV 10 10 1

28 R. integrifolia Grizzly Lake, Yukon GR 9 9 4 3

29 R. integrifolia Top of the World, Yukon TW 10 10 1

30 R. integrifolia Keno Hill, Yukon KH 10 10 3 1

31 R. integrifolia North Canol Rd, Yukon NC 1 0 1 1

32 R. integrifolia Rose-Lapie, Yukon RL 10 10 1 1

33 R. integrifolia Outpost Mtn Yukon OM 3 3 1

34 R. integrifolia Kluane Rock Glacier, Yukon KR 10 10 1

35 R. integrifolia Montana Mtn 04, Yukon MT 10 10 1

36 R. integrifolia Montana Mtn 05, Yukon MM 10 10 1

(36)

Site Site Number amplified Number digested Number sequenced Number sequenced

number Species Site Name Code

psbA-trnH psbA-trnH PsbA- trnH ITS

38 R. integrifolia Three Guardsmen's Pass, BC TG 10 10 2 1

39 R. integrifolia Mt. Fetterly, BC MF 10 10 1

40 R. integrifolia Moose Mtn, BC MO 8 8 2

41 R. integrifolia Estshi Creek, BC EC 6 6 1

42 R. integrifolia Hannah, BC HA 9 9 1

43 R. integrifolia Mt. Tommy Jack, BC TJ 3 3 1

44 R. integrifolia Atna, BC AT 6 6 2 1

45 R. integrifolia Insect Creek, BC IC 8 8 2

46 R. integrifolia Hudson Bay Mtn. 06, BC HU 1 1 1

47 R. integrifolia Mt. Couture, BC MC 8 8 2 1

48 R. integrifolia Hudsons Bay Mtn, 04 BC HB 8 8 1

49 R. integrifolia Thornhill Mtn, BC TM 9 9 1

50 R. integrifolia Towustasin Hill, BC TH 1 1 1 1

51 R. integrifolia Bonanza Beach, BC BB 10 10 1 1

52 R. integrifolia Gimli Peak, BC GP 9 9 1

53 R. integrifolia Fisher Peak, BC FP 10 10 1

54 R. integrifolia Reynolds Pass, Montana RT 10 10 1 1

55 R. integrifolia Goat Flats, Montana GF 10 10 1 1

56 R. integrifolia Beartooth Pass, Wyoming BP 10 10 2 1

57 R. integrifolia Medicine Bow Mtn, Wyoming MB 11 11 1 1

58 R. rhodantha Snowy Range, Wyoming SR 10 10 1 1

59 R. rhodantha Uinta Mtns, Utah UM 6 6 1 1

60 R. integrifolia Arapaho Rim, Colorado AR 2 2 0

61 R. integrifolia Above Arapaho, Colorado AA 1 1 0

62 R. rhodantha S. Arapaho Col. Colorado AC 5 5 1

63 R. rhodantha 4th of July Mine, Colorado FM 9 9 1 1

64 R. rhodantha 4th of July Valley, Colorado FV 10 10 1

65 R. integrifolia Loveland Pass, Colorado LP 10 10 1 1

66 R. integrifolia Schofield Pass, Colorado SP 9 9 1

67 R. integrifolia Cottonwood Pass, Colorado CW 9 9 1 1

68 R. integrifolia Handies Meadow, Colorado HM 10 10 1 2

69 R. rhodantha American Basin, Colorado AB 10 10 2 1

70 R. integrifolia La Plata Canyon, Colorado LA 10 10 1 1

71 R. integrifolia Mt Dana, California MD 10 10 1 1

72 R. integrifolia Paiute Pass, California PP 10 10 1

73 R. rosea Nuuk, Greenland NU 1 0 1 1

74 R. rosea Nasarsuaq, Greenland NA 1 0 1 1

75 R. rosea Nunavik HH, Quebec NH 5 0 2 1

(37)

2.3.2.3 RFLP sampling of the psbA-trnH region

Using the samples initially sequenced for the psbA-trnH region (Table 3), I searched for variable restriction sites using the on-line freeware program Webcutter 2.0 (Heiman, 1997). I identified five restriction enzymes (ApoI, BfaI, BstXI, MseI, and NsiI) for further testing (Table 5). The effectiveness of each restriction enzyme (RE) was tested on four amplified samples and the products run on 2% TBE agarose gels for approximately 45 minutes with a 50 bp reference ladder, followed by staining and observation under UV light. Of the five restriction enzymes (RE’s) that cut the psbA-trnH region at variable restriction sites in the sequenced samples, four produced repeatable and unambiguous banding patterns.

Table 5: Restriction Enzyme recognition sequences and numbers of restriction sites for the psbA-trnH region in Rhodiola integrifolia.

Enzyme & supplier Recognition Sequence Number and Positions of variable restriction sites

APO1, New England Biolabs R/AATTY 3 cut sites. Cuts once or twice at position 86, 243 or 293

BstXI, New England Biolabs CCANNNNN/NTGG 1 cut site. Does not cut or cuts once at position 244

MseI, New England Biolabs T/TAA 4 cut sites. Cuts 1 to 3 times at positions 133, 167, 202, or 267 NsiI, New England Biolabs ATGCA/T 1 cut site. Does not cut or cuts

once at position 132

I performed restriction digests using the four selected REs on each of 614 samples amplified for the psbA-trnH region. Gel concentrations and migration times were adjusted in order to optimize band resolution. For ApoI and MseI the optimal gel concentration was 3% run on a single tier for approximately 70 to 75 minutes. BstXI and NsiI, yielding fewer bands, could be run in two tiers on a 2% gel for approximately 30 to 40 minutes. The

(38)

combination of the four restriction enzymes yielded a total of 11 variable restriction sites over all samples examined (Table 6).

Table 6: Restriction digest cut site positions and patterns of cut sites for each haplotype

found in psbA-trnH sequences of R.integrifolia. Numbers below restriction enzyme names refer to the 5’ position along the psbA-trnH sequence where the cut site or indel occurs.

Haplotype ApoI ApoI BstXI MseI MseI MseI NsiI ApoI MseI MseI Number 243 293 244 133 167 267 132 18-23 72-79 94-99 1 0 0 0 0 1 0 0 0 0 0 2 0 1 1 0 1 0 0 0 0 0 3 0 0 1 0 0 0 0 0 0 0 4 0 0 1 1 1 0 1 0 0 0 5 0 0 1 0 1 0 0 0 0 0 6 1 0 0 0 1 0 0 0 0 0 7 0 0 1 0 1 1 0 0 0 0 8 0 0 1 0 1 0 0 1 0 0 9 0 1 1 0 1 1 0 0 0 0 10 0 0 1 0 0 0 0 0 0 1 11 0 0 1 0 1 0 0 0 1 0

2.3.2.4 Restriction site haplotype determination

From photographs of gels I could detect restriction digest fragments corresponding to the presence or absence of restriction sites within the psbA-trnH spacer. BstXI and NsiI bands were easily distinguishable, and variation in fragment length due to the

presence/absence of indels was discernable in BstXI banding patterns. The various combinations of fragments and banding patterns found for ApoI and MseI were more complex (Figure 5).

By cross-referencing with sequence data and comparing gel patterns with the 50 base pair reference ladder on the gels, I was able to ascertain the lengths of restriction fragments in the digest photographs. If band lengths did not add up to the approximate total length of the PCR product (375 bp), other scenarios were considered. These included the presence

(39)

a) b)

Figure 5: Gel images of ApoI (a), and MseI (b) restriction enzyme banding patterns. Band sizes are given once for each banding

(40)

of more than one band of the same length, the presence of very short bands that migrated to the end of the gel and thus were not detected, and the possibility of length variation in the PCR product (resulting from an insertion or deletion). In several cases these

irregularities were resolved by sequencing the samples in question. I determined restriction digest haplotypes by combining the fragment patterns and lengths of all four restriction enzyme digests. Each unique combination of fragment patterns and indels was deemed a haplotype and given a numeric code (Table 6).

2.4 Data Analyses

2.4.1 Phylogenetic analyses

2.4.1.1 Sequence assembly and alignment

I assembled forward and reverse sequence chromatograms into consensus sequences using SeqMan II (SeqMan 5.07, DNAStar Inc., Madison, WI). Sequences were aligned in ClustalX 1.81 (Thompson et al. 1997). All sequences were checked against sequence traces (chromatograms). I used the online service “Reverse

Complement” (http://www.bioinformatics.org/ sms/index.html) to get complementary sequences where needed. In preparation for phylogenetic analyses, the aligned and trimmed sequence datasets were converted to NEXUS format (Maddison et al. 1997) using ClustalX.

2.4.1.2 ITS Analyses

I determined ITS sequence haplotypes for R. integrifolia and R. rhodantha by grouping similar sequences from the ClustalX alignments. These groupings were

(41)

version 1.21 (Clement et al. 2000) (see section 2.4.2). Each R. integrifolia and R.

rhodantha ITS sequence haplotype was given a lower case letter code.

For phylogenetic analyses, two different ITS datasets were used. The first was made up of sequences of 16 species of Rhodiola (including R. rosea) obtained from Genbank (Mayuzumi and Ohba, 2002) (Appendix 2), together with six sequences of R.

integrifolia, one of R. rhodantha and two of R. rosea from my samples. This dataset was

used to determine the phylogenetic position of the three North American Rhodiola species (R. integrifolia, R. rosea and R. rhodantha) within the genus. On the basis of its reportedly very close relationship with the genus Rhodiola (Mayuzumi and Ohba, 2004), a Genbank sequence of Pseudosedum sp. was used here as an outgroup. The second dataset consisted of 19 individuals of R. integrifolia and one of R. rhodantha together with one sequence of R. rosea used here as an outgroup. Only individuals for which both ITS and psbA-trnH sequences were available were included in this dataset which was used to clarify the position of R. integrifolia with respect to the other two North American Rhodiola species.

2.4.1.2.1 Maximum Parsimony and Maximum Likelihood Analyses using PAUP

PAUP 4.0 beta test version 4.0b (Swofford, 2002) was used to analyze each dataset for the best tree, using three different methods: neighbour-joining, maximum parsimony (MP), and maximum likelihood (ML). Initially, neighbour-joining trees were produced. I then carried out heuristic searches using the Maximum Parsimony (MP) optimality criterion. I performed each heuristic search with 100 replicates added randomly (holding one tree at each step). In all analyses gaps were treated as missing data (causing one single base repeat to be omitted). Starting trees were obtained stepwise

(42)

and added in a random sequence. Tree-bisection–reconnection (TBR) branch swapping, and “multrees” were in effect. All characters were treated as unordered and equally weighted. A bootstrap consensus tree was estimated from 500 bootstrap replicates.

In order to determine the best maximum likelihood (ML) model I used the online version of ModelTest 3.8, ModelTest Server 1.0 (http://darwin.uvigo.es/software/ modeltest_server.html) (Posada, 2006). This freeware program allows the user to determine the most appropriate model of nucleotide substitution to use for a specific dataset in a maximum likelihood phylogenetic analysis. It selects from among 56 models and presents an output file with the best model based on both Hierarchical Likelihood Ratio Tests (hLRTs) and the Akaike Information Criterion (AIC), with a set of

commands that can be inserted into a PAUP block or added from the command-line in a terminal interface version of PAUP (Posada and Crandall, 1998). Although the hLRTs have been the more often used model selection strategy, I chose the AIC, considered by Posada & Buckley (2004) to be the superior approach. The optimal ML analysis model was determined to be the Tamura/Nei (1993) equal base frequency model with gamma shape parameter distribution (TrNef+G), (AIC), which assumes base frequencies to be equal and the proportion of invariable sites to be zero. Starting trees were obtained via stepwise addition with random addition of sequences. Any branch lengths less than or equal to 1e-08 were collapsed (creating polytomies). Maximum likelihood phylogenetic analyses in PAUP were run on the two 600 bp ITS sequence datasets following this model.

(43)

2.4.1.3 Phylogenetic analyses of Chloroplast DNA sequences

I assembled and aligned sequences of the psbA-trnH spacer region of the chloroplast genomes of R. integrifolia, R. rhodantha and R. rosea as outlined above. Sequence haplotypes for R. integrifolia and R. rhodantha were determined by grouping similar ClustalX sequence alignments, and haplotype matrices and networks were produced as outlined below. Each sequence haplotype was given an upper case letter code. Rhodiola rosea was used as the outgroup in all cpDNA sequence and RFLP analyses. Analyses were performed on datasets in NEXUS format containing one example of each haplotype. Indels in the chloroplast sequences were treated as single characters, as was a four-base inversion (See Appendix 3 for sequence alignments).

Two sets of psbA-trnH sequences were used in phylogenetic analyses. The first was made up of one sequence of each haplotype. This included 27 sequences of R.

integrifolia, one R. rhodantha sequence, and one sequence of R. rosea included as the

outgroup. The second dataset was made up of a subset of the samples that were also sequenced for ITS. This dataset contained 19 R. integrifolia sequences and one R.

rhodantha sequence.

Phylogenetic trees were constructed for both datasets according to both the Maximum Parsimony and the Maximum Likelihood optimality criteria. ModelTest was run on these datasets, as described above, to determine the optimal model for maximum likelihood phylogenetic analyses. ML analysis was performed by PAUP following the TIM+I+G model (AIC) using nucleotide frequencies calculated from the dataset (A=0.3872, C=0.1486, G=0.1186 and T=0.3456). Starting trees were obtained via

Systematic and phylogeographic implications of molecular variation in the western North American roseroot, Rhodiola integrifolia (Crassulaceae)

Systematic and phylogeographic implications of molecular

variation in the western North American roseroot,

Rhodiola integrifolia (Crassulaceae).

Systematic and phylogeographic implications of molecular

variation in the western North American roseroot,

Rhodiola integrifolia (Crassulaceae).

Abstract

Table of Contents

List of Tables

List of Figures

Acknowledgements

Dedication

Chapter 1: INTRODUCTION

Beringia

Beringia

Siberia

Chapter 2: METHODS