Citation
Evolutionary and ecological implications of polyploidy in Eutrema Edwardsii (Brassicaceae)

Material Information

Title:
Evolutionary and ecological implications of polyploidy in Eutrema Edwardsii (Brassicaceae)
Creator:
Mastin, Jared Edmond ( author )
Place of Publication:
Denver, CO
Publisher:
University of Colorado Denver
Publication Date:
Language:
English
Physical Description:
1 electronic file (54 pages) : ;

Thesis/Dissertation Information

Degree:
Master's ( Master of science)
Degree Grantor:
University of Colorado Denver
Degree Divisions:
Department of Integrative Biology, CU Denver
Degree Disciplines:
Biology
Committee Chair:
Bruederle, Leo P.
Committee Members:
Hufft, Rebecca
Wunder, Michael

Subjects

Subjects / Keywords:
Cruciferae ( lcsh )
Chromosome numbers ( lcsh )
Chromosome numbers ( fast )
Cruciferae ( fast )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Abstract:
Polyploidy is a ubiquitous phenomenon that is most frequent in the arctic where glaciation cycles put selective pressures on populations by repeated fragmentation. Several decades of research suggest that polyploids would have been more fit than diploids in novel habitats as glaciers receded. The higher fitness of polyploids has been attributed to increased genetic material and novel gene products, which result in phenotypic plasticity and rapid adaptation. These concepts lead to the question; Does polyploidy increase the ecological tolerances of a species thereby facilitating colonization outside the niche of the progenitors? To answer this question, a polyploid complex was investigated to address three objectives: 1) determine the polyploid mode of origin; 2) document the distribution polyploid cytotypes; and 3) discover the relationship between ploidy and niche breadth. ( ,,, )
Abstract:
Eutrema edwardsii R. Br. (Brassicaceae) is an arctic-alpine mustard with a near circumpolar distribution. Its closest relative, Eutrema penlandii Rollins, is a federally listed, threatened species that is endemic to the Mosquito Range in the Southern Rocky Mountains of Colorado. Together these species comprise a polyploid complex in North America, for which we conducted chromosome counts, flow cytometry, and allozyme analysis to elucidate the polyploid origins of E. edwardsii and model the niche of polyploid cytotypes to discover environmental correlates with respect to ploidy.
Abstract:
Results obtained from mitotic counts of two populations of E. penlandii reveal this taxon to be diploid. Diploidy was confirmed using flow cytometry for an additional 15 individuals representing four populations. Previously published chromosome counts for E. edwardsii reveal a polyploid complex of tetraploid, hexaploid, and octaploid populations for which an autopolyploid origin has been presumed. However, allozyme analysis revealed an allopolyploid origin for E. edwardsii, as evidenced from fixed heterozygosity at six loci. Flow cytometry revealed all three expected cytotypes among 53 populations. Niche models were created for tetraploid and hexaploid populations to calculate niche overlap (D = 0.354, I = 0.654), which revealed divergence.
Abstract:
The data reported herein support the recognition of E. penlandii as taxonomically distinct, reveal cryptic variation within E. edwardsii, and support niche divergence between tetraploids and hexaploids.
Bibliography:
Includes bibliographical references.
System Details:
System requirements: Adobe Reader.
General Note:
n3p
Statement of Responsibility:
by Jared Edmond Mastin.

Record Information

Source Institution:
University of Colorado Denver
Holding Location:
Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
on10129 ( NOTIS )
1012944334 ( OCLC )
on1012944334
Classification:
LD1193.L45 2017m M37 ( lcc )

Downloads

This item has the following downloads:


Full Text
EVOLUTIONARY AND ECOLOGICAL IMPLICATIONS OF POLYPLOIDY
IN EUTREMA EDWARDS!! (BRASSICACEAE) By
JARED EDMOND MASTIN B.S., University of Colorado Denver, 2015
A thesis submitted to the Faculty of the Graduate School of the University of Colorado in partial fulfilment of the requirements for the degree of Master of Science Biology Program
2017


This thesis for the Master of Science degree by Jared Edmond Mastin has been approved for the Integrative Biology Program
by
Leo P Bruederle, Chair Rebecca Hufft Michael Wunder
Date: May 13, 2017


Mastin, Jared Edmond (M.S., Biology Program)
Evolutionary and Ecological Implications of Polyploidy in Eutrema edwardsii (Brassicaceae) Thesis directed by Associate Professor Leo P Breuderle
ABSTRACT
Polyploidy is a ubiquitous phenomenon that is most frequent in the arctic where glaciation cycles put selective pressures on populations by repeated fragmentation. Several decades of research suggest that polyploids would have been more fit than diploids in novel habitats as glaciers receded. The higher fitness of polyploids has been attributed to increased genetic material and novel gene products, which result in phenotypic plasticity and rapid adaptation. These concepts lead to the question; Does polyploidy increase the ecological tolerances of a species thereby facilitating colonization outside the niche of the progenitors? To answer this question, a polyploid complex was investigated to address three objectives: 1) determine the polyploid mode of origin; 2) document the distribution polyploid cytotypes; and 3) discover the relationship between ploidy and niche breadth.
Eutrema edwardsii R. Br. (Brassicaceae) is an arctic-alpine mustard with a near circumpolar distribution. Its closest relative, Eutrema penlandii Rollins, is a federally listed, threatened species that is endemic to the Mosquito Range in the Southern Rocky Mountains of Colorado. Together these species comprise a polyploid complex in North America, for which we conducted chromosome counts, flow cytometry, and allozyme analysis to elucidate the polyploid origins of E. edwardsii and model the niche of polyploid
cytotypes to discover environmental correlates with respect to ploidy.


Results obtained from mitotic counts of two populations of E. penlandii reveal this taxon to be diploid. Diploidy was confirmed using flow cytometry for an additional 15 individuals representing four populations. Previously published chromosome counts for E. edwardsii reveal a polyploid complex of tetraploid, hexaploid, and octaploid populations for which an autopolyploid origin has been presumed. However, allozyme analysis revealed an allopolyploid origin for E. edwardsii, as evidenced from fixed heterozygosity at six loci. Flow cytometry revealed all three expected cytotypes among 53 populations. Niche models were created for tetraploid and hexaploid populations to calculate niche overlap (D = 0.354, / = 0.654), which revealed divergence.
The data reported herein support the recognition of E. penlandii as taxonomically distinct, reveal cryptic variation within E. edwardsii, and support niche divergence between tetraploids and hexaploids.
The form and content of this abstract are approved. I recommend its publication.
Approved: Leo P Bruederle
IV


ACKNOWLEDGEMENTS
I would like to acknowledge my advisor, Leo P. Bruederle, Rebecca Hufft (Associate Director of Applied Conservation, Denver Botanic Gardens) and Michael Wunder (Associate Professor, University of Colorado Denver) for serving on my thesis committee and for helpful conversations and advice about flow cytometry, data analysis and niche modeling. I also acknowledge: Robert Laport for our many meetings about niche modeling and flow cytometry; James Salmen (Director of Facilities and Laboratory Manager, University of Colorado Denver), who provided technical support with respect to flow cytometry; Eileen Yakish, Hannah Tystad, and Daniel Harper, then undergraduates at CU Denver who collected preliminary allozyme data for this study; and Suzanne Meinig, who collected flow cytometry data and georeferenced herbarium samples. The research was supported by a contract awarded to LPB by the U.S. Fish SiWildlife Service (F10AC00603).
v


TABLE OF CONTENTS
CHAPTER
I. POLYPLOID ORIGINS OF EUTREMA EDWARDSII.......................................1
Introduction.................................................................1
Materials and Methods........................................................2
Results......................................................................7
Discussion...................................................................9
Figures and Tables..........................................................14
II. ECOLOGICAL NICHE MODELING REVEALS DIVERGENCE BETWEEN TETRAPLOID
AND HEXAPLOID POPULATIONS...................................................21
Introduction................................................................21
Materials and Methods.......................................................26
Results.....................................................................28
Discussion..................................................................30
Tables and Figures..........................................................38
REFERENCES.........................................................................43
VI


CHAPTER I
ALLOPOLYPLOID ORIGINS OF EUTREMA EDWARDSII Introduction
Although flowering plants are under-represented in the literature addressing cryptic species (Bickford et al. 2006), there is a growing body of research addressing these cryptic species in the flora of the arctic in particular (e.g., Grundt et al. 2006). The underlying explanations proposed to explain this phenomenon include recent speciation, incomplete divergence, and reticulate evolution, with selection, inbreeding, and genetic drift proposed as mechanisms leading to cryptic variation. It is also clear that polyploidy — particularly autopolyploidy in plants — is an important and overlooked mechanism leading to the formation of cryptic species (Soltis et al. 2007). However, there has been a resistance to recognize multiple cytotypes as distinct — even when they confer reproductive isolation — as well as a reliance upon the morphological species concept, or some variant thereof (Soltis et al. 2007). Herein, we report the results of multidisciplinary research addressing the systematics of the Eutrema edwardsii R. Br. species complex (Brassicaceae), with a goal of contributing to a better understanding of E. penlandii Rollins, a federally listed Colorado endemic.
Until 1985, two Eutrema species were recognized from North America: E. edwardsii and E. penlandii, when Weber (1985) published the combination E. edwardsii R. Br. subsp. penlandii (Rollins) W.A. Weber, thereby reducing the rank of this taxon to subspecies. More recently, E. penlandii was subsumed into E. edwardsii, as part of a larger revision to the genus following a critical evaluation of Neomartinella Pilger, Platycraspedum O.E. Schulz,
1


Taphrospermum C.A. Meyv and Thellungiella O.E. Schulz; this resulted in the aforementioned genera being subsumed into Eutrema (Al-Shehbaz and Warwick 2005, Warwick et al. 2006). Therein, it was suggested that the variation exhibited by E. edwardsii could easily accommodate E. penlandii. As part of a larger study addressing conservation genetics in E. penlandii (Hardwick 1997), fixed heterozygosity was observed in populations of E. edwardsii suggesting to us an allopolyploid origin for the latter.
Although it has long been assumed that the aforementioned two species form a polyploid complex, for which we presupposed autopolyploidy for E. edwardsii (e.g., Jprgensen et al. 1958), no chromosome counts have been reported for E. penlandii, in particular. Herein, we take multiple approaches using chromosome counts, flow cytometry, and allozyme analysis to better understand the systematics of E. edwardsii s.l., and to test the hypothesis that E. penlandii and E. edwardsii form an autopolyploid complex. Whereas chromosome counts and flow cytometry are used to document ploidy for E. penlandii, allozyme analysis is used to elucidate the polyploid origin of E. edwardsii.
Materials and Methods
Natural history. Eutrema edwardsii and E. penlandii are perennial herbs that occupy a variety of imperfectly drained arctic and alpine tundra habitats, such as meadows, margins of ponds and stream banks, and solifluction slopes (Aiken et al. 2011). Eutrema edwardsii is widespread, with a near circumpolar distribution from the Chukotka in Asia across North America to Norway, where it is common in Svalbard (Fig. 1). Eutrema edwardsii also occurs in the alpine of the White Mountains, Alaska Range, Chugach Mountains, Wrangell-St. Elias Mountains, and the Mackenzie Mountains of the northern Rocky Mountains in western
2


North America and in the Altai Mountains, Pamir Mountains, and Himalayas in Asia. In contrast, E. penlandii is a narrow endemic that is restricted in distribution to the Mosquito Range in the Southern Rocky Mountains of Colorado, where typically it occupies alpine peatlands overlying calcareous substrates. Relatively little is known about the natural history (e.g., reproductive biology) of either species. Although chromosome counts conducted for E. edwardsii have revealed tetraploid (4x=28), hexaploid (6x=42), and octaploid (8x=56) populations, where n = 7 (Table 1), no counts have been reported for E. penlandii (Warwick and Al-Shebhaz 2006).
Field Collections. Plant material, including leaf tissue for allozyme analysis and flow cytometry and seeds for chromosome counts, was collected between 1995 and 2014. As part of a larger study using allozyme analysis to characterize population genetic diversity and structure in E. penlandii, Hardwick (1997) collected leaf tissue from seven populations of the narrowly distributed Colorado endemic and three populations of its widespread congener E. edwardsii from Alaska (Table 2). L Bruederle subsequently collected leaf tissue from an additional five populations of E. penlandii and seven populations of E. edwardsii from Alaska and Yukon Territory; K. Marr (Royal BC Museum) provided material from a Siberian population of E. edwardsii. Approval for tissue collection from E. penlandii was granted by the U.S. Fish and Wildlife Service under the authority of permits PRT-704930 and TE12513A-0.
Chromosome counts. Counts for this research were obtained by N. Luebke (Milwaukee Public Museum) from squashes of root tips harvested from seedlings germinated specifically for this purpose. In the process of collecting tissue for population genetic and
3


molecular systematics research, a small number of fruits were harvested. Seeds were germinated in 0.005 M giberellic acid in a Percival environmental chamber set at 15°C; attempts were made to cultivate those seedlings that were not harvested in a research greenhouse for subsequent use for these purposes. As such, either whole seedlings or root tips were harvested within 2-4 hours after light conditions were restored and placed in a saturated solution of paradichlorobenzene for two hours, after which they were fixed in Farmer's solution, in which they were stored. Squashes were prepared by macerating the root tip in a drop of 1% aceto-orcein on a microscope slide cleaned with 95% ethanol, heating for two mins at 50° C, and squashing. Squashes were scanned at lOOx and 400x for meiotic figures and photographed at lOOOx. The chromosome count for E. penlandii was obtained from seed collected at the Treasurevault Mountain site, for which a voucher has been deposited with the Katherine Kalmbach Herbarium (KHD) at the Denver Botanic Gardens.
Flow cytometry. Relative genome size was determined from silica-dried samples to confirm ploidy for E. penlandii and E. edwardsii based upon chromosome counts, and to document ploidy at sites across the range of E. penlandii. Flow cytometry methods followed Dolezel et al. (2007).
Genome size was obtained for three individuals from each population, depending upon the amount of tissue that was available. Standards were grown in the research greenhouse at the University of Colorado Denver, with leaf tissue harvested and silica gel dried for subsequent use. Pisum sativum 'Bush Sugar' differed enough from both Eutrema species to ensure no overlap in scatter plots and histograms generated by flow cytometer (FCM)
4


software. The standards were included as part of each assay in order to estimate ploidy for
the sample peak and to adjust sample values based on differences in standard value, which fluctuated among assays. The fluorescence value for the sample was multiplied by the ratio of standard fluorescence value to the average value for the standard across all assays.
Between l-2cm2 of leaf tissue was minced using a razor blade along with an internal standard in 1ml of LB01 buffer solution pH 7.5 for 1-2 minutes. The homogenate was mixed using a pipettor and then incubated on ice for 15 minutes. The solution comprising nuclei in suspension was filtered through 42pm nylon mesh into a 1.5ml Eppendorf tube (Dolezel et al. 2007). RNase-A (0.25pl at 100mg/l) was added to 0.5ml of the suspension and allowed to incubate on ice for 15 minutes. Propidium iodide (PI, 25pl at lg/ml) was added to the suspension, which was gently mixed and incubated on ice for 3-5 minutes in the dark. The solution was then injected into the FCM with the sample rate set to 'slow' and run until approximately 100,000 events were recorded. The nuclei suspension, without the addition of PI, could be kept in a refrigerator overnight without a decrease in the quality of flow cytometry output.
Clusters of G1 nuclei outside the region of the standard were gated to create a histogram and calculate mean peak value and coefficient of variation (CV). Assays with a CV over 5% were discarded. Five consecutive measurements were obtained to collect roughly 500-1000 events in the gated region. The average of the five mean peak values was recorded for each sample. Histograms were created that revealed the relative difference between the sample and the standard.
5


Ploidy was estimated recognizing that E. penlandii is diploid (2n), as documented through the aforementioned chromosome count, and comparing the mean fluorescence of both taxa following Dolezel et al. (2007).
Allozyme analysis. Soluble enzymatic proteins were extracted from two fresh leaves ground in 0.25ml of a cold Tris-HCI grinding buffer modified by the addition of 4% w/v PVP-40 and 1% w/v 2-mercaptoethanol (Soltis et al. 1983); a small amount of sea sand facilitated grinding. Extracts were adsorbed onto 1.5 x 11mm wicks cut from No. 17 Whatman chromatography paper and stored at -80°C until electrophoresis.
Protein extracts were applied to each of three different starch gel and electrode buffer systems (Nielsen and Johansen 1986): lithium-borate, gel buffer pH 8.3/electrode buffer pH 8.1; histidine-HCl-tris, gel buffer pH 7.5/electrode buffer pH 7.5; tris-citrate, gel buffer pH 7.8/electrode buffer pH 8.7. Electrophoresis was conducted at 4°C until a bromophenol blue marker had migrated approximately 12cm.
Following electrophoresis, gels were sliced horizontally into approximately 1.5mm thick slices, with interior slices incubated in substrate specific stains (Soltis et al. 1983) previously revealed to be informative with respect to the polyploid origin of E. edwardsii (Bruederle unpub. data). The lithium borate gel was stained for phosphoglucoisomerase (PGI-2, specifically); superoxide dismutase (SOD-1); and triose-phosphate isomerase (TPI-1); the histidine-HCL gel was stained for malate dehydrogenase (MDH-1); and the tris-citrate gel was stained for aspartate aminotransferase (AAT-1). Enzyme nomenclature follows the International Union of Biochemistry (1984).
6


Designation of loci and alleles was based on relative mobility of proteins following Hardwick (1997), with the most rapidly migrating loci receiving sequentially higher number and the alleles at a locus similarly assigned sequential higher lowercase letters. No formal genetic analyses (e.g., controlled crosses) were performed to document patterns of inheritance. Putative genetic loci and genotypes were inferred from the known substructure and intracellular compartmentalization of enzymes, as well as electrophoretic patterns observed in individuals presumed heterozygous at polymorphic loci. For each population, data were collected as individual genotypes from which ploidy was extrapolated based upon dosage with support provided using flow cytometry with the same population samples. Deviations from Hardy-Weinberg expectations assuming disomic inheritance were tested using Chi-square analysis.
Results
Chromosome counts, DNA content, and ploidy. Chromosome counts reported here from the Treasurevault Mountain site revealed E. penlandii to be diploid, with 2x = 14 and n = 7 (Fig. 2). Fluorescence during flow cytometry ranged from 48,860-53,024, with a mean fluorescence of 49,907 for E. penlandii (Table 3).
By contrast, fluorescence for E. edwardsii populations ranged from 102,601 - 180,523, suggesting that our sample included both tetraploid and hexaploid individuals (Fig. 3). Whereas three populations from the Arctic Coastal Plain in northern Alaska and the Chugach Range in southern Alaska were tetraploid, those populations sampled from Central Alaska and Yukon were hexaploid. Coefficient of variation for these measurements were between 2-5%; typically, 500-2000 nuclei were detected in each run.
7


Although field observations, such as relatively tall habit and large leaf morphology observed in some individuals, suggested infrequent autopolyploidy in some populations of E. penlandii, this was not substantiated by flow cytometry — all individuals of E. penlandii sampled here were found to be diploid. In contrast, E. edwardsii was exclusively polyploid, as expected based upon counts previously reported in the literature; populations sampled here are either tetraploid or hexaploid, with no evidence of diploidy. Although octaploids have been reported in the literature, none were observed as part of this study.
Comparisons between populations off. penlandii and f. edwardsii provided evidence supporting polyploidy in the latter. Eutrema edwardsii populations sampled for this study were tetraploid or, more commonly hexaploid. Furthermore, relative genome size in hexaploid f. edwardsii was greater than three times that of f. penlandii, which could be the result of genome modification after polyploidization. Or, this could be the result of another, unknown, cryptic taxon with a slightly larger genome being one of the other progenitors to f. edwardsii.
Allozyme analysis. Six loci exhibited disomic inheritance and fixed heterozygosity in one or more of the E. edwardsii populations: AAT-2, MDH-1, PGD-3, PG1-2, SOD-1, and TPI-1. Chi-squared analyses revealed that genotypic frequencies at all six loci deviated from expectations of independent assortment of chromosomes (p < 0.05), refuting the hypothesis of an autopolyploid origin for f. edwardsii.
Seventeen alleles were observed at the six polymorphic loci in f. edwardsii — four alleles comprising TPI-1 and MDH-1, three alleles at AAT-1, and two alleles each for the other loci. Eutrema penlandii shared nine of the alleles found in f. edwardsii, suggesting a
8


progenitor-derivative relationship for this species. For example, whereas E. penlandii has
the b, c, and d alleles, tetraploid E. edwardsii show fixed heterozygosity for the "b" and "c" alleles (i.e., bbcc), with all the hexaploids also having the "a" allele (e.g., aabbcc), presumably contributed by a third progenitor.
We also found evidence for multiple origins of E. edwardsii based upon variability for dosage at loci that exhibited fixed-heterozygosity. Hexaploid populations off. edwardsii exhibited as many as eight different heterozygous genotypes per locus (i.e., TPI-1), suggesting that the progenitors to this taxon were polymorphic at these loci.
Although these data are not unequivocal, f. penlandii or a close relative appears to have contributed one genome to tetraploid and to hexaploid f. edwardsii. Furthermore, the allozyme data suggest that one or more extinct or yet to be described diploid Eutrema s.l. species contributed the remaining genome(s) through multiple origins from polymorphic populations.
Discussion
Here, I present genetic evidence documenting the allopolyploid origin of tetraploid and hexaploid populations of f. edwardsii involving two or more different progenitors. In tetraploids having an allopolyploid origin, hybridization between two individuals with divergent genomes is followed by chromosome doubling resulting in polyploid progeny characterized by two complete genomes (Soltis and Soltis 2000). In contrast to autopolyploidy, where chromosomes segregate and pair independently during meiosis (polysomic inheritance), chromosomes are inherited disomically with allopolyploidy; that is, chromosomes of the same genome pair during meiosis, which typically results in fixed-
9


heterozygosity at loci that were divergent in the parental taxa. Although these represent
the extreme forms of polyploidy, the six tetraploid and hexaploid populations of E. edwardsii studied here exhibited a pattern of disomic inheritance of chromosomes at each of six loci, four of which exhibited fixed-heterozygosity.
Polyploid species may result from multiple events of polyploidization involving either the same parental progenitors (polytropic), or different progenitors (polyphyletic) (Brochmann et al. 2004). Polyploids involving multiple progenitors can be difficult to discern; like-wise, polyploidy resulting from a single event that with subsequent recombination, gene flow between polyploids or diploids, or mutations can cause an overestimate of the number of events of polylploidization (Wallace 2003). Although our data reveal multiple origins for E. edwardsii, they also suggest that tetraploids and hexaploids arose from hybridization events involving the same two or three progenitors. These genomic combinations can have effects on closely related taxa resulting in species misidentification. Hidden variation in morphologically cryptic species can lead to false conclusions of relatedness when a species may have greater variation than observed, indicating divergence. Recently diverged species that have undergone polyploidization may appear to be autopolyploid; however, disomic inheritance still occurs (Brochmann et al. 2004).
We also provide evidence suggesting that E. penlandii is the closest living descendant of one of the diploid progenitors of E. edwardsii (Paun et al. 2009). Although we are unable to identify the closest living descendant(s) of the other progenitor(s), attempts at genome reconstruction coupled with the number of loci at which fixed-heterozygosity was observed
10


suggests that they were genetically divergent representatives of Eutrema s.l. Given the morphological similarity of E. penlandii and E. edwardsii, it seems likely that the progenitors were morphologically cryptic, despite genetic differentiation, as has been demonstrated in other genera, such as Draba (Grundt et al. 2005). Given the genetic differentiation that we see among the progenitors, it seems likely that hybridization resulted in increased unreduced gamete formation in the hybrids, with whole genome duplication resulting from the fusion of these unreduced gametes (Paun et al. 2009). Different factors have been correlated with unreduced gamete formation (Ramsey and Schemske 1998; Mason and Pires 2015), several of which may have contributed to the formation of E. edwardsii (e.g., hybridization, stress).
Polyploidy is extremely common in the flora of the arctic, where fluctuations in climate have resulted in cycles of fragmentation and range expansion (Brochmann et al. 2004). Genetic drift, exacerbated by selfing would have resulted in population differentiation, which has been correlated with an increased ability to colonize new sites associated with deglaciation (Brochmann et al. 2004).
As previously discussed, taxonomic treatments of E. penlandii differ markedly, with the recent treatments of Al-Shehbaz and Warwick (2005) and Weber and Wittmann (2012) subsuming E. penlandii into E. edwarsdsii. However, genetic data presented herein reveal that the taxonomic boundaries between E. edwardsii and E. penlandii are obscured due to a progenitor-derivative relationship involving allopolyploidy and morphologically cryptic progenitors.
11


Eutrema penlandii is a relictual Arcto-Tertiary element in the Colorado flora (Weber 2003), with the greatest species diversity within the genus found in Asia. Given the strong evidence for an allopolyploid origin for E. edwardsii, the diploid progenitors were presumably sympatric in high latitudes at the end of the Tertiary and during the Pleistocene, when hybridization coupled with unreduced gamete formation — we found no evidence of somatic doubling (e.g., mixoploidy) — and selfing resulted in whole genome duplication and the cytotypes observed in the polymorphic, polyploid E. edwardsii. Given the genetic differences deduced from the allozyme data, the progenitor taxa were genetically well differentiated, much more so than expected for sibling species. And while morphometric analyses examining species boundaries between E. penlandii and E. edwardsii reveal significant differences, they are admittedly subtle (Regier, Pansing, and Bruederle, unreported data). This uncoupling between morphological and genetic differentiation highlights the importance of recognizing cryptic species, especially with regard to polyploid complexes. Polyploidization between genetically divergent species has immediate consequences for genetic diversity of the progeny. Fixed heterozygosity can preserve genetic diversity when drift occurs due to founder effect or when plants self (Abbott and Brochmann 2003). Populations of autopolyploids, however, will be more susceptible to loss of genetic diversity due to polysomic inheritance coupled with drift or selfing. Regardless, multiple forms of evidence, in addition to the cytotype differences reported herein — molecular sequence data, morphology, and distribution — reveal E. penlandii to be distinct. Furthermore, it is reproductively isolated within E. edwardsii s.l. and, as such, we propose recognition of this taxon at the species level, despite it being morphologically cryptic.
12


Based on the alleles present in each species, fixed heterozygosity, and ploidy there is
evidence of the E. penlandii genome in both the tetraploids and hexaploids and that there must have been multiple genome duplication events and at least one additional unidentified progenitor. This is supported by variation in dosage observed at fixed-heterozygous loci in both tetraploids and hexaploids. Variation in dosage of alleles at fixed-heterozygous loci is the result of multiple polyploidization events from a polymorphic progenitor (Wyatt et al. 1989).
Finally, although we studied only tetraploid and hexaploid populations off. edwardsii, we cannot exclude the possibility that additional taxa were involved in the formation of this taxon, nor can we make conclusions regarding the origin of octaploid populations. Although octaploidy has been reported for this taxon (Knaben 1968), no octaploid populations were included as part of this research and, as such, we are unable to address its origin here.
Ongoing research addressing polyploidy in f. edwardsii s.l. includes mapping cytotypes across the range of f. edwardsii and ecological niche modeling of cytotypes comprising the f. edwardsii species complex to detect ecological differentiation.
13


Figures and Tables
â–  -vCr -
• VS^
■# • •7*
•v:*-.
.r /H
**-f* •* *
Fig. 1. Distribution of E. edwardsii s.l. Occurrences obtained from GBIF; map created by Dr. P. Anthamatten from the Department of Geography and Environmental Sciences, University of Colorado Denver. Near circumpolar distribution shows highest frequency in Beringia (Alaska and Eastern Siberia) and Northern Canada, extending to Svalbard, Norway. Additional occurrences in the Altai Mountains, Pamir Mountains, and Himalayas in Asia. Occurrences in Colorado represent E. penlandii Rollins.
14


Table 1. Locations lor Eutrema penlandii Rollins and E. edwardsii R. Br. (Brasslcaceae) populations sampled for chromosome counts1, flow cytometry1, and allozyme analysis*. Coordinates for E. penlandii are withheld due to federal listing as threatened.
Site Latitude Longitude Elevation (m) Population Size Application
Slue Lakes NA NA 3958 <10 2
Cameron Amphrtheater (3A) NA NA 3978-4039 -3000 3
Cooney Lake NA NA 3767-3889 500-600 3
Hilltop Mine {Hilltop Mine 13A) NA NA 3933 200-300 3
Hoosier Ridge NA NA 3872 -300 3
c Peerless Mine {near Peerless Mine/Horseshoe Basin ISA) NA NA 3952.0 <50 2,3
"5 Horseshoe Gulch {near Peerless Mine/Horseshoe Basin 15C) NA NA 3758 200-300 2,3
Q.
a E Kite Lake (4A) NA NA 3780-3812 200+ 3
Hj Mosquito Pass (8A-E) NA NA 4009 1000+ 3
Uj Mount Buckskin (North Mosquito Creek 16A-D) NA NA 3872 500+ 3
Mount Sheridan (14B) NA NA 3984 200+ 3
North London (American Flats/London Mountain 9A) NA NA 3857 1000+ 3
Treasurevault Mountain NA NA 3901 <100 1
Weston Pass NA NA 3612-3627 100-200 2,3
Barrow, East 71.23917 -156.3358 -1 500-1000 2,3
_â–  Barrow, West 71.26412 -156.8296 9.1 500-1000 2,3
a vj

1 O Campbell Creek 61.06879 -149.6056 ~823.9 ~100 2,3

5 -g Dempster Highway 64.95168 -138.2713 933.6 -100 3
Eagle Summit 65.4849 -145.404 1125.3 -100 2,3


Table 2. Allele frequencies at six polymorphic loci in populations of E. penlandii Rollins and E. edwardsii R. Br, (Brassicaceae), where n - sample size and H0 - observed heterozygosity._____________________________________________________________________________________
Eutremo edwardsii (N = 5} Eutrema peniandii (N =â–  12)
site (ploidy) site
Locus Allele Barrow East, AK (4n) Barrow West, AK (4n) Campbell Creek, AK (4n) Dempster Highway, AK (6n) Eagle Summit, AK (On) Cameron Amphitheater, CO Cooney Lake, CO Hilltop Mine, CO Hoosier Ridge, CO Horseshoe Gulch 1, CO Horseshoe Gulch 2, CO Kite Lake, CO Mosquito Pass, CO Mt Buckskin, CO Mt Sheridan, CO North London Mine, CO Weston Pass, CO
AAT-1 3 0.000 0.000 0.412 0.597 0.556 1.000 1.000 1,000 1.000 na 1.000 1.000 1.000 1.000 na 1.000 1.000
b 1.000 0.990 0.588 0.403 0.444 0.000 0.000 0,000 0.00 na 0.000 0.000 0.000 0.000 na 0.000 0.000
c 0.000 0.010 0.000 0.000 0.000 0.000 0.000 0.000 0.000 na 0.000 0.000 0.000 0.000 na 0.000 0.000
n 25 25 17 24 24 25 25 22 26 na 17 7 25 25 na 14 25
Hj 0.000 0.040 1.000 1.000 1.000 0.000 0.000 0.000 0.000 na 0.000 0.000 0.000 0.000 na 0.000 0.000
MOH-1 3 0.000 0.000 0.000 0,590 0.667 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000
b 0.750 0.750 0.750 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000
c 0.000 0.000 0.000 0.410 0.333 1.000 1.000 1,000 1.000 1,000 1.000 1.000 1.000 1.000 1.000 1.000 1.000
d 0.250 0.250 0.250 0.000 0.000 0.000 0.000 0,000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000
n 25 22 34 24 24 25 25 24 26 19 17 8 25 25 8 14 25
Hj 1.000 1.000 1.000 1.000 1.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000
PGD-3 0 0.000 0.000 0.000 0.000 0.000 0.000 0.030 0.000 0.025 0.000 0.000 0.000 0.000 0,040 0.000 0.000 0.000
3 0.000 0.000 0.500 0.1G7 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000
b 1.000 1.000 0.500 0.833 1.000 1.000 0.970 1,000 0.975 1,000 1.000 1.000 1.000 0.960 1.000 1.000 1.000
n 25 22 34 20 24 25 25 24 20 19 17 8 25 25 8 14 25
Hj 0.000 0.000 1.000 0.500 0.000 0.000 0.120 0.000 0.050 0.000 0.000 0.000 0.000 0.080 0.000 0.000 0.000


1.000 1.000 1.000 1.000 1.000 0.000 0.000 0.000 0.038 0.053 na 0.000 0.020 0.000 0.250 0.000 0.000
LI
o
o
Ln
Ln
Ln
Ln
Ln
Ln
Ln

crv
Ln
Ln
I^J
Ln
CL n cr CU X u 3 cr CU X u 3 cr cu
O o o o h-» o o h-» o O
8 Ln o Ln o 8 o o Ln Ln o Ln o o o rvj Ln Ln C3 Ln c
O O O O o o o o
o o o o h-» o o i-» o o
8 Ln o Ln o 8 o o Ln Ln o Ln o b o Ln Ln o Ln C3
o O O o o o o o
o o o o I—1 o o i-» o o
8 Ln o Ln o 8 b o crv Ln o Ln o b o LU -&â–  Ln o Ln o
o O O o o o o o
o o o o M o o o o o
8 u> LU u> LD u> U> o o Ln u> LD b â– sj Ln O Ln b crv Lo LD
o U> U> U> o o "-4 CD
o O O O M p o o O O
8 U> LD U> LD U> Ll> o o h-1 00 u> u> b â– sj bo N> vl CD k> KD
O U> U> U> o Ln CD N>
o O O O o rw Ln o h-1 o rvj Ln H* O
b & U> crv 8 8 o o b o b o b o 8 8
O o O O o o O O
I—1 o o o o Ln o h-1 o Ln H* CD
8 8 8 8 o o b o b o b o 8 8
O o o o o o O O
o o o o o ro o h-1 o 24 o
b Ln 8 2 8 o o b o b o b o 8 8
CD o N-> o o o O O
O o O o o o H* o o CD
b 00 b i—1 8 8 o o ro crv b o b o M rvj cr> Id -t* b Ln
i—1 O o o Ln N> CD
o o o o o o o O
8 u> b 8 b c h-1 CO b o b o b c 19 8 8
O Ln Ln O o o O O
3 Cu 3 Cu 3 Cu 3 Cu 3 CU 3 cu 3 cu 3 CU o b C: 17 h^> 8 o 8
o O o
O O O o O H1 O I-1 CD
8 8 8 8 o e 00 b o o o o o 00 8 8
O O O O o o O O
o o o o o Ln o H1 o Ln o
b 00 b o b r--» b o b C: b o o o o o 8 8
o o o o O O
o o o o o Ln o H1 o Ln 1-* o
s 8 b o 8 b C: b o o o o o 8 8
o O o O o o O O
o o o o o o H1 o CD
u> 'si 8 b 8 o e 00 b o o o o o 00 8 8
Ln O Ln O o o O O
O o O o o o H1 o h-1 o
'vJ 00 8 N> 8 o C: 14 b o o o o o 14 8 8
O O o o O O
O o o o N* Ln o H1 o Ln H* CD
8 8 8 b o o C: b o o o o o 8 8
O O O o o O O
Locus
Allele
Barrow East, AK (4n)
Barrow West, AK (4n)
Campbell Creek, AK (4n)
Dempster Highway, AK (6n)
Eagle Summit, AK (6n)
Cameron
Amphitheater, CO Cooney Lake, CO
Hilltop Mine, CO
Hoosier Ridge, CO
Horseshoe Gulch 1, CO
Horseshoe Gulch 2, CO
Kite Lake, CO
Mosquito Pass, CO
Mt Buckskin, CO
Mt Sheridan, CO
North London Mine, CO
Weston Pass, CO
cu
cr
rr>
Psj
n
o
u
Eutrema edwardsii (N - 5} Eutrema penlandii (N = 12)
site (plordy) site (plordy}


Fig. 2. Chromosome count from root-tip squash of Eutrema penlandii Rollins (Brassicaceae), 14 chromosomes (2n=2x=14) performed by N. Luebke (Milwaukee Public Museum). Root tips were saturated in paradichlorobenzene and fixed in Farmer's solution; chromosomes were stained with 1% aceto-orcein and photographed at lOOOx.
18


Table 3. Flow cytometry fluorescence data for E. penlandii Rollins and E. edwardsii R. Br.(Brassicaceae) populations sampled for ploidy determination. An internal standard Pisum sativum 'Bush Sugar' was used to account for machine fluctuation.
Taxon Site # Sample CV Standard CV Est. Ploidv 2n
E. edwardsii
Barrow East, AK 03 102601 4.6 441379 4.6 4.1 4x
07 108701 4.7 490736 4.0 4.4 4x
10 102973 4.7 487346 4.5 4.1 4x
Barrow West, AK 03 107623 4.8 515001 4.6 4.3 4x
12 105710 2.7 511906 6.1 4.2 4x
29 107684 4.4 470868 4.8 4.3 4x
Campbell Creek, AK 23 109796 4.8 479433 4.4 4.4 4x
27 112054 4.8 472957 4.3 4.5 4x
30 110633 4.7 489050 3.9 4.4 4x
Denali Hwy, AK 03 178615 4.8 565773 4.3 7.2 6x
10 167368 4.8 524336 4.0 6.7 6x
12 167274 4.6 521411 4.1 6.7 6x
Eagle Summit, AK 15 168828 3.7 601424 3.1 6.8 6x
20 171481 4.7 502898 4.4 6.9 6x
25 169061 4.2 520323 4.5 6.8 6x
12 Mile Summit 02 180523 4.7 515548 4.4 7.2 6x
06 168750 4.3 500931 4.3 6.8 6x
10 171250 4.4 507964 3.9 6.9 6x
E. penlandii
Blue Lake 04 49493 4.5 516318 4.8 2.0 2x
05 49312 3.9 547391 4.7 2.0 2x
06 49796 4.2 517080 4.6 2.0 2x
Weston Pass 05 49153 3.3 550131 2.6 2.0 2x
07 49003 4.1 527668 2.5 2.0 2x
08 48860 3.9 519437 3.3 2.0 2x
Horseshoe Gulch 01 49490 6.0 552199 2.9 2.0 2x
08 49350 4.6 516684 2.7 2.0 2x
10 49385 4.6 511149 2.7 2.0 2x
Peerless Mine 1 04 50691 4.1 562289 2.8 2.0 2x
17 49286 4.6 538854 3.2 2.0 2x
Peerless Mine 2 07 53024 4.0 559630 3.0 2.1 2x
10 50545 3.9 527425 2.9 2.0 2x
17 51295 3.6 506255 3.3 2.1 2x
19


Fig. 3. Flow cytometry histogram showing diploid, tetraploid and hexaploid cytotypes of Eutrema edwardsii s.l. R. Br. (Brassicaceae) (left to right). Leaf tissue from three individuals was simultaneously chopped in a buffer to release nuclei, which were stained with propidium iodide (PI) for visualization on the FL2 channel. Fluorescence peaks indicate two and three-fold increases in nucleus size for tetraploids and hexaploids, respectively.
20


CHAPTER II
ECOLOGICAL NICHE MODELING REVEALS DIVERGENCE BETWEEN TETRAPLOID AND HEXAPLOID POPULATIONS OF EUTREMA EDWARDSII R BR. (BRASSICACEAE)
Introduction
Polyploidy or whole genome duplication (WGD) is a common and recurring phenomenon in plants that is now believed to have been involved in the origin of all angiosperms (Doyle and Sherman-Broyles 2017). The ubiquity of polyploidy suggests that there is an increase in fitness conferred by increased genetic diversity, which can alter ecological tolerances and increase adaptability.
There are several mechanisms by which polyploidy can occur, all of which involve unreduced gametes: 1) one-step mechanism, involving two unreduced gametes; 2) autotetraploid triploid bridge, where a triploid is formed and subsequently reproduces with an unreduced gamete; 3) allotetraploid triploid bridge, same as the aforementioned, but involving divergent species; 4) higher-ploidy one-step; and 5) hybridization of autopolyploids (Mason and Pires 2015). The rate of unreduced gamete formation can increase when genetic and environmental factors cause meiotic aberrations, such as pre-meiotic doubling, premature cytokinesis, mutation in synapsis, lack of spindle fiber segregation, and post-meiotic fusion (Sora et al. 2016).
Environmental stress has been shown to increase the frequency of unreduced gamete formation in greenhouse and growth chamber experiments (Ramsey and Schemske 1998); fluctuating temperatures and nutrient stress correlated with higher rates of unreduced
21


gamete formation. Similarly, hybridization has been shown to result in increased unreduced
gamete formation. Ramsey and Schemske (1998) reported a 49-fold increase in unreduced gamete formation for hybrids versus non-hybrids among 22 species of flowering plants. This suggests that adverse conditions can promote polyploidy via the production of unreduced gametes, which in turn, can facilitate adaptation to a new niche.
Niche is generally used to refer to a subset of environmental conditions in which a species can survive and reproduce (Sillero 2011), but often includes biotic interactions and dispersal. However, several different niche concepts exist. The fundamental niche is the broadest and includes the range of all abiotic variables where a species can maintain a viable population (Sillero 2011). The realized niche is where species interactions, such as competition, have not excluded a taxon from their fundamental niche. The occupied niche is the portion of the realized niche in which populations are not restricted by geography, distribution limitations or historical factors (Sillero 2011). The occupied niche is closest to what is being estimated from presence only niche modeling; here, "niche" refers to occupied niche, unless otherwise specified.
Polyploidization may broaden the niche of a species by increasing ecological tolerances as a result of genome duplication. Added chromosomes and genetic material can effect immediate anatomical, morphological, and physiological changes in polyploids (Baker et al. 2017). Similarly, hybridization can have an additive effect on niche, broadening ecological tolerance to that of both progenitors combined, or a transgressive effect, changing niche tolerances as a result of novel gene products (Soltis et al. 2014). Heterosis, or hybrid vigor, is a phenomenon that describes faster growth rate, larger flowers, seeds and pollen and is
22


often exploited in crop breeding; however, hybrids are often unstable or sterile (Miller et al.
2012). Allopolyploidy alleviates the problems associated with hybrids and stabilizes heterosis. Just as intermediate phenotypes are expected from hybrids, intermediate niches can be expected for allopolyploids, such that ideal environmental conditions comprise entirely, or lie between, the tolerances of each progenitor (Harbert et al. 2014). Furthermore, the added genetic diversity conferred by increased heterozygosity and phenotypic plasticity is thought to increase environmental tolerances (Brochmann et al. 1992, Levin 2000). This increased adaptability allowed polyploids to colonize new habitats outside the niche of their progenitors.
Current research is modeling ecological niche in polyploid complexes in order to elucidate environmental factors that influence polyploid distribution and niche differentiation (e.g., McIntyre 2012, Theodoridis et al. 2013, Glennon et al. 2014). Data extracted from climate databases and localities of populations can generate models that predict the probability that a species occurs in a given area (Thompson et al. 2014).
Likewise, niche models of one cytotype can be compared to another to test whether different cytotypes have similar niche requirements (Godsoe et al. 2013).
There are three possible outcomes beside null difference that can result from niche modeling analysis (Glennon et al. 2014): 1) niche expansion, which would indicate that polyploidy confers increased environmental tolerance, allowing them to disperse to habitat outside the niche of progenitors; 2) niche shift, would indicate a change in niche requirements due to genome duplication, which would still support the hypothesis of polyploidy as an evolutionary mechanism (Van Daijk and Bakx-Schotman 1997); and 3) niche
23


contraction, which would suggest that polyploidy does not increase environmental tolerances or, that polyploids have been restricted to limited habitat due to inbreeding, local adaptation or competition from diploid populations. Previous work has provided a large framework into which to fit my model and, as a consequence, any of these results will be informative of niche differentiation among cytotypes.
Several recently published studies provide evidence for ployploids having different niche conditions, wider niche breadth, and a trend toward harsher niche conditions in comparison to diploids. In Chamerion angustifolium (L.) Holub (Onagraceae), Thompson et al. (2014) showed that autotetraploid populations occupied habitats that were warmer and drier than diploid populations. However, not all polyploids exhibit a positive correlation between niche breadth and ploidy. Theodoridis et al. (2013) demonstrated a decrease in niche breadth as ploidy increased in four closely related species of allopolyploid Primula sect. Aleuritai (Primulaceae). They found that although niche breadth decreased with ploidy, each cytotype had a distinct niche based on seven variables of temperature and precipitation.
The results of both Thomposon et al. (2014) and Theodoridis et al. (2013) provide evidence that polyploids have adaptations to novel conditions that allow them to colonize habitat unsuitable to diploids. The distribution of polyploids from an ecological perspective can offer insight to how and why polyploidy is so ubiquitous, particularly in the arctic, where polyploidy is more frequent (Brochmann et al. 2004).
The arctic flora was greatly influenced by glaciation cycles that began in the late Pliocene (Abbott and Brochmann 2003). Plants that survived in the arctic from the Tertiary to the present are those that previously occurred in wetland and riparian habitats (Murray
24


1981). Early in the Quaternary, plants that occurred in the arctic were largely destroyed or
fragmented by advancing ice-sheets (Abbott and Brochmann 2003). However, the region of Northeast Russia and Northwest America, Beringia, remained ice free throughout the Quaternary, there by serving as a refuge. As glaciers receded, some arctic taxa were more successful at recolonizing the deglaciated habitat. The repeated fragmentation, recolonization and reunion of previously isolated populations have created a complex mixture of hybrid species that have been stabilized by genome doubling (Abbott and Brochmann 2003).
Reticulate polyploidization and hybridization has resulted in increasingly high-ploid hybrids characterized by high or fixed-heterozygosity, which maintains genetic diversity in newly formed populations experiencing inbreeding or bottlenecks (Abbott and Brochmann 2003). These hybrid polyploids appear to be better suited for colonizing the recently deglaciated areas, as suggested by the higher ploidies found in glaciated areas and lower ploidies found in the refugia (Brochamann et al. 2004).
Few studies have addressed polyploid niche modeling in arctic species, where polyploidy is more frequent. Here, I use modeling to compare ecological niche in tetraploid and hexaploid populations of E. edwardsii, with a goal of testing the null hypothesis of no difference. For these purposes, ploidy was determined by flow cytometry from herbarium tissue and field collected samples representing the range of E. edwardsii in North America.
A Principle Components Analysis was performed using climatic data to reveal environmental correlates with respect to ploidy. Ecological niche models were then generated to estimate niche breadth, niche overlap, niche identity and niche similarity. The ultimate goal of this
25


research was to determine if there is divergence between tetraploid and hexaploid E. edwardsii, as theory predicts.
Materials and Methods
Field and Herbarium collections. Ninety-three populations off. edwardsii were sampled for ploidy determination from across North America, as well as Beringia, specifically (Table 4). Here, populations are represented by individuals, as no mixo-ploid populations were detected among f. edwardsii populations (Chapter 1). Leaf tissue was obtained from 79 herbarium accessions across northwestern North America from St. Mathew Island, Alaska, USA to Devon Island, Nunavut, CAN, as well as 14 field collections from Barrow, Alaska to Dawson, Yukon, CAN (including those from Chapter 1). Leaf tissue was harvested from herbarium accessions maintained at the University Herbarium in the Museum of Natural History at the University of Colorado Boulder (COLO) and the Herbarium at the Museum of the North at the University of Alaska Fairbanks (ALA). One or two leaves were harvested from all accessions obtained since 1979 that had adequate tissue for flow cytometry. Three individuals, collected prior to 1979 (one as old as 1948) were sampled to test the efficacy of FCM ploidy determination from older samples. Tissue was stored in silica gel until assayed.
Flow cytometry. Methods followed Dolezel et al. (2007) as modified by Mastin et al. (2017). Fluorescence values were compared to that of f. penlandii (2n) to determine ploidy. The protocol used to estimate ploidy from herbarium accession was modified as follows due to the limited amount of tissue available: 1) Solanum lycopersicum CV. Stupicke (2C DNA = 1.96pg) and Glycine max CV. Polanka (2C DNA = 2.50pg) were used as standards; 2) due to overlap of nuclei size, the standards were assayed individually once for every 6-10 samples;
26


3) a higher threshold for coefficient of variation (CV = 10%) was set, due to lower tissue quality, as compared to silica dried samples; and 4) genome size was obtained for only one individual from each herbarium collection, which here, represents a population.
Environmental niche modeling. Environmental data layers for modeling were obtained from Worldclim.org (Hijmans et al. 2005) at 30 arc-seconds and trimmed down to the North American range of E. edwardsii (Latitude 50° - 85°, Longitude "10° - 175°). Principle Components Analysis (PCA) was performed in R (ver. 3.3.1; R Core Team 2016) with 19 bioclimatic variables and altitude to identify the variables with the highest contribution to the variance among samples. The highest contributing variables the first two axes that described 83% of the variation among variables were chosen as candidates. Reliable maximum entropy niche modeling has been shown to be effective with as few as five occurrences for every explanatory variable used (Godsoe et al. 2013). This rule was used as the maximum number of variables for niche modeling.
Niche models were generated in Maxent (ver. 3.3.3k; Phillips et al., 2004, 2006), a machine learning application that builds models by finding the probability distribution of maximum entropy (Phillips et al. 2006). This produces a model with the most uniform distribution of suitability, for a given set of occurrences, across all variables. Maxent niche models were used in ENMTools (ver. 1.2; Warren et al. 2008) to calculate niche breadth, niche overlap (Schoener's 'D', and 'I', Warren et al. 2008), and niche identity in order to test the hypothesis of niche divergence between cytotypes.
Niche breadth is the proportion of environmental conditions that is occupied within the available range of conditions in a given area. Comparisons of niche breadth of multiple taxa
27


using the same geographic range indicate which taxon occurs in a wider range of the available environmental conditions. Niche overlap estimates the proportion of similarity between models, where 0 = no overlap and 1 = identical models. D is a metric that was developed before presence only niche modeling and assumes that the variance of each variable reflects the relative use of that variable; as such, the mean of a variable confers the highest density. /, however, removes that assumption so that the variance reflects probability distributions (Warren et al. 2008). The niche identity test estimates the overlap of one model to a model generated from randomizing the samples of both models. Overlap of the original models was compared to the distribution of overlap from the randomized models. Niche identity test whether two populations were drawn from the same distribution of environmental variables (Warren et al. 2008). Lower identity scores indicate greater divergence between the distribution of variables occupied by each species.
Maxent model parameters that differed from the default were set as follows: random test percentage = 25%; replicates = 15; replicated run type was 'subsample'; max iterations = 5000; random seed = true, add samples to background = true.
Results
Flow cytometry. Of the 79 herbarium accessions sampled for flow cytometry, 47 (59.5%) provided interpretable results. Flow cytometry revealed three cytotypes from among the herbarium and field collected samples of E. edwardsii, including 26 tetraploids, 33 hexaploids, and two octaploids (Table 4). Samples without specific coordinate data that could not be georeferenced, duplicate samples that were considered the same population, and samples with missing environmental data were not included in the niche modeling,
28


leaving 21 tetraploids and 30 hexaploids; due to low sample size, octaploids were not included in Maxent modeling (Godsoe et al. 2013). Fluorescence values for tetraploids were 92,985 - 115,724 (mean= 102,558, sd= 5842) and hexaploids were 130,003 - 179,377 (mean= 155,021, sd= 10,387); CVs ranged from 3.1 to 7.3 (Table 4).
Environmental niche modeling. The first two axes of the PCA describe 83% of the variation. The variables with the highest loading on these components were BIO-2 (mean diurnal range, mean of monthly temperature ranges), BIO-4 (temperature seasonality), BIO-5 (max temperature of the warmest month), BIO-12 (annual precipitation), and BIO-16 (precipitation of the wettest quarter) (Fig. 4). The results of a PCA revealed that hexaploids occupy habitat that has greater diurnal range, greater temperature seasonality, higher maximum temperatures, with moderately higher precipitation maxima, as compared to tetraploids.
Niche models were created in Maxent for tetraploids (n=21) and hexaploids (n=30) (Fig. 5). Area under the curve (AUC) scores from receiver operating characteristic (ROC) curves of Maxent models for tetraploid and hexaploid E. edwardsii were 0.901 ± 0.037 and 0.928 ± 0.065, respectively. This score indicates how well the model predicts the presence of occurrences, such that an AUC of 0.500 would indicate random predictions and AUC = 1.000 would indicate no false positives and no false negatives. BIO-5 (precipitation seasonality) had the highest contribution to the tetraploid model, both for model gain when used by itself and loss-of-gain when omitted. For the hexaploid model, BIO-2 (mean diurnal range) had the highest gain in isolation and the highest loss when omitted from the model.
29


Niche breadth was calculated for each cytotype model revealing higher breadth for tetraploids (niche breadth = 0.521) as compared to hexaploids (niche breadth = 0.159). Maxent models revealed niche overlap of tetraploids and hexaploids (D = 0.354, / = 0.654) that was lower than expected for niche equivalency (Table 5). Niche identity of Maxent tetraploid and hexaploid models were 0.639 and 0.589, respectively. This indicates that the tetraploid model has greater overlap with polyploids combined than does the hexaploid model.
Suitability maps created in Maxent revealed a spatially broad range comprising mostly coastal and low elevation areas for tetraploids and a narrow range of alpine regions for hexaploids (Fig. 5). Suitability is highest for tetraploids across all coastal areas in North America from the Pacific coast across northern Canada to Greenland and Iceland. The hexaploid model shows highest suitability in eastern Alaska and Yukon Territory, and a small area of moderate suitability north of Hudson Bay.
Discussion
Polyploidy is ubiquitous, especially in the arctic where it characterizes up to 80% of plant species (Abbott and Brochmann 2003, Soltis and Soltis 2009), including a large proportion of allopolyploids. The high frequency of polyploids in the arctic suggests that polyploidy may increase ecological tolerance or adaptability in new or changing habitats. The proposed success of polyploids, and allopolyploids specifically, is presumably due to the added genetic material and novel information resulting from genome doubling and hybridization, respectively (Brochmann et al. 2004). The added genetic material thereby increases the
30


phenotypic plasticity of individuals and diversity of populations, facilitating adaptation to
habitats outside the niche of the progenitors.
For synthetic allopolyploids, these changes can occur immediately after formation or after as few as five generations of selfing (Chen and Ni 2006). If gene expression and regulatory changes occur at ecologically important loci, rapid adaptation can result. Pires et al. (2004) showed that the synthetic allopolyploid Brassica napus segregated into early and late flowering lineages after only six generations of selfing. This rapid phenotypic change can result in adaptation to new environments. More specifically, changes in gene expression and regulation due to environmental conditions are expected to promote successful colonization of polyploid populations in novel niches (Chen and Ni 2006). Thus, polyploids would have an advantage in novel niches resulting from environmental change.
Ploidy determination. Although flow cytometry ploidy determination using herbarium tissue has previously been demonstrated to be successful by Roberts (2007) and others, here I show that ploidy can be determined from material that is well over 25 years old. The success rate reported here for herbarium samples was 60% and comprises samples spanning the entire range of collection dates (Table 4). Unsuccessful assays result from nuclei degradation most likely associated with the age of the plant when collected and drying conditions. The use of herbarium accessions for ploidy determination and ecological niche modeling will allow studies to use larger data sets with minimal sampling effort and to allow niche comparisons across time.
Ploidy data for five populations of E. edwardsii suggest no mixo-ploid populations in this sample, suggesting populations are comprised of a single cytotype, and thus populations of
31


cytotypes are assumed to be distinct. Although mixo-ploid populations are common among
polyploid species, discrete populations are expected under minority cytotype exclusion (Levin 1975), such that low frequency cytotypes will be eliminated from populations due to the low frequency of ploidy-matched gametes. Likewise, allopolyploids are expected to have a higher fitness in habitat outside the niche of the lower-ploidy cytotypes and therefore expected to either colonize novel niches to avoid competition with progenitors or outcompete progenitors in the parental population.
These results suggest that tetraploid and hexaploid populations of E. edwardsii occupy divergent niches and add support to the hypothesis that polyploidy facilitates adaptation to new and changing environments based on the following evidence: Niche differentiation was observed between tetraploid and hexaploid populations of E. edwardsii (D = 0.354, / = 0.654); hexaploids occupy a niche that has both higher temperature and precipitation maximums than that of tetraploids (Fig. 4), with a smaller geographic range of suitable habitat (Fig. 5). The difference between niche means supports niche shift occurred for hexaploids, whereas a smaller niche breadth also supports niche contraction. The apparent niche shift may be the result of niche intermediacy relative to the progenitors or niche novelty resulting from the union of previously divergent genomes, providing unique gene products, and thus phenotypes, relative to either progenitor (Soltis et al. 2014). Subsequent evolution can limit the niche through homeolog loss resulting in local adaptation. Alternatively, the disruption of a cytotype may be limited by the available habitat that is within the new niche. That is to say, hexaploids may have a broader fundamental niche, but
32


the occupied niche is smaller than tetraploids due to competition and availability of suitable
habitat.
Niche modeling. A Principle Components Analysis revealed five variables that contribute most to the variation observed among all variables. These variables describe the most variance among environmental variables of the two cytotypes and thus the direction of differentiation hexaploids underwent.
Ecological niche modeling revealed divergence between tetraploid and hexaploid populations of E. edwardsii, with hexaploids occupying a niche that has greater temperature range with higher maxima and more precipitation as compared to tetraploids. These results support the hypothesis that increased ploidy facilitated adaptation to habitats that are outside the range of the progenitors in E. edwardsii.
Ecological niche models provided estimates of overlap between ploidy levels that revealed divergence between cytotypes. Niche breadth for tetraploids is higher than that of hexaploids, which indicates that hexaploids occupy a narrower niche compared to tetraploids, which supports the hypothesis of niche contraction. However, hexaploids may be restricted in their range due to limited availability of habitat that is outside the niche of tetraploids. It may be that the fundamental niche of hexaploids is much broader than the occupied niche depicted in this paper. Presence only niche modeling provides an estimate of a subset of the occupied niche, which is limited to the scope of variables used. The observed niche shift supports the hypothesis of niche divergence between cytotypes.
Niche overlap is expected to decrease with increasing divergence among taxa, such that overlap is higher for more closely related species; thus, high niche overlap would be
33


expected for infraspecific taxa. Results for E. edwardsii indicate less overlap than that from
other studies in which niche differentiation was concluded (e.g., McIntyre 2012). However,
D tends to overestimate overlap because of the amount of equally unsuitable habitat. Niche overlap increases when the models have large areas of habitat that are unsuitable for both cytotypes. Therefore, niche divergence may be greater than indicated by this statistic alone. The / statistic for niche overlap removes the assumption that presences represent the full range of variables occupied by the taxon, which makes this statistic more conservative. Even under this relaxed assumption / = 0.654 suggests divergence between cytotypes.
The identity test provides a more robust estimation of the equivalency of two sets of data (Warren et al. 2008) by treating both cytotypes as one group and randomly sampling to generate many replicate models. The low overlap between replicate models and each original model provides further support that these two cytotypes do not occupy similar niches. Whether the differentiation in niches is caused by local adaptation, novel genetic information from a potential third progenitor to the hexaploids, or a direct consequence of genome duplication cannot be elucidated form these data. Nevertheless, the degree of niche overlap combined with the difference in niche breadth observed suggests that polyploidization facilitated novel niche adaptation in E. edwardsii.
There are inherent limitations associated with presence only niche modeling that have been addressed in the past few years (Sillero 2011, Warren 2012). Most notably is the unknown relationship between the occupied niche (represented by the model) and fundamental niche. It is important to recall that niche models represent the occupied niche that may be restricted due to variables not considered in niche models, such as geography,
34


soil composition, dispersal, and competition. Conversely, the fundamental niche may be
smaller than the model when a species is locally adapted to microclimates or rare niches.
The variables used to generate models may also be questioned as, in many ENM, including this study, variables are not chosen for their influence over biological tolerances, but rather for their contribution to the variance among variables. However, Warren (2012) argued that as long as the assumptions of ENM are understood, meaningful conclusions can still be made. The assumptions he describes are as follows: ENMs are at least some subset of the niche that a taxon occupies, whether or not the variables used are the most influential to the niche; the explanatory power of ENMs is, at worst, a result of spatial autocorrelation among variables, but would still inform comparisons between models. That is, most climatic variables are correlated, so even if the variables used to generate models are not the most influential to the niche, some aspect about those sets of conditions is. Therefore, models do not need to be perfect representations of a species niche to detect divergence.
Based on these assumptions, it can be stated that under any circumstance, it is not possible to calculate the fundamental niche of a species; rather, niche modeling elucidates the range within variables that characterize a niche (Godsoe 2010, Sillero 2011). Likewise, ecological niche models are better understood as a simulation of a species distribution, rather than a description of the niche (Jimenez-Valverde et al. 2008, Sillero 2011). Therefore, we can be confident that differentiation observed between niche models represents real differentiation for at least a subset of a species' niche.
35


Pleistocene glaciations created an environment characterized by large scale habitat destruction and fragmentation. As glaciers receded, habitat was sporadically opened offering habitat void of competition but, likely, less suitable than refugia where Eutrema presumably persisted. These characteristics are ideal for neo-polyploids, which are less fit in refugia due to minority cytotype exclusion and potentially more fit in novel habitat due to increased genomes and/or hybridization. Multiple Eutrema linages existed prior to and possibly throughout polyploidization, which provided enough variation to adapt to a harsh and ever-changing environment. Today, all but one of those lineages are either extinct or not described, which limits our ability to elucidate relationships between progenitor and progeny.
Future Directions. Environmental niche model analysis of North American Eutrema will continue by modeling the niche of the diploid, E. penlandii, and the niche of polyploids (E. edwardsii) combined. Niche models of diploids will be extrapolated to the range of E. edwardsii to determine niche overlap of the two species. This will provide a context in which to evaluate niche evolution in North American Eutrema. However, it may not be possible to determine if divergence between the two species is a result of niche intermediacy or post-polyploidization adaptation, because only one progenitor is described. Therefore, genetic studies need to be performed to better understand the relationship between E. penlandii and E. edwardsii and the degree of divergence between progenitors. It would be helpful to know the timescale for genome duplication and hybridization events as well as the contribution of each genome to the to polyploids, assuming there were multiple origins and possibly independent origins for different cytotypes.
36


Future studies should also examine microclimates, vegetation, and other biotic and abiotic variables to further describe niches and elucidate differentiation in other ecological dimensions. Eutrema grows in specific sites that might share temperature and precipitation conditions with areas not suitable for this species, such as alpine habitats that are not wetlands. Similarly, there may be competition or seed dispersal limitations that are not considered in this analysis. The study of the ecological implications of polyploidy will provide further insight into the evolutionary history of E. edwardsii and broaden our knowledge of the impact polyploidy has had on the arctic flora.
37


Tables and Figures
Table 4. Flow cytometry data for E. edwardsii samples harvested by Leo P. Bruederle or from herbaria at University of Colorado (COLO) and University of Alaska (ALA).
Source / Herbarium Site Identifier Collection Date Latitude Longitude Fluorescence CV Ploidy
LPBlab 702201101 07/02/11 64.1038 -140.8778 167535.02 4.23 6x
LPBlab 705201101 07/05/11 65.3996 -145.9763 173507.45 4.47 6x
LPBlab 710201102 07/10/11 68.1845 -149.4251 174252.62 4.72 6x
LPBlab 710201103 07/10/11 68.5513 -149.4949 161730.30 5.08 6x
LPBlab 711201102 07/11/11 68.6481 -149.4027 167702.18 4.16 6x
LPBlab 711201103 07/11/11 68.3032 -149.3595 170165.70 4.45 6x
LPBlab 717201002 07/17/10 64.6657 -138.3914 167189.04 4.41 6x
LPBlab 721201001 07/21/10 71.2641 -156.8296 104758.49 4.66 4x
LPBlab 721201002 07/21/10 71.2516 -156.5060 109264.12 4.57 4x
LPBlab 722201001 07/22/10 71.2392 -156.3358 107005.55 4.56 4x
LPBlab 0704201101a 07/04/11 63.0727 -145.7279 171085.47 4.73 6x
LPBlab 0706201101a 07/06/11 65.4849 -145.4040 169789.97 4.17 6x
LPBlab 0717201001b 07/17/10 64.9517 -138.2713 146708.01 4.41 6x
LPBlab Campbell Creek N/A 61.0688 -149.6056 111558.98 4.64 4x
COLO COLO 346724 07/10/80 70.3667 -148.5333 97587.37 6.49 4x
COLO COLO 419537 07/13/84 70.0167 -148.7500 97782.74 4.95 4x
COLO COLO 419673 07/25/84 70.3000 -149.2000 99092.80 4.95 4x
ALA V071453 08/04/79 70.2833 -148.5000 206863.83 5.57 8x
ALA V117025 06/13/94 65.6333 -146.7556 155475.61 4.92 6x
ALA VI18847 06/28/94 61.2472 -149.5875 109699.34 5.00 4x
ALA V119651 06/14/95 65.6167 -147.3667 148780.81 6.57 6x
ALA V119938 07/23/95 64.4270 -173.2189 99559.50 5.19 4x
ALA V121861 07/03/96 65.1167 -144.0500 135727.82 4.92 6x
ALA V122304 06/18/95 65.8389 -145.8833 147233.24 4.88 6x
ALA V122877 07/21/97 68.6000 -156.5000 95988.25 4.88 4x
ALA V123470 07/15/97 63.6667 -160.7000 95546.51 4.98 4x
ALA V127658 07/14/99 75.0667 -98.5000 104675.67 5.37 4x
ALA V129584 07/26/99 69.7667 -122.0833 185531.21 6.54 6x
ALA V131261 08/22/98 63.8278 -146.8492 157154.37 4.91 6x
ALA V133443 08/23/99 76.5191 -86.7662 103031.74 4.86 4x
ALA V133475 08/26/99 74.5333 -82.7833 115762.52 3.91 4x
ALA V134036 06/26/01 67.2700 -163.6700 106758.54 4.90 4x
ALA V134203 07/01/01 67.1133 -163.3633 96527.73 4.92 4x
38


Table 4. Cont.
Source / Herbarium Site Identifier Collection Date Latitude Longitude Fluorescence CV Ploidy
ALA V137839 07/05/02 68.5367 -161.4664 100766.76 4.77 4x
ALA V139646 07/28/02 68.2394 -152.1944 97054.02 4.86 4x
ALA V139812 07/19/02 68.1950 -152.7444 158004.68 4.28 6x
ALA V139958 07/20/02 68.1900 -152.7400 147925.88 8.46 6x
ALA V140421 06/21/01 68.6110 -149.6437 158840.05 6.25 6x
ALA V140748 07/29/02 68.5644 -152.5344 153299.58 4.77 6x
ALA V141037 07/27/02 68.3333 -151.0333 173159.74 9.00 6x
ALA V141828 07/11/02 64.6250 -165.6750 101525.72 4.83 4x
ALA V146945 07/07/02 64.5398 -143.7021 164457.88 4.87 6x
ALA V151202 07/19/03 61.3163 -142.2703 152331.41 4.86 6x
ALA V153299 06/08/03 62.4000 -141.9500 152403.85 5.21 6x
ALA V158770 07/18/97 60.6001 -172.9300 100831.20 4.32 4x
ALA V158827 07/22/97 60.5500 -172.9167 103520.74 4.21 4x
ALA V160941 06/11/07 66.8000 -141.0667 166189.84 4.91 6x
ALA V174587 06/22/09 64.5240 -143.5907 148235.13 4.84 6x
ALA V79106 06/28/84 63.8000 -148.9333 144776.87 7.80 6x
ALA V82602 08/01/85 68.6167 -149.3167 148647.32 4.75 6x
ALA V83059 07/07/84 70.0000 -144.4333 108508.59 8.76 4x
ALA V95009 08/13/87 69.1167 -105.0833 205937.14 6.40 8x
ALA V97542 06/08/88 64.5728 -163.7542 100373.33 7.58 4x
39


Hexaploid Tetraploid
PC1 (55.5% explained var.)
Fig. 4. PCA plot showing the loading of Bioclim 19 variables and altitude (downloaded from worldclim.org) for the first two components, PCI and PC2, which comprise 83% of the variation among variables. Tetraploid niche space is outline in blue and hexaploid niche space is outlined in red.
40


Fig. 5. Maxent niche models of North American Eutrema edwardsii R. Br. (Brassicaceae) for tetraploid (n = 21; top) and hexaploid (n = 30; bottom) populations. Maps show potential suitability, based on input variables of temperature and precipitation, ranging from 0 (blue) to 1 (red).
41


Table 5. Modeling statistics from ENMTools. Overlap is a symmetrical calculation, whereas Identity is calculated as original cytotype model compared to models from a combined distribution of both cytotypes.
Cytotype Niche Breadth Overlap (D) Overlap (1) Identity
Tetraploid 0.521 0.354 0.654 0.639
Hexaploid 0.159 0.354 0.654 0.589
42


REFERENCES
Abbott RJ and Brochmann C. 2003. History and evolution of the arctic flora: in the footsteps of Eric Hulten. Molecular Ecology. 12: 299-313.
Al-Shehbaz IA and S.l. Warwick. 2005. A synopsis of Eutrema (Brassicaceae). Harvard Papers in Botany 10: 129-135.
Chen ZJ and Ni Z. 2006. Mechanisms of genomic rearrangements and gene expression changes in plant polyploids. BioEssays. 28: 240-252.
Baker RL, Yarkhunova Y, Vidal K, Ewers BE, Weinig C. 2017. Polyploidy and the relationship between leaf structure and function: implications for correlated evolution of anatomy, morphology and physiology in Brassica. BMC Plant Biology. 17(3) DOI 10.1186/sl2870-016-0957-3
Bickford D, Lohman DL, Sodhi NS, Ng PKL, Meier R, Winker K, Ingram KK, Das I. 2006. Cryptic species as a window on diversity and conservation. Trends Ecol. Evol. 22: 148-155.
Bocher TW and K. Larsen. 1950. Chromosome numbers of some arctic or boreal flowering plants. Medd. Grpnl. 1947(6).
Brochmann C, Soltis PS, Soltis DE. 1992. Recurrent formation and polyphyly of Nordic polyploids in Draba (Brassicaceae). American Journal of Botany. 79(6): 673-688.
Brochmann C, Brysting AK, Alsos IG, Borgen L, Grundt HH. 2004. Biological relevance of
polyploidy: ecology to genomics Polyploidy in arctic plants. Society 82(4): 521-536.
Dolezel J, Greilhuber J, Suda J. 2007. Estimation of Nuclear DNA content in Plants Using Flow Cytometry. Nature Protocols. 2(9): 2233-2244.
Fayette K and Bruederle LP. 2001. Morphometric analyses support specific status for Eutrema penlandii Rollins (Brassicaceae). Botany Abstract
Grundt HH, Obermayer R, Borgen L. 2005. Ploidal levels in the arctic-alpine polyploidy Draba lacteal (Brassicaceae) and its low-ploid relatives. Bot. J. Linn. Soc. 147: 333-347.
Grundt HH, Kjplner S, Borgen L, Rieseberg LH, Brochmann C. 2006. High biological species diversity in the arctic flora. Proceedings of the National Academy of Sciences 103(4): 927-975.
Glennon KL, Ritchie ME, Segraves KA. 2014. Evidence for shared broad-scale climatic niche of diploid and polyploidy plants. Ecology Letters. 17: 574-582.
43


Godsoe W, Larson MA, Glennon KL and Segraves KA. 2013. Polyploidization in Heuchera
cylindrca (Saxifragaceae) did not result in a shift in climatic requirements. American Journal of Botany. 100(3): 496-508.
Hardwick RC. 1997. Population Genetics of the Rare Alpine Endemic Eutrema penlandii Rollins (Brassicaceae). Master's Thesis. University of Colorado Denver.
Harbert RS, Brown AHD, Doyle JJ. 2014. Climate niche modeling in the perennial Glycine
(Leguminosae) allopolyploidy complex. American Journal of Botany. 104(4): 710-721.
Husband BC and Schemske DW. 1998. Cytotype distribution at a diploid-tetraploid contact zone in Chamerion (Epilobium) angustifolium (Onagraceae). American Journal of Botany. 85(12): 1688-1694.
Jprgensen CA, Sprensen T, Westergaard M. 1958. The flowering plants of Greenland: A taxonomical and cytological survey. Kongel. Danske Vidensk. Selsk. Biol. Skr. 9: 1-172.
Jimenez-Valverde A, Gomez JF, Lobo JM, Baselga A, Hortal J. 2008. Challenging species distribution models: the case of Maculinea nausithous in the Iberian Peninsula. Annales Zoologici Fennici. 45(3): 200-210.
Knaben G. 1968. Chromosome numbers of flowering plants from central Alaska USA polyploidy. Nytt Magasin for Botanikk (Oslo) 15(3): 240-254.
Kron P and Husband BC. 2012. Using flow cytometery to estimate pollen DNA content: Improved methodologies and applications. Annals of Botany. 110: 1067-1078.
Levin DA. 1975. Minority cytotype exclusion in local plant populations. Taxon, 35-43.
Levin DA. 2000. The origin, expansion, and demise of plant species. Oxford University Press.
Mason AS and Pires JC. 2015. Unreduced gametes: meiotic mishap or evolutionary history?. Trends in Genetics. 31(1): 5-10.
McAllister C, Blaine R, Kron P, Bennett B, Garrett H, Kidson J, Miller AJ. 2015. Environmental correlates of cytotype distribution in Andropogon gerardii (Poaceae). American journal of botany. 102(1): 92-102.
McIntyre PJ. 2012. Polyploidy associated with altered and broader ecological niches in the Claytonia perfoliata (Portulacaceae) species complex. American Journal of Botany. 99(4): 655-662
44


Miller M, Zhang C, Chen ZJ. 2012. Ploidy and Hybridity Effects on Growth Vigor and Gene Expression in Arbidopsis thaliana Hybrids and Their Parents. G3: Genes, Genomes, Genetics. 2: 505-513.
Murray DF. 1981. The role of arctic refugia in the evolution of the arctic vascular flora- a
Beringian perspective, pp. 11-20. In G. G. E. Scudder and J. L. Reveal, eds. Evolution today, proceedings of ICSEB II. Hunt Institute for Botanical Documentation, Pittsburgh.
Otto SP and Whitton J. 2000. Polyploid incidence and evolution. Annu. Rev. Genet. 34: 401-437.
Paun O, Forest F, Fay ME, Chase MW. 2006. Hybrid speciation in angiosperms: parental divergence drives ploidy. New Phytol. 182: 507-518.
Paun O, Forest F, Fay MF, Chase MW. 2009. Hybrid speciation in angiosperms: parental divergence drives ploidy. New Phytologist. 182: 507-518.
Pires JC, Zhao J, Schranz ME, Leo EJ, Quijada PA, Lukens LN and Osborn TC. 2004. Flowering time divergence and genomic rearrangements in resynthesized Brassica polyploids (Brassicaceae). Biological Journal of the Linnean Society. 82: 675-688.
R Core Team (2016). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/.
Ramsey J and Schemske DW. 1998. Pathways, mechanisms, and rates of polyploidy formation in flowering plants. Annual Rev. Ecol. Syst. 29: 467-501.
Roberts AV. 2007. The Use of Bead Beating to Prepare Suspensions of Nuclei for Flow
Cytometry from Fresh Leaves, Herbarium Leaves, Petals and Pollen. Cytometry Part A. 71A:1039-1044.
Sillero N. 2011. What does ecological modeling model? A proposed classification of
ecological niche models based on their underlying methods. Ecological Modelling. 222: 1343-1346.
Sokolovskaja AP and Strelkova OS. 1941. Polyploidy and karyological races under conditions in the Arctic. C.R. (Doklady) Acad. Sci. .U.S.S.R. 32: 144.
Soltis, DE, Haufler CH, Darrow DC, Gastony GJ. 1983. Starch-gel electrophoresis of ferns - A compilation of grinding buffers, gel and electrode buffers, and staining schedules. Amer. Fern J.73: 9-27.
45


Soltis DE, Visger CJ, Soltis PS. 2014. The polyploidy revolution then...and now: Stebbins revisited. American Journal of Botany. 101(7): 1057-1078.
Soltis PS and Soltis DE. 2000. The role of genetic and genomic attributes in the success of polyploids. Proceedings of the National Academy of Sciences 97(13): 7051-7057.
Soltis DE, Soltis PE, Schemske DW, Hancock JF, Thompson JN, Husband BC, Judd WS. 2007. Autopolyploidy in Angiosperms: Have we grossly underestimated the number of species. Taxon 56(1): 13-30.
Soltis PS and Soltis DE. 2009. The role of genetic and genomic attributes in the success of polyploids. PNAS. 97(13): 7051-7057.
Sora D, Kron P, Husband BC. 2016. Genetic and environmental determinants of unreduced gamete production in Brassica napus, Sinapis arvensis and their hybrids. Heredity. 117: 440-448.
Theodoridis S, Randin C, Broennimann O, Patsiou T, Conti E. 2013. Divergent and narrower climatic niche characterize polyploidy species of European primroses in Primula sect. Aleuritia. Journal of Biogeography. 40: 1278-1289
Thompson KA, Husband BC and Maherali H. 2014. Climatic niche differences between diploid and tetraploid cytotypes of Chamerion angustifolium (Onagraceae).
American Journal of Botany. 101(11): 1868-1875 .
Van Dijk P and Bakx-Schotman T. 1997. Chloroplast DNA phylogeography and cytotype geography in autopolyploid Plantago media. Molecular Ecology. 6: 345-352.
Wallace LE. 2003. Molecular Evidence for Allopolyploid Speciation and Recurrent Origins in Platanthera huronensis (Orchidaceae). International Journal of Plant Science. 164(6): 907-916.
Warren DL, Glor RE, Turelli M. 2008. Environmental niche equivalency versus conservatism: quatitative approaches to niche evolution. Evolution. 62(11): 2868-2883.
Warren DL. 2012. In defense of 'niche modeling'. Trends in Ecology and Evolution. 27(9): 497-500.
Warwick, S.I., Al-Shehbaz, I.A., Sauder, C.A. 2006. Phylogenetic position of Arabis arenicola and generic limits of Aphragmus and Eutrema (Brassicaceae) based on sequences of nuclear ribosomal DNA. Canadian Journal of Botany 84: 269-281.Table 1. Locations for Eutrema penlandii Rollins and E. edwardsii R. Br. (Brassicaceae) populations sampled for chromosome counts1, flow cytometry2, and allozyme analysis3.
46


Weber WA. 1985. New names and combinations, principally in the Rocky Mountain USA Flora V. Phytologia 58: 382-384.
Weber WA. 2003. The Middle Asian Element in the Southern Rocky Mountain flora of the western United States: a critical biogeographical review. J. Biogeogr. 30: 649-685.
Weber WA, Wittmann RC. 2012. Colorado Flora: Eastern Slope. Afield guide to the vascular plants. Fourth Ed. University Press of Colorado. Boulder, CO.
Wyatt R, Stoneburner A, Odrzykoski I J. 1989. Bryophyte isozymes: systematics and
evolutionary implications. In: Soltis DE, Soltis PS (eds) Isozymes in Plant Biology, Dioscorides Press, Portland: 221-240.
47


48


Full Text

PAGE 1

EVOLUTIONARY AND ECOLOGICAL IMPLICATIONS OF POLYPLOIDY IN EUTREMA EDWARDSII (BRASSICACEAE) B y JARED EDMOND MASTIN B.S., University of Colorado Denver, 2015 A thesis submitted to the Faculty of the Graduate School of the University of Colorado in partial fulfilment of the requirements for the degree of Master of Science Biology Program 2017

PAGE 2

ii This thesis for the Master of Science degree by Jared Edmond Mastin has been approved for the Integrative Biology Program by Leo P Bruederle, Chair Rebecca Hufft Michael Wunder Date: May 1 3, 2017

PAGE 3

iii Mastin, Jared Edmond (M.S., Biology Program) Evolutionary and E cological I mplications of P olyploidy in Eutrema edwardsii (Brassicaceae) Thesis directed by Associate Professor Leo P Breuderle ABSTRACT Polyploidy is a ubiquitous phenomenon that is most frequent in the arctic where glaciation cycles put selective pressures on populations by repeated fragmentation. Several decades of research suggest that polyploids would have been more fit than d iploids in novel habitats as glaciers receded. The higher fitness of polyploids has been attributed to increased genetic material and novel gene products, which result in phenotypic plasticity and rapid adaptation. These concepts lead to the question; Does polyploidy increase the ecological tolerances of a species thereby facilitating colonization outside the niche of the progenitors? To answer this question, a polyploid complex was investigated to address three objectives: 1) determine the polyploid mode o f origin; 2) document the distribution polyploid cytotypes; and 3) discover the relationship between ploidy and niche breadth. Eutrema edwardsii R. Br. ( Brassicaceae) is an arctic alpine mustard with a near circumpolar distribution. Its closest relative, E utrema penlandii Rollins, is a federally listed, threatened species that is endemic to the Mosquito Range in the Southern Rocky Mountains of Colorado . Together these species comprise a polyploid complex in North America, for which we conducted chromosome c ounts, flow cytometry, and allozyme analysis to elucidate the polyploid origins of E. edwardsii and model the niche of polyploid cytotypes to discover environmental correlates with respect to ploidy.

PAGE 4

iv Results obtained from mitotic counts of two populations of E. penlandii reveal this taxon to be diploid . Diploidy was confirmed using flow cytometry for an additional 15 individuals representing four populations . Previously published chromosome counts for E. edwardsii reveal a polyploid complex of tetraploid, hexaploid, and octaploid populations for which an autopolyploid origin has been presumed . However, allozyme analysis revealed an allopolyploid origin for E. edwardsii, as evidenced from fixed heterozygosity at six loci . Flow cytometry revealed all three ex pected cytotypes among 53 populations. Niche models were created for tetraploid and hexaploid populations to calculate niche overlap ( D = 0.354, I = 0.654), which revealed divergence. The data reported herein support the recognition of E. penlandii as tax onomically distinct, reveal cryptic variation within E. edwardsii , and support niche divergence between tetraploids and hexaploids. The form and content of this abstract are approved. I recommend its publication. Approved: Leo P Bruederle

PAGE 5

v ACKNOWLEDGEMENTS I would like to acknowledge my advisor, Leo P. Bruederle, Rebecca Hufft (Associate Director of Applied Conservation, Denver Botanic Gardens) and Michael Wunder (Associate Professor, University of Colorado Denver) for serving on my thesis committee and for helpful conversations and advice about flow cytometry, data analysis and niche modeling . I also ac knowledge : Robert Laport for our many meetings about niche modeling and flow cytometry; James Salmen (Director of F acilities and Laboratory Manager, University of Colorado Denver) , who provided technical support with respect to flow cytometry; Eileen Yakish, Hannah Tystad, and Daniel Harper, then undergraduates at CU Denver who collected preliminary allozyme data for t his study; and Suzanne Meinig, who collected flow cytometry data and georeferenced herbarium samples . The research was supported by a contract awarded to LPB by the U.S. Fish &Wildlife Service (F10AC00603) .

PAGE 6

vi TABLE OF CONTENTS CHAPTER I. POLYPLOID ORIG INS OF EUTREMA EDWARDSII Materials and Methods ................................................................................................2 Results................ ........................................................................................................ ..7 Discussion .................................................................................................................. .. 9 Figures and Ta bles ..................................................................................................... 14 II. ECOLOGICAL NICHE MODELING REVEALS DIVERGENCE BETWEEN TETRAPLOID Introduction.................................................................................. ..............................21 Materials and Methods................................................................. ............................. 26 Results ........................................................................................... ..............................28 Tables and Figures....................................................... ................. ............................. 38 REFERENCES

PAGE 7

1 CHAPTER I ALLOPOLYPLOID ORIGINS OF EUTREMA EDWARDSII Introduction Although flowering plants are under represented in the literature addressing cryptic species (Bickford et al. 2006), there is a growing body of research addressing these cryptic species in the flora of the arctic in particular (e.g., Grundt et al. 2006) . T he underlying explanations proposed to explain this phenomenon include recent speciation, incomplete divergence, and reticulate evolution, with selection, inbreeding, and genetic drift proposed as mechanisms leading to cryptic variation . It is also clear t hat polyploidy particularly autopolyploidy in plants is an important and overlooked mechanism leading to the formation of cryptic species (Soltis et al. 2007) . However, there has been a resistance to recognize multiple cytotypes as distinct even when they confer reproductive isolation as well as a reliance upon the morphological species concept, or some variant thereof (Soltis et al. 2007) . Herein, we report the results of multidisciplinary research addressing the systematics of the Eutrema edwardsi i R. Br. species complex (Brassicaceae), with a goal of contributing to a better understanding of E. penlandii Rollins, a federally listed Colorado endemic. Until 1985, two Eutrema species were recognized from North America: E. edwardsii and E. penlandii , when Weber (1985) published the combination E. edwardsii R. Br. subsp. penlandii (Rollins) W.A. Weber, thereby reducing the rank of this taxon to subspecies . More recently, E. penlandii was subsumed into E. edwardsii , as part of a larger revision to the ge nus following a critical evaluation of Neomartinella Pilger , Platycraspedum O.E. Schulz ,

PAGE 8

2 Taphrospermum C.A . Mey. , and Thellungiella O.E. Schulz ; this resulted in the aforementioned genera being subsumed into Eutrema (Al Shehbaz and Warwick 2005, Warwick et al. 2006) . Therein, it was suggested that the variation exhibited by E. edwardsii could easily accommodate E. penlandii . As part of a larger study addressing conservation genetics in E. penlandii (Hardwick 1997), fixed heterozygosity was observed in populations of E. edwardsii suggesting to us an allopolyploid origin for the latter. Although it has long been assumed that the aforementioned two species form a polyploid complex, for w hich we presupp osed autopolyploidy for E. edwardsii (e.g., Jørgensen et al. 1958), no chromosome counts have been reported for E. penlandii , in particular . Herein, we take multiple approaches using chromosome counts, flow cytometry, and allozyme analysis to better unders tand the systematics of E. edwardsii s.l. , and to test the hypothesis that E. penlandii and E. edwardsii form an autopolyploid complex . Whereas chromosome counts and flow cytometry are used to document ploidy for E. penlandii , allozyme analysis is used to elucidate the polyploid origin of E. edwardsii . Materials and Methods Natural history . Eutrema edwardsii and E. penlandii are perennial herbs that occupy a variety of imperfectly drained arctic and alpine tundra habitats, such as meadows, margins of ponds and stream banks, and solifluction slopes (Aiken et al. 2011) . Eutrema edwardsii is widespread, with a near circumpolar distribution from the Chukotka in Asia across North America to Norway, where it is common in Svalbard (Fig. 1) . Eutrema edwardsii also occurs in the alpine of the White Mountains, Alaska Range, Chugach Mountains, Wrangell St. Elias Mountains, and the Mackenzie Mountains of the northern Rocky Mountains in western

PAGE 9

3 North America and in the Altai Mountains, Pamir Mountains, and Himalaya s in Asia . In contrast, E. penlandii is a narrow endemic that is restricted in distribution to the Mosquito Range in the Southern Rocky Mountains of Colorado, where typically it occupies alpine peatlands overlying calcareous substrates . Relatively little i s known about the natural history (e.g., reproductive biology) of either species . Although chromosome counts conducted for E. edwardsii have revealed tetraploid (4x=28), hexaploid (6x=42), and octaploid (8x=56) populations, where n = 7 (Table 1), no counts have been reported for E. penlandii (Warwick and Al Shebhaz 2006). Field Collections . Plant material, including leaf tissue for allozyme analysis and flow cytometry and seeds for chromosome counts, was collected between 1995 and 2014 . As part of a larger study using allozyme analysis to characterize population genetic diversity and structure in E. penlandii , Hardwick (1997) collected leaf tissue from seven populations of the narrowly distributed Colorado endemic and three populations of its widespread cong ener E. edwardsii from Alaska (Table 2) . L Bruederle subsequently collected leaf tissue from an additional five populations of E. penlandii and seven populations of E. edwardsii from Alaska and Yukon Territory; K. Marr (Royal BC Museum) provided material f rom a Siberian population of E. edwardsii . Approval for tissue collection from E. penlandii was granted by the U.S. Fish and Wildlife Service under the authority of permits PRT 704930 and TE12513A 0. Chromosome counts . Counts for this research were obtaine d by N. Luebke (Milwaukee Public Museum) from squashes of root tips harvested from seedlings germinated specifically for this purpose . In the process of collecting tissue for population genetic and

PAGE 10

4 molecular systematics research, a small number of fruits w ere harvested . Seeds were germinated in 0.005 M giberellic acid in a Percival environmental chamber set at 15°C; attempts were made to cultivate those seedlings that were not harvested in a research greenhouse for subsequent use for these purposes . As suc h, either whole seedlings or root tips were harvested within 2 4 hours after light conditions were restored and placed in a saturated solution of paradichlorobenzene for two hours, after which they were fixed in . Squashes were prepared by macerating the root tip in a drop of 1% aceto orcein on a microscope slide cleaned with 95% ethanol, heating for two mins at 50° C, and squashing . Squashes were scanned at 100x and 400x for meiotic figures and photographed at 10 00x . The chromosome count for E. penlandii was obtained from seed collected at the Treasurevault Mountain site, for which a voucher has been deposited with the Katherine Kalmbach Herbarium (KHD) at the Denver Botanic Gardens . Flow cytometry . Relative genome size was determined from silica dried samples to confirm ploidy for E. penlandii and E. edwardsii based upon chromosome counts, and to document ploidy at sites across the range of E. penlandii . Flow cytometry methods followed (2007). Genome size was obtained for three individuals from each population, depending upon the amount of tissue that was available . Standards were grown in the research greenhouse at the University of Colorado Denver, with leaf tissue harvested and silic a gel dried for subsequent use . Eutrema species to ensure no overlap in scatter plots and histograms generated by flow cytometer (FCM)

PAGE 11

5 software. The standards were included as part of each assay in order to estimate ploidy for the sample peak and to adjust sample values based on differences in standard value, which fluctuated among assays. The fluorescence value for the sample was multiplied by the ratio of standard fluorescence value to the average value for the standard across all assays . Between 1 2cm 2 of leaf tissue was minced using a razor blade along with an internal standard in 1ml of LB01 buffer solution pH 7.5 for 1 2 minutes . The homogenate was mixed using a pipettor and then incubated on ice fo r 15 minutes . The solution comprising nuclei in al. 2007) . RNase mg/l) was added to 0.5 ml of the suspension and allowed to incubate on ice for 15 minute s . suspension, which was gently mixed and incubated on ice for 3 5 minutes in the dark . The approximately 100,000 events were recorded. The nuclei suspension, without the addition of PI, could be kept in a refrigerator overnight without a decrease in the quality of flow cytometry output. Clusters of G1 nuclei outside the region of the standard were gated to create a histogram and calculate mean peak value and coefficient of variation ( CV ) . Assays with a CV over 5% were discarded . Five consecutive measurements were obtained to collect roughly 500 1000 events in the gated region . The average of the five mean peak values was recorded for each sample . Histograms were created that revealed the relative difference between the sample and the standard.

PAGE 12

6 Ploidy was estimated recognizing that E. penlandii is diploid (2n), as documented through the aforementioned chromosome count, and comparing the mean fluorescence of Allozyme analysis . Soluble enzymatic proteins were extracted from two fresh leaves gro und in 0 .25 ml of a cold Tris HCl grinding buffer modified by the addition of 4% w/v PVP 40 and 1% w/v 2 mercaptoethanol (Soltis et al. 1983); a small amount of sea sand facilitated grinding . Extra cts were adsorbed onto 1.5 x 11 mm wicks cut from No. 17 What man chromatography paper and stored at 80°C until electrophoresis. Protein extracts were applied to each of three different starch gel and electrode buffer systems (Nielsen and Johansen 1986): lithium borate, gel buffer pH 8.3/electrode buffer pH 8.1; his tidine HCl tris, gel buffer pH 7.5/electrode buffer pH 7.5; tris citrate, gel buffer pH 7.8/electrode buffer pH 8.7 . Electrophoresis was conducted at 4°C until a bromophenol blue marke r had migrated approximately 12 cm. Following electrophoresis, gels were sliced horizontally in to approximately 1.5 mm thick slices, with interior slices incubated in substrate specific stains (Soltis et al. 1983) previously revealed to be informative with respect to the polyploid origin of E. edwardsii (Bruederle unpub. data) . The lithium borate gel was stained for phosphoglucoisomerase (PGI 2, specifically); superoxide dismutase (SOD 1); and triose phosphate isomerase (TPI 1); the histidine HCL gel was stained for malate dehydrogenase (MDH 1); and the tris citrate gel was stain ed for aspartate aminotransferase (AAT 1) . Enzyme nomenclature follows the International Union of Biochemistry (1984).

PAGE 13

7 Designation of loci and alleles was based on relative mobility of proteins following Hardwick (1997), with the most rapidly migrating loc i receiving sequentially higher number and the alleles at a locus similarly assigned sequential higher lowercase letters . No formal genetic analyses (e.g., controlled crosses) were performed to document patterns of inheritance . Putative genetic loci and ge notypes were inferred from the known substructure and intracellular compartmentalization of enzymes, as well as electrophoretic patterns observed in individuals presumed heterozygous at polymorphic loci . For each population, data were collected as individu al genotypes from which ploidy was extrapolated based upon dosage with support provided using flow cytometry with the same population samples . Deviations from Hardy Weinberg expectations assuming disomic inheritance were tested using Chi square analysis . Results Chromosome counts, DNA content, and ploidy . Chromosome counts reported here from the Treasurevault Mountain site revealed E. penlandii to be diploid, with 2x = 14 and n = 7 (Fig. 2) . Fluorescence during flow cytometry ranged from 48,860 53,024, with a mean fluorescence of 49,907 for E. penlandii (Table 3). By contrast, fluorescence for E. edwardsii populations ranged from 102,601 180,523, suggesting that our sample included both tetr aploid and hexaploid individuals (Fig. 3) . Whereas three populations from the Arctic Coastal Plain in northern Alaska and the Chugach Range in southern Alaska were tetraploid, those populations sampled from Central Alask a and Yukon were hexaploid . Coeffici ent of variation for these measurements were between 2 5%; typically, 500 2000 nuclei were detected in each run.

PAGE 14

8 Although field observations, such as relatively tall habit and large leaf morphology observed in some individuals, suggested infrequent autopo lyploidy in some populations of E. penlandii , this was not substantiated by flow cytometry all individuals of E. penlandii sampled here were found to be diploid . In contrast, E. edwardsii was exclusively polyploid, as expected based upon counts previousl y reported in the literature; populations sampled here are either tetraploid or hexaploid, with no evid ence of diploidy . Although octa ploids have been reported in the literature, none were observed as part of this study. Comparisons between populations of E. penlandii and E. edwardsii provided evidence supporting polyploidy in the latter . Eutrema edwardsii populations sampled for this study were tetraploid or, more commonly hexaploid . Furthermore, relative genome size in hexaploid E. edwardsii was greater t han three times that of E. penlandii, which could be the result of genome modification after polyploidization . Or, this could be the result of another, unknown, cryptic taxon with a slightly larger genome being one of the other progenitors to E. edwardsii . Allozyme analysis . Six loci exhibited disomic inheritance and fixed heterozygosity in one or more of the E. edwardsii populations: AAT 2, MDH 1, PGD 3, PGI 2, SOD 1, and TPI 1 . Chi squared analyses revealed that genotypic frequencies at all six loci devia ted from expectations of independent assortment of chromosomes (p < 0.05), refuting the hypothesis of an autopolyploid origin for E. edwardsii . Seventeen alleles were observed at the six polymorphic loci in E. edwardsii four alleles comprising TPI 1 and MDH 1, three alleles at AAT 1, and two alleles each for the other loci . Eutrema penlandii shared nine of the alleles found in E. edwardsii , suggesting a

PAGE 15

9 progenitor derivative relationship for this species . For example, whereas E. penlandii has the b, c, and d alleles, tetraploid E. edwardsii alleles (i.e., bbcc), with all the presumably contributed by a third progenitor. W e also found evidence for multiple origins of E. edwardsii based upon variability for dosage at loci that exhibited fixed heterozygosity . Hexaploid populations of E. edwardsii exhibited as many as eight different heterozygous genotypes per locus (i.e., TPI 1), suggesting that the progenitors to this taxon were polymorphic at these loci. Although these data are not unequivocal, E. penlandii or a close relative appears to have contributed one genome to tetraploid and to hexaploid E. edwardsii . Furthermore, th e allozyme data suggest that one or more extinct or yet to be described diploid Eutrema s.l. species contributed the remaining genome(s) through multiple origins from polymorphic populations . Discussion Here, I present genetic evidence documenting the all opolyploid origin of tetraploid and hexaploid populations of E. edwardsii involving two or more different progenitors . In tetraploids having an allopolyploid origin, hybridization between two individuals with divergent genomes is followed by chromosome dou bling resulting in polyploid progeny characterized by two complete genomes (Soltis and Soltis 2000) . In contrast to autopolyploidy, where chromosomes segregate and pair independently during meiosis (polysomic inheritance), chromosomes are inherited disomic ally with allopolyploidy; that is, chromosomes of the same genome pair during meiosis, which typically results in fixed -

PAGE 16

10 heterozygosity at loci that were divergent in the parental taxa . Although these represent the extreme forms of polyploidy, the six tetra ploid and hexaploid populations of E. edwardsii studied here exhibited a pattern of disomic inheritance of chromosomes at each of six loci, four of which exhibited fixed heterozygosity. Polyploid species may result from multiple events of polyploidization involving either the same parental progenitors (polytropic), or different progenitors (polyphyletic) (Brochmann et al. 2004) . Polyploids involving multiple progenitors can be difficult to discern; like wise, polyploidy resulting from a single event that wi th subsequent recombination, gene flow between polyploids or diploids, or mutations can cause an overestimate of the number of events of polylploidization (Wallace 2003) . Although our data reveal multiple origins for E. edwardsii, they also suggest that te traploids and hexaploids arose from hybridization events involving the same two or three progenitors . These genomic combinations can have effects on closely related taxa resulting in species misidentification . Hidden variation in morphologically cryptic sp ecies can lead to false conclusions of relatedness when a species may have greater variation than observed, indicating divergence . Recently diverged species that have undergone polyploidization may appear to be autopolyploid; however, disomic inheritance s till occurs (Brochmann et al. 2004). We also provide evidence suggesting that E. penlandii is the closest living descendant of one of the diploid progenitors of E. edwardsii (Paun et al. 2009) . Although we are unable to identify the closest living descenda nt(s) of the other progenitor(s), attempts at genome reconstruction coupled with the number of loci at which fixed heterozygosity was observed

PAGE 17

11 suggests that they were genetically divergent representatives of Eutrema s.l . Given the morphological similarity of E. penlandii and E. edwardsii , it seems likely that the progenitors were morphologically cryptic, despite genetic differentiation, as has been demonstrated in other genera, such as Draba (Grundt et al. 2005). Given the genetic differentiation that we see among the progenitors, it seems likely that hybridization resulted in increased unreduced gamete formation in the hybrids, with whole genome duplication resulting from the fusion of these unreduced gametes (P aun et al. 2009). Different factors have been correlated with unreduced gamete formation (Ramsey and Schemske 1998; Mason and Pires 2015), several of which may have contributed to the formation of E. edwardsii (e.g., hybridization, stress) . Polyploidy is extremely c ommon in the flora of the arctic, where fluctuations in climate have resulted in cycles of fragmentation and range expansion (Brochmann et al. 2004) . Genetic drift, exacerbated by selfing would have resulted in population differentiation, which has been co rrelated with an increased ability to colonize new sites associated with deglaciation (Brochmann et al. 2004) . As previously discussed, taxonomic treatments of E. penlandii differ markedly, with the recent treatments of Al Shehbaz and Warwick (2005) and W eber and Wittmann (2012) subsuming E. penlandii into E. edwarsdsii . However, genetic data presented herein reveal that the taxonomic boundaries between E. edwardsii and E. penlandii are obscured due to a progenitor derivative relationship involving allopol yploidy and morphologically cryptic progenitors.

PAGE 18

12 Eutrema penlandii is a relictual Arcto Tertiary element in the Colorado flora (Weber 2003), with the greatest species diversity within the genus found in Asia . Given the strong evidence for an allopolyploid origin for E. edwardsii , the diploid progenitors were presumably sympatric in high latitudes at the end of the Tertiary and during the Pleistocene, when hybridization coupled with unreduced gamete formation we found no evidence of somatic doubling (e.g., mixoploidy) and selfing resulted in whole genome duplication and the cytotypes observed in the polymorphic, polyploid E. edwardsii . Given the genetic differences deduced from the allozyme data, the progenitor taxa were genetically well differentiated, m uch more so than expected for sibling species . And while morphometric analyses examining species boundaries between E. penlandii and E. edwardsii reveal significant differences, they are admittedly subtle (Regier, Pansing, and Bruederle, unreported data) . This uncoupling between morphological and genetic differentiation highlights the importance of recognizing cryptic species, especially with regard to polyploid complexes . Polyploidization between genetically divergent species has immediate consequences for genetic diversity of the progeny . Fixed heterozygosity can preserve genetic diversity when drift occurs due to founder effect or when plants self (Abbott and Brochmann 2003) . Populations of autopolyploids, however, will be more susceptible to loss of gene tic diversity due to polysomic inheritance coupled with drift or selfing . Regardless, multiple forms of evidence, in addition to the cytotype differences reported herein molecular sequence data, morphology, and distribution reveal E. penlandii to be dis tinct . Furthermore, it is reproductively isolated within E. edwardsii s.l. and, as such, we propose recognition of this taxon at the species level, despite it being morphologically cryptic .

PAGE 19

13 Based on the alleles present in each species, fixed heterozygosit y, and ploidy there is evidence of the E. penlandii genome in both the tetraploids and hexaploids and that there must have been multiple genome duplication events and at least one addit ional unidentified progenitor. This is supported by variation in dosage observed at fixed heterozygous loci in both tetraploids and hexaploids. Variation in dosage of alleles at fixed heterozygous loci is the result of multiple polyploidization events from a polymorphic progenitor (Wyatt et al. 1989). Finally, although we st udied only tetraploid and hexaploid populations of E. edwardsii , we cannot exclude the possibility that additional taxa were involved in the formation of this taxon, nor can we make conclusions regarding the origin of octaploid populations . Although octaploidy has been reported for this taxon (Knaben 1968), no octaploid populations were included as part of this research and, as such, we are unable to address its origin here . Ongoing research addressing polyploidy in E. edwardsii s.l . includes mapping cytotypes across the range of E. edwardsii and ecological niche modeling of cytotypes comprising the E. edwardsii species complex to detect ecological differentiation.

PAGE 20

14 Figures and Tables Fig . 1. Distribution of E. edwardsii s.l. Occurrences obtained from GBIF; map created by Dr. P. Anthamatten from the Department of Geography and Environmental Sciences, University of Colorado Denver . Near circumpolar distribution shows highest frequency in Beringia (Alaska and Ea stern Siberia) and Northern Canada, extending to Svalbard, Norway. Additional occurre nces in the Altai Mountains, Pamir Mountains, and Himalayas in Asia . Occurrences in Colorado represent E. penlandii Rollins.

PAGE 21

15

PAGE 22

16

PAGE 23

17

PAGE 24

18 Fig. 2. Chromosome count from r oot tip squash of E utrema penlandii Rollins (Brassicaceae) , 14 chromosomes (2n=2x=14) performed by N. Luebke (Milwaukee Public Museum ) . Root tips were saturated in paradichlorobenzene and fixed ; chromosomes were stained with 1% aceto orcein and photographed at 1000x.

PAGE 25

19 Table 3. Flow cytometry fluorescence data for E. penlandii Rollins and E. edwardsii R. Br.(Brassicaceae) populations sampled for ploidy determination. An internal standard, Pisum sativum . Taxon Site # Sample CV Standard CV Est. Ploidy 2n E. edwardsii Barrow East, AK 03 102601 4.6 441379 4.6 4.1 4x 07 108701 4.7 490736 4.0 4.4 4x 10 102973 4.7 487346 4.5 4.1 4x Barrow West, AK 03 107623 4.8 515001 4.6 4.3 4x 12 105710 2.7 511906 6.1 4.2 4x 29 107684 4.4 470868 4.8 4.3 4x Campbell Creek, AK 23 109796 4.8 4794 33 4.4 4.4 4x 27 112054 4.8 472957 4.3 4.5 4x 30 110633 4.7 489050 3.9 4.4 4x Denali Hwy, AK 03 178615 4.8 565773 4.3 7.2 6x 10 167368 4.8 524336 4.0 6.7 6x 12 167274 4.6 521411 4.1 6.7 6x Eagle Summit, AK 15 168828 3.7 601424 3.1 6.8 6x 20 171481 4.7 502898 4.4 6.9 6x 25 169061 4.2 520323 4.5 6.8 6x 12 Mile Summit 02 180523 4.7 515548 4.4 7.2 6x 06 168750 4.3 500931 4.3 6.8 6x 10 171250 4.4 507964 3.9 6.9 6x E. penlandii Blue Lake 04 49493 4.5 516318 4.8 2.0 2x 05 49312 3.9 547391 4.7 2.0 2x 06 49796 4.2 517080 4.6 2.0 2x Weston Pass 05 49153 3.3 550131 2.6 2.0 2x 07 49003 4.1 527668 2.5 2.0 2x 08 48860 3.9 519437 3.3 2.0 2x Horseshoe Gulch 01 49490 6.0 552199 2.9 2.0 2x 08 49350 4.6 516684 2.7 2.0 2x 10 49385 4.6 511149 2.7 2.0 2x Peerless Mine 1 04 50691 4.1 562289 2.8 2.0 2x 17 49286 4.6 538854 3.2 2.0 2x Peerless Mine 2 07 53024 4.0 559630 3.0 2.1 2x 10 50545 3.9 527425 2.9 2.0 2x 17 51295 3.6 506255 3.3 2.1 2x

PAGE 26

20 Fig. 3. Flow cytometry h istogram showing diploid, tetraploid and hexaploid cytotypes of Eutrema edwardsii s.l. R. Br. (Brassicaceae) (left to right) . Leaf tissue from three individuals was simultaneously chopped in a buffer to release nuclei , which were stained with pr opidium iodide (PI) for visualization on the FL2 channel. Fluorescence peaks indicate two and three fold increases in nucleus size for tetraploids and hexaploids, respectively .

PAGE 27

21 CHAPTER II ECOLOGICAL NICHE MODELING REVEALS DIVERGENCE BETWEEN TETRAPLOID AND HEXAPLOID POPULATI ONS OF EUTREMA EDWARDSII R BR. (BRASSICACEAE) Introduction Polyploidy or whole genome duplication (WGD) is a common and recurring phenomenon in plants that is now believed to have been involved in the origin of all angiosperms (Doyle and Sherman Broyles 2017) . The ubiquity of polyploidy suggest s that there is an increase in fitness conferred by increased genetic diversity , which can alter ecological tolerances and increase adaptability . There are several mechanisms by which polyploidy can occur, all of which involve unreduced gametes: 1) one step mechanism, involving two unreduced gametes; 2) autotetraploid triploid bridge, where a triploid is formed and subsequently rep roduces with an un reduced gamete ; 3) allotetraploid triploid bridge, same as the aforementioned, but involving divergent species; 4) higher ploidy one step; and 5) hybridization of autopolyploids (Mason and Pires 2015). The rate of unreduced gamete formation can increase wh en genetic and environmental factors cause meiotic aberrations, such as pre meiotic doubling, premature cytokinesis, mutation in synapsis, lack of spindle fiber segregation , and post meiotic fusion ( Sora et al. 2016). Environmental stress ha s been shown to increase the frequency of unreduced gamete formation in greenhouse and growth chamber experiments ( Ramsey and Schemske 1998) ; fluctuating temperature s and nutrient stress correlated with higher rates of unreduced

PAGE 28

22 gamete formation. Similarly, hybridizat ion has been shown to result in increase d unreduced gamete formation. Ramsey and Schemske (1998 ) reported a 49 fold increase in unreduced gamete formation for hybrids versus non hybrids among 22 species of flowering plants. This suggests that adverse conditions can promote polyploidy via the production of unreduced gametes , which in turn , can facilitate adaptation to a new niche . N iche is generally used to refer to a subset of environmental conditions in which a species can survive and reproduce ( Sillero 2011) , but often includes biotic interactions and dispersal . However, several different niche concepts exist . The fundamental niche is the broadest and includes the range of all abiotic variables where a species can maintain a viable population (Si llero 2011). The realized niche is where species interactions, such as comp eti tion, have not excluded a taxon from their fundamental niche. T he occupied niche is the portion of the realized niche in which populations are not restricted by geography, distri bution limitations or historical factors (Sillero 2011). The occupied niche is closest to what is being estimated from presence only niche modeling; here, occupied niche , unless otherwise specified. Polyploidization may b roaden the niche of a species by increasing ecological tolerances as a result of genome duplication. Added chromosomes and genetic material can effect immediate anatomical, morphological , and physiological changes in polyploids (Baker et al. 2017). Similarly, hybridizatio n can have a n additive effect on niche , broaden ing ecological tolerance to that of both progenitors combined , or a transgressive effect, changing niche tolerances as a result of novel gene products (Soltis et al. 2014) . Heterosis, or hybrid vigor, is a phe nomenon that describes faster growth rate, larger flowers, seeds and pollen and is

PAGE 29

23 often exploited in crop breeding; however, hybrids are often unstable or sterile (Miller et al. 2012). Allop olyploidy alleviates the problems associated with hybrids and stabilizes heterosis. Just as intermediate phenotypes are expected from hybrids, intermediate niches can be expected for allopolyploids, such that ideal environmental conditions comprise entirel y, or lie between , the tolerances of each progenitor (Harbert et al. 2014). Furthermore, t he added genetic diversity conferred by increased heterozygosity and phenotypic plasticity is thought to increase environmental tolerances (Brochmann et al. 1992, Lev in 2000). This increased adaptability allowed polyploids to colonize new habitats outside the niche of their progenitors . Current research is modeling ecological niche in polyploid complexes in order to elucidate environmental factors that influence polypl oid distribution and niche differentiation ( e.g., McIntyre 2012, Theodoridis et al. 2013, Glennon et al. 2014 ). Data extracted from climate databases and localities of populations can generate models that predict the probability that a species occur s in a given area (Thompson et al. 2014) . Likewise, niche models of one cytotype can be compared to another to test whether different cytotypes have similar niche requirements (Godsoe et al. 2013). There are three possible outcomes beside null difference that ca n result from niche modeling analysis (Glennon et al. 2014) : 1) n iche expansion , which would indicate that polyploidy confers increased environmental tolerance, allowing them to disperse to habitat outside the niche of progenitors ; 2) n iche s hift, would in dicate a change in niche requirements due to genome duplication, which would still support the hypothesis of polyploidy as an evolutionary mechanism (Van Daijk and Bakx Schotman 1997) ; and 3) n iche

PAGE 30

24 contraction , which would suggest that polyploidy does not increase environmental tolerances or, that polyploids have been restricted to limited habitat due to inbreeding , local adaptation or competition from di ploid populations. Previous work has provided a large framework into which to fit my model and, as a con sequence, any of these results will be informative of niche differentiation among cytotypes. Several recently published studies provide evidence for ployploids having different niche conditions, wider niche breadth, and a trend toward harsher niche conditi ons in comparison to diploids. In Chamerion angustifolium (L.) Holub (Onagraceae), Thompson et al. (2014) showed that autotetraploid populations occupied habitats that were warmer and drier than diploid populations. However, not all polyploids exhibit a po sitive correlation bet ween niche breadth and ploidy. Theodoridis et al. (2013) demonstrated a decrease in niche breadth as ploidy increased in four closely related species of allopolyploid Primula sect. Aleuritai (Primulaceae). They found that although nic he breadth decreased with ploidy, each cytotype had a distinct niche based on seven variables of temperature and precipitation . The results of both Thomposon et al. (2014) and Theodoridis et al. ( 2013) provide evidence that polyploids have adaptations to novel conditions that allow them to colonize habitat un suitable to diploids. The distribution of polyploids from an ecological perspective can offer insight to how and why polyploidy is so ubiquitous, particularly in the arctic, where polyploidy is more fr equent (Brochmann et al. 2004). The arctic flora was greatly influenced by glaciation cycles that began in the late Pliocene (Abbott and Brochmann 2003) . Plants that survived in the arctic from the Tertiary to the present are those that previously occurre d in wetland and riparian habitats (Murray

PAGE 31

25 1981) . Early in the Quaternary, plants that occurred in the arctic were largely destroyed or fragmented by advancing ice sheets (Abbott and Brochmann 2003). However, the region of Northeas t Russia and Northwest Am erica, Beringia, remained ice free throughout the Quaternary, there by serving as a refuge. As glaciers receded, some arctic taxa were more successful at recolonizing the deglaciated habitat. The repeated fragmentation, recolonization and reunion of previo usly isolated populations have created a complex mixture of hybrid species that have been stabilized by genome doubling (Abbott and Brochmann 2003) . Reticulate polyploidization and hybridization has resulted in increasingly high ploid hybrids characterize d by high or fixed heterozygosity, which maintains genetic diversity in newly formed populations experiencing inbreeding or bottleneck s (Abbott and Brochmann 2003). These hybrid polyploids appear to be better suited for colonizing the recently deglaciated areas, as suggested by the higher ploidies found in glaciated areas and lower ploidies found in the refugia (Brochamann et al. 2004). Few studies have addressed polyploid niche modeling in arctic species, where polyploidy is more frequent. Here , I use modeling to compare ecological niche in tetraploid and hexaploid populations of E. edwardsii , with a goal of testing the null hypothesis of no difference. For these purposes, p loidy was determined by flow cytometry from herbarium tissue and field colle cted samples representing the range of E. edwardsii in North America . A Principle Components A nalysis was performed using climatic data to r eveal environmental correlates with respect to ploidy. Ecological n iche models were then generated to estimate niche breadth, niche overlap , niche identity and niche similarity. The ultimate goal of this

PAGE 32

26 research was to determine if there is divergence between tetraploid and hexaploid E. edwardsii , as theory predicts. Materials and Methods Field and Herbarium colle ctions. Ninety three populations of E. edwardsii were sampled for ploidy determination from across North America , as well as Beringia, specifically (Table 4) . Here, populations are represented by individuals , as no mixo ploid populations w ere detected among E. edwardsii population s ( Chapter 1). Leaf tissue was obtained from 79 herbarium accessions across northwestern North America from St. Mathew Island, Alaska , USA to Devon Island, Nunavut, C A N, as well as 14 field collect ions from Barrow, Alaska to Dawson , Yukon, C AN (including those from Chapter 1) . Leaf tissue was harvested from herbarium accessions maintained at the University Herbarium in the Museum of Natural History at the University of Colorado Boulder (COLO) and the Herbarium at the Museum of the North at the University of Alaska Fairbanks (ALA) . One or two leave s were harvested from all accessions obtained since 19 79 that had adequate tissue for flow cytometry . Three individuals, collected prior to 19 79 ( one as old as 19 48 ) were sampled to test the efficacy of FCM ploidy determination from older samples. Tissue was stored in silica gel until assayed. Flow cytometry . (2017) . F luorescence values were compared to that of E. penlandii (2n) to determine ploidy . The protocol used to estimate ploidy from h erbarium accession was modified as follows d ue to the limited amount of tissue available: 1) Solanum lycopersicum CV. Stupicke (2C DNA = 1.96pg) and Glycine max C V. Polanka (2C DNA = 2.50pg) were used as standards ; 2) due to overlap of nuclei size , the standards were assayed individually once for every 6 10 samples;

PAGE 33

27 3) a higher threshold for coefficient of variation (CV = 10%) was set , due to lower tissue quality , as compared to silica dried samples ; and 4) genome size was obtained for only one individual from each herbarium collection, which here, represents a population. Environmental niche modeling. Environmental data layers for modeling were obtained from W orld clim .org (Hijmans et al. 2005) at 30 arc seconds and trimmed down to the North American range of E. edwardsii (Latitude 50 85 , Longitude 10 175 ) . Principle Components Analysis (PCA) was performed in R (ver. 3.3.1; R Core Team 2016) with 19 biocli matic variables and altitude to identify the variables with the highest contribution to the variance among samples. The highest contributing variables the first two axes that described 83% of the variation among variables were chosen as candidates. Reliabl e maximum entropy niche modeling has been shown to be effective with as few as five occurrences for every explanatory variable used (Godsoe et al. 2013). This rule was used as the maximum number of variables for niche modeling. Niche models were generated in Maxent (ver. 3.3.3k; Phillips et al., 2004, 2006), a machine learning application that builds models by finding the probability distribution of maximum entropy (Phillips et al. 2006). This produces a model with the most unifo rm distribution of suitability, for a given set of occurrences, across all variables. Maxent niche models were used in ENMTools (ver. 1.2; Warren et al. 2008) to calculate niche breadth, niche overlap ( Warren et al. 2008), and niche identity in order to test the hypothesis of niche divergence between cytotypes. Niche breadth is the proportion of environmental conditions that is occupied within the available range of conditions in a given area. Comparisons of niche breadth of m ultiple taxa

PAGE 34

28 using the same geographic range indicate which taxon occurs in a wider range of the available environmental conditions . Niche overlap estimates the proportion of similarity between models, where 0 = no overlap and 1 = identical models. D is a metric that was developed before presence only niche modeling and assume s that the variance of each variable reflects the rel ative use of that variable; as such, the mean of a varia ble confers the highest density . I , however, removes that assumption so that the variance reflect s probability distributions (Warren et al. 2008). The niche identity test estimates the overlap of one model to a model generated from randomizing the samples of both models. Overlap of the original models was compared to the d istribution of overlap from the randomized models. Niche identity test whether two populations were drawn from the same distribution of environmental variables (Warren et al. 2008). Lower identity scores indicate greater divergence between the distribution of variables occupied by each species. Maxent model parameters that differed from the default were set as follows: random test percentage = 25%; iterations = 5 000 ; random seed = true, add samples to background = true . Results Flow cytometry. Of the 79 herbarium accessions sampled for flow cytometry , 47 (5 9 . 5 %) provided interpretable results . Flow cytometry revealed three cytotypes from among the herbarium and field collected samples of E. edwardsii , in c luding 26 tetraploid s , 3 3 hexaploids , and two octaploids (Table 4 ) . S amples without specific coordinate data that could not be georeferenced , duplicate samples that were considered the same population , and samples with missing environmental d ata were not included in the niche modeling ,

PAGE 35

29 leaving 21 tetraploids and 30 hexaploids; due to low sample size, octaploids were not included in Maxent modeling (Godsoe et al. 2013) . F luorescence values for tetr aploids were 92,985 115,724 (mean= 102,558, sd= 5842 ) and hexaploids were 130 ,003 179,377 (mean= 155,021, sd= 10,387 ) ; CV s ranged from 3.1 to 7.3 (Table 4 ). Environmental niche modeling. The first two axes of the PCA describe 83% of the variation . The variables with the highest loading on these c omponents were BIO 2 (mean diurnal range, mean of monthly temperature ranges), BIO 4 (temperature seasonality), BIO 5 (max temperature of the warmest month), BIO 12 (annual precipitation), and BIO 16 (precipitation of the wettest quarter) (F ig. 4 ) . The res ults of a PCA revealed that hexaploids occupy habitat that has greater diurnal range, greater temperature seasonality, higher maximum temperatures, with moderately higher precipitation maxima, as compared to tetraploids. Niche models were created in Maxent for tetraploids (n=21 ) and hexaploids (n=30) (Fig. 5 ). Area under the curve (AUC) scores from receiver operating characteristic (ROC) curves of M axent models for tetraploid and hexaploid E. edwardsii were 0.901 ± 0.037 and 0.928 ± 0.065, respectively. Thi s score indicates how well the model predicts the presence of occurrences, such that an AUC of 0.500 would indicate random predictions and AUC = 1.000 would indicate no false positives and no false negatives . BIO 5 (precipitation seasonality) had the highe st contribution to the tetraploid model, both for model gain when used by itself and loss of gain when omitted. For the hexaploid model, BIO 2 (mean diurnal range) had the highest gain in isolation and the highest loss when omitted from the model.

PAGE 36

30 Niche b readth was calculated for each cytotype model revealing higher breadth for tetraploids (niche breadth = 0.521) as compared to hexaploids (niche breadth = 0.159). Maxent models revealed niche overlap of tetraploids and hexaploids ( D = 0.354, I = 0.654 ) that was lower than expected for niche equivalency ( T able 5 ). Niche identity of Maxent tetraploid and hexaploid models were 0.639 and 0.589, respectively. This indicates that the tetraploid model has greater overlap with polyploids combined than does the hexap loid model. Suitability maps created in Maxe nt revealed a spatially broad range comprising mostly coastal and low elevation areas for tetraploids and a narrow range of alpine regions for hexaploids ( F ig. 5 ). Suitability is highest for tetraploids across a ll coastal areas in North America from the Pacific coast across northern Canada to Greenland and Iceland. The hexaploid model shows highest suitability in eastern Alaska and Yukon Territory, and a small area of moderate suitability north of Hudson Bay . Di scussion Polyploidy is ubiquitous, especially in the arctic where it characterizes up to 80% of plant species (Abbott and Brochmann 2003, Soltis and Soltis 200 9) , including a large proportion of allopolyploids . The high frequency of polyploids in the arctic suggests that polyploidy may increase ecological tolerance or adaptability i n new or changing habitats. The proposed success of polyploids , and allopolyploids specifically, is presumably due to the added genetic material and novel informatio n resulting from genome doubling and hybridization, respectively (Brochmann et al. 2004). The added genetic material thereby increases the

PAGE 37

31 phenotypic plasticity of individuals and diversity of populations, facilitating adaptation to habitats outside the ni che of the progenitors . For synthetic allopolyploids, these changes can occur immediately after formation or after as few as five generations of selfing (Chen and Ni 2006). If gene expression and regulatory changes occur at ecologically important loci , ra pid adaptation can result. Pires et al. (2004) showed that the synthetic allopolyploid Brassica napus segregated into early and late flowering lineages after on ly six generations of selfing. This rapid phenotypic change can result in adaptation to new envi ronments. More specifically, changes in gene expression and regulation due to environmental conditions are expected to promote successful colonization of polyploid populations in novel niches (Chen and Ni 2006). Thus, polyploids would have an advantage in novel niches resulting from environmental change. Ploidy determination. Although f low cytometry ploidy determination using herbarium tissue has previously been demonstrated to be successful by Roberts ( 2007) and others , h ere I show that ploidy can be determined from material that is well over 25 years old . The success rate reported here for herbarium samples was 60 % and comprises samples spanning the entire range of collection dates ( T able 4 ). Unsuccessful assays result fr om nuclei degradation most likely associated with the age of the plant when collected and drying conditions. The use of herbarium accessions for ploidy determination and ecological niche modeling will allow studies to use larger data sets with minimal samp ling effort and to allow niche comparisons across time. Ploidy data for five populations of E. edwardsii suggest no mixo ploid populations in this sample , suggesting populations are comprised of a single cytotype, and thus populations of

PAGE 38

32 cytoty p es are assu med to be distinct . Although mixo ploid populations are common among polyploid species, discrete populations are expected under minority cytotype exclusion (Levin 1975), such that low frequency cytotypes will be eliminated from populations due to the low frequency of ploidy matched gametes. Likewise, allopolyploids are expected to have a higher fitness in habitat outside the niche of the lower ploidy cytotyp es and therefore expected to either colonize novel niches to avoid competition with progenitors or outcompete progenitors in the parental population. These results suggest that tetraploid and hexaploid populations of E. edwardsii occupy divergent niches an d add support to the hypothesis that polyploidy facilit ates adaptation to new and changing environments based on the following evidence: Niche differentiati on was observed between tetraploid and hexaploid populations of E. edwardsii ( D = 0.354 , I = 0.654 ) ; h exaploids occupy a niche that has both higher temperature and precipitation maximums than that of tetraploids (Fig. 4 ), with a smaller geographic range of suitable habitat (Fig. 5 ). The difference between niche means supports niche shift occurred for hexaploids, whereas a smaller niche breadth also supports niche contraction. The apparent niche shift may be the result of niche intermediacy relative to the progenitors or niche novelty resulting from the union of previously divergent genomes, providing u nique gene products , and thus phenotypes, relative to either progenitor (Soltis et al. 2014) . Subsequent evolution can limit the niche through homeolog loss resulting in local adaptation . Alternatively, the disruption of a cytotype may be limited by the av ailable habitat that is within the new niche. That is to say, hexaploids may have a broader fundamental niche, but

PAGE 39

33 the occupied niche is smaller than tetraploids due to competition and availability of suitable habitat. Niche modeling. A Principle Componen ts A nalysis revealed five variables that contribute most to the variation observed among all variables. These variables describe the most variance among environmental variables of the two cytotypes and thus the direction of differentiation hexaploids under went. Ecological niche modeling revealed divergence between tetraploid and hexaploid populations of E. edwardsii , with hexaploids occupying a niche that has greater temper ature range with higher maxima and more precipitation as compared to tetraploids. These results support the hypothesis that increased ploidy facilitated adaptation to habitats that are outside the range of the progenitors in E. edwardsii . Ecological niche models provided estimates of overlap between ploidy levels that revealed divergence between cytotypes. Niche breadth for tetraploids is higher than that of hexaploids, which indicates that hexaploids occupy a narrower niche compared to tetraploids, which supports the hypothesis of nic he contraction. However, hexaploids may be restricted in their range due to limited availability of habitat that is outside the niche of tetraploids. It may be that the fundamental niche of hexaploids is much broader than the occupied niche depicted in thi s paper. Presence only niche modeling provides an estimate of a subset of the occupied niche, which is limited to the scope of variables used. The observed niche shift supports the hypoth esis of niche divergence between cytotypes. Niche overlap is expected to decrease with increasing divergence among taxa, such that overlap is higher for more closely related species; thus, high niche overlap would be

PAGE 40

34 expected for infraspecific taxa. R esults for E. edwardsii indicate less overlap than that from other studies in which niche differentiation was concluded ( e.g. , McIntyre 2012). However, D tends to overestimate overlap because of the amount of equally unsuitable habitat. Niche overlap increases when the models have large areas of habitat that are unsuitable for b oth cytotypes. Therefore, niche divergence may be greater than indicated by this statistic alone . The I statistic for niche overlap removes the assumption that presences represent the full range of variables occupied by the taxon, which makes this statisti c more conservative. Even under this relaxed assumption I = 0.654 suggests divergence between cytotypes . The identity test provides a more robust estimation of the equivalency of two sets of data (Warren et al. 2008) by treating both cytotypes as one group and randomly sampling to generate many replicate models. The low overlap bet ween replicate models and each original model provides further support that these two cytotypes do not occupy similar niches. Whether the differentiation in niches is caused by lo cal adaptation, novel genetic information from a potential third progenitor to the hexaploids , or a direct consequence of genome duplication cannot be elucidated form these data. Nevertheless, the degree of niche overlap combined with the difference in nic he breadth observed suggests that polyploidization facilitated novel niche adaptation in E. edwardsii . There are inherent limitations associated with presence only niche modeling that have been addressed in the past few years (Sillero 2011, Warren 2012). Most notably is the unknown relationship between the o ccupied niche (represented by the model) and fundamental niche. It is important to recall that niche models represent the occupied niche that may be restricted due to variables not consider ed in niche m odels, such as geography,

PAGE 41

35 soil composition, dispersal , and competition . Conversely, the fundamental niche may be smaller than the model when a species is locally adapted to microclimates or rare niches. The variables used to generate models may also be que stioned as , in many ENM, including this study, variables are not chosen for their influence over biological tolerances , but rather for their contribution to the variance among variable s. However, Warren (2012) argued that as long as the assumptions of ENM are understood, meaningful conclusions can still be made. The assumptions he describes are as follows: ENMs are at least some subset of the niche that a taxon occupies , whether or not the variables used are the most influential to the niche; the explanator y power of ENMs is , at worst , a result of spatial autocorrelation among variables, but would still inform comparisons between models. That is, most climatic variables are correlated, so even if the variables used to generate models are not the most influen tial to the niche, some aspect about th ose sets of conditions is . T herefore, models do not need to be perfect representations of a species niche to detect divergence. Based on these assumptions, it can be stated that under any circumstance, it is not possible to calculate the fundamental niche of a species; rather, niche modeling elucidates the range within variables that characterize a niche (Godsoe 2010, Sillero 2011) . Likewise, ecological niche models are better understood as a simulation of a species distribution , rather than a description of the niche (Jiménez Valverde et al. 2008, Sillero 2011). Therefore, we can be confident that differentiation observed between n iche models represents real differentiation for at least a subset of a species niche.

PAGE 42

36 Pleistocene glaciations created an environment characterized by large scale habitat destruction and fragmentation. As glaciers receded, habitat was sporadically opened offering habitat void of competition but, likely, less suitable than refugia where Eutrema presumably persisted. These characteristics are ideal for neo polyploids, which are less fit in refugia due to minority cytotype exclusion and potentially more fit i n novel habitat due to increased genomes and/or hybridization. Multiple Eutrema linages existed prior to and possibly throughout polyploidization, which provided enough variation to adapt to a harsh and ever changing environment. Today, a ll but one of thos e lineages are either extinct or not described, which limits our ability to elucidate relationships between progenitor and progeny. Future Directions. E nvironmental niche model analysis of North American Eutrema will continue by modeling the niche of the diploid, E. penlandii , and the niche of polyploids ( E. edwardsii) combined . Niche models of diploids will be extrapolated to the range of E. edwardsii to determine niche overlap of the two species . This will prov ide a context in which to evaluate niche evolution in North American Eutrema . However, it may not be possible to determine if divergence between the two species is a result of niche intermediacy or post polyploidization adaptation, because only one progeni tor is described . Therefore, genetic studies need to be performed to better understand the relationship between E. penlandii and E. edwardsii and the degree of divergence between progenitors. It would be helpful to know the timescale for genome duplication and hybridization events as well as the contribution of each genome to the to polyploids, assuming there were multiple origins and possibly independent origins for different cytotypes.

PAGE 43

37 Fut ure studies should also examine microclimates, vegetation, and other biotic and abiotic variables to further describe niches and elucidate differentiation i n other ecological dimensions . Eutrema grows in specific sites that might share temperature and prec ipitation conditions with areas not suitable for this species , such as alpine habitats that are not wetlands . Similarly, there may be competition or seed dispersal limitations that are not co nsidered in this analysis. The study of the ecological implicatio ns of polyploidy will provide further insight into the evolutionary history of E. edwardsii and broaden our knowledge of the impact polyploidy has had on the arctic f lora .

PAGE 44

38 Tables and Figures Table 4 . Flow cytometry data for E. edwardsii samples harvested by Leo P. Bruederle or from herbaria at University of Colorado (COLO) and University of Alaska (ALA) . Source / H erbarium Site Identifier Collection Date Latitude Longitude Fluorescence CV Ploidy LPB lab 702201101 07/02/11 64.1038 140.8778 167535.02 4.23 6x LPB lab 705201101 07/05/11 65.3996 145.9763 173507.45 4.47 6x LPB lab 710201102 07/10/11 68.1845 149.4251 174252.62 4.72 6x LPB lab 710201103 07/10/11 68.5513 149.4949 161730.30 5.08 6x LPB lab 711201102 07/11/11 68.6481 149.4027 167702.18 4.16 6x LPB lab 711201103 07/11/11 68.3032 149.3595 170165.70 4.45 6x LPB lab 717201002 07/17/10 64.6657 138.3914 167189.04 4.41 6x LPB lab 721201001 07/21/10 71.2641 156.8296 104758.49 4.66 4x LPB lab 721201002 07/21/10 71.2516 156.5060 109264.12 4.57 4x LPB lab 722201001 07/22/10 71.2392 156.3358 107005.55 4.56 4x LPB lab 0704201101a 07/04/11 63.0727 145.7279 171085.47 4.73 6x LPB lab 0706201101a 07/06/11 65.4849 145.4040 169789.97 4.17 6x LPB lab 0717201001b 07/17/10 64.9517 138.2713 146708.01 4.41 6x LPB lab Campbell Creek N/A 61.0688 149.6056 111558.98 4.64 4x COLO COLO 346724 07/10/80 70.3667 148.5333 97587.37 6.49 4x COLO COLO 419537 07/13/84 70.0167 148.7500 97782.74 4.95 4x COLO COLO 419673 07/25/84 70.3000 149.2000 99092.80 4.95 4x ALA V071453 08/04/79 70.2833 148.5000 206863.83 5.57 8x ALA V117025 06/13/94 65.6333 146.7556 155475.61 4.92 6x ALA V118847 06/28/94 61.2472 149.5875 109699.34 5.00 4x ALA V119651 06/14/95 65.6167 147.3667 148780.81 6.57 6x ALA V119938 07/23/95 64.4270 173.2189 99559.50 5.19 4x ALA V121861 07/03/96 65.1167 144.0500 135727.82 4.92 6x ALA V122304 06/18/95 65.8389 145.8833 147233.24 4.88 6x ALA V122877 07/21/97 68.6000 156.5000 95988.25 4.88 4x ALA V123470 07/15/97 63.6667 160.7000 95546.51 4.98 4x ALA V127658 07/14/99 75.0667 98.5000 104675.67 5.37 4x ALA V129584 07/26/99 69.7667 122.0833 185531.21 6.54 6x ALA V131261 08/22/98 63.8278 146.8492 157154.37 4.91 6x ALA V133443 08/23/99 76.5191 86.7662 103031.74 4.86 4x ALA V133475 08/26/99 74.5333 82.7833 115762.52 3.91 4x ALA V134036 06/26/01 67.2700 163.6700 106758.54 4.90 4x ALA V134203 07/01/01 67.1133 163.3633 96527.73 4.92 4x

PAGE 45

39 Table 4 . Cont. Source / Herbarium Site Identifier Collection Date Latitude Longitude Fluorescence CV Ploidy ALA V137839 07/05/02 68.5367 161.4664 100766.76 4.77 4x ALA V139646 07/28/02 68.2394 152.1944 97054.02 4.86 4x ALA V139812 07/19/02 68.1950 152.7444 158004.68 4.28 6x ALA V139958 07/20/02 68.1900 152.7400 147925.88 8.46 6x ALA V140421 06/21/01 68.6110 149.6437 158840.05 6.25 6x ALA V140748 07/29/02 68.5644 152.5344 153299.58 4.77 6x ALA V141037 07/27/02 68.3333 151.0333 173159.74 9.00 6x ALA V141828 07/11/02 64.6250 165.6750 101525.72 4.83 4x ALA V146945 07/07/02 64.5398 143.7021 164457.88 4.87 6x ALA V151202 07/19/03 61.3163 142.2703 152331.41 4.86 6x ALA V153299 06/08/03 62.4000 141.9500 152403.85 5.21 6x ALA V158770 07/18/97 60.6001 172.9300 100831.20 4.32 4x ALA V158827 07/22/97 60.5500 172.9167 103520.74 4.21 4x ALA V160941 06/11/07 66.8000 141.0667 166189.84 4.91 6x ALA V174587 06/22/09 64.5240 143.5907 148235.13 4.84 6x ALA V79106 06/28/84 63.8000 148.9333 144776.87 7.80 6x ALA V82602 08/01/85 68.6167 149.3167 148647.32 4.75 6x ALA V83059 07/07/84 70.0000 144.4333 108508.59 8.76 4x ALA V95009 08/13/87 69.1167 105.0833 205937.14 6.40 8x ALA V97542 06/08/88 64.5728 163.7542 100373.33 7.58 4x

PAGE 46

40 Fig. 4 . PCA plot showing the loading of Bioclim 19 variables and altitude (downloaded from worldclim.org) for the first two components , PC1 and PC2 , which comprise 8 3 % of the variation among variables. Tetraploid niche space is outline in blue and hexaploid niche space is outlined in red .

PAGE 47

41 Fig. 5 . Maxent niche models of North American E utrema edwardsii R. Br. (Brassicaceae) for tetraploid ( n = 21 ; top) and hexaploid ( n = 30; bottom) populations. Maps show potential suitability, based on input variables of temperature and precipitation , ranging from 0 (blue) to 1 (red) .

PAGE 48

42 Table 5 . Mode ling statistics from ENMTools. Overlap is a symmetrical calculation, whereas Identity is calculated as original cytotype model compared to models from a combined distribution of both cytotypes . Cytotype Niche Breadth Overlap (D) Overlap (I) Identity Tetraploid 0.5 21 0.354 0.654 0.63 9 Hexaploid 0.1 59 0.354 0.654 0.589

PAGE 49

43 REFERENCES Abbott RJ and Brochmann C. 2003. History and evolution of the arctic flora: in the footsteps of Eric Hultén . Molecular Ecology . 12: 299 313. Al Shehbaz IA and S.I. Warwick . 2005 . A synopsis of Eutrema (Brassicaceae) . Harvard Papers in Botany 10: 129 135. Chen ZJ and Ni Z. 2006. Mechanisms of genomic rearrangements and gene expression changes in plant polyploids. BioEssays . 28: 240 252. Baker RL, Yarkhunova Y, Vidal K, Ewers BE, Weinig C. 2017. Polyploidy and the relationship between leaf structure and fu nction: implications for correlated evolution of anatomy, morphology and physiology in Brassica . BMC Plant Biology . 17(3) DOI 10.1186/s12870 016 0957 3 Bickford D, Lohman DL, Sodhi NS, Ng PKL, Meier R, Winker K, Ingram KK, Das I. 2006. Cryptic species as a window on diversity and conservation. Trends Ecol. Evol. 22: 148 155. Böcher TW and K. Larsen . 1950 . Chromosome numbers of some arctic or boreal f lowering plants . Medd. Grønl. 1947(6). Brochmann C, Soltis PS, Soltis DE. 1992. Recurrent formation and polyphyly of Nordic polyploids in Draba (Brassicaceae). American Journal of Botany. 79(6): 673 688. Brochmann C, Br ysting AK, Alsos IG, Borgen L, Grundt HH. 2004. Biological relevance of polyploidy: ecology to genomics Polyploidy in arctic plants. Society 82(4): 521 536. Cytometry. Nature Protocols. 2(9): 2233 2244. Fayette K and Bruederle LP. 2001. Morphometric analyses support specific status for Eutrema penlandii Rollins (Brassicaceae). Botany Abstract Grundt HH, Obermayer R, Borgen L. 2005. Ploidal levels in the arctic alpine polyploidy Draba lacteal (Brassicaceae) and its low ploid relatives. Bot. J. Linn. Soc . 147: 333 347. Grundt HH, Kjølner S, Borg en L, Rieseberg LH, Brochmann C . 2006 . High biological species diversity in the arctic flora . Proceedings of the National Academy of Sciences 103(4): 927 975. Glennon KL, Ritchie ME, Segraves KA. 2014. Evidence for shared broad scale climatic niche of dip loid and polyploidy plants. Ecology Letters . 17: 574 582.

PAGE 50

44 Godsoe W, Larson MA, Glennon KL and Segraves KA. 2013. Polyploidization in Heuchera cylindrca (Saxifragaceae) did not result in a shift in climatic requirements. American Journal of Botany. 100(3 ): 496 508. Hardwick RC. 1997 . Population Genetics of the Rare Alpine Endemic Eutrema penlandii University of Colorado Denver. Harbert RS, Brown AHD, Doyle JJ. 2014. Climate niche modeling in the perennial Glycine (Leguminosae) allopolyploidy complex. American Journal of Botany . 104(4): 710 721. Husband BC and Schemske DW. 1998. Cytotype distribution at a diploid tetraploid contact zone in Chamerion ( Epilobium ) angustifolium (Onagraceae). American Journal of Botan y . 85(12): 1688 1694. Jørgensen CA, Sørensen T, Westergaard M . 1958. The flowering plants of Greenland: A taxonomical and cytological survey. Kongel. Danske Vidensk. Selsk. Biol. Skr . 9: 1 172. Jiménez Valverde A, Gómez JF, Lobo JM, Baselga A, Hortal J. 2008 . Challenging species distribution models: the case of Maculinea nausithous in the Iberian Peninsula. Annales Zoologici Fennici. 45(3): 200 210. Knaben G. 1968. Chromosome numbers of flowering plants from central Alaska USA polyploidy. Nytt Magasi n for Botanikk (Oslo ) 15(3): 240 254. Kron P and Husband BC. 2012. Using flow cytometery to estimate pollen DNA content: Improved methodologies and applications. Annals of Botany . 110: 1067 1078. Levin DA. 1975. Minority cytotype exclusion in local plan t populations. Taxon , 35 43. Levin DA. 2000. The origin, expansion, and demise of plant species. Oxford University Press. Mason AS and Pires JC. 2015. Unreduced gametes: meiotic mishap or evolutionary history?. Trends in Genetics . 31(1): 5 10. McAllist er C, Blaine R, Kron P, Bennett B, Garrett H, Kidson J, Miller AJ. 2015. Environmental correlates of cytotype distribution in Andropogon gerardii (Poaceae). American journal of botany . 102(1): 92 102. McIntyre PJ. 2012. Polyploidy associated with altered and broader ecological niches in the Claytonia perfoliata (Portulacaceae) species complex. American Journal of Botany. 99(4): 655 662

PAGE 51

45 Miller M, Zhang C, Chen ZJ. 2012. Ploidy and Hybridity Effects on Growth Vigor and Gene Expression in Arbidopsis thaliana Hybrids and Their Parents. G3: Genes, Genomes, Genetics. 2: 505 513. Murray DF. 1981. The role of arctic refugia in the evolution of the arctic vascular flora a Beringian perspective, pp. 11 20. In G. G. E. Scudder and J. L. Reveal, eds. Evolution toda y, proceedings of ICSEB II. Hunt Institute for Botanical Documentation, Pittsburgh. Otto SP and Whitton J. 2000. Polyploid incidence and evolution. Annu. Rev. Genet . 34: 401 437. Paun O, Forest F, Fay ME, Chase MW. 2006. Hybrid speciation in angiosperms: parental divergence drives ploidy. New Phytol . 182: 507 518. Paun O, Forest F, Fay MF, Chase MW. 2009. Hybrid speciation in angiosperms: parental divergence drives ploidy. New Phy tologist . 182: 507 518. Pires JC, Zhao J, Schranz ME, Leo EJ, Quijada PA, Lukens LN and Osborn TC. 2004. Flowering time divergence and genomic rearrangements in resynthesized Brassica polyploids (Brassicaceae). Biological Journal of the Linnean Society . 8 2: 675 688. R Core Team (2016). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R project.org/. Ramsey J and Schemske DW. 1998. Pathways, mechanisms, and rates of polyploid y formation in flowering plants . Annual Rev. Ecol. Syst. 29: 467 501. Roberts AV. 2007. The Use of Bead Beating to Prepare Suspensions of Nuclei for Flow Cytometry from Fresh Leaves, Herbarium Leaves, Petals and Pollen. Cytometry Part A. 71A: 1039 1044. Sillero N. 2011. What does ecological modeling model? A proposed classification of ecological niche models based on their underlying methods. Ecological Modelling . 222: 1343 1346. Sokolovskaja AP and Strelkova OS . 1941 . Polyploidy and karyological races under conditions in the Arctic. C.R. (Doklady) Acad. Sci. .U.S.S.R . 32: 144. Soltis, DE, Haufler CH, Darrow DC, Gastony GJ. 1983. Starch gel electrophoresis of ferns A compilation of grinding buffers, gel and electrode buffers, and staining schedules. A mer. Fern J .73: 9 27.

PAGE 52

46 revisited. American Journal of Botany . 101(7): 1057 1078. Soltis PS and Soltis DE. 2000. The role of genetic and genomic attributes in the succ ess of polyploids. Proceedings of the National Academy of Sciences 97(13): 7051 7057. Soltis DE, Soltis PE, Schemske DW, Hancock JF, Thompson JN, Husband BC, Judd WS . 2007. Autopolyploidy in Angiosperms: Have we grossly underestimated the number of specie s. Taxon 56(1): 13 30. Soltis PS and Soltis DE. 2009. The role of genetic and genomic attributes in the success of polyploids. PNAS . 97(13): 7051 7057. Sora D, Kron P, Husband BC. 2016. Genetic and environmental determinants of unreduced gamete production in Brassica napus, Sinapis arvensis and their hybrids. Heredity . 117: 440 448. Theodoridis S, Randin C, Broennimann O, Patsiou T, Conti E. 2013. Divergent and narr ower climatic niche characterize polyploidy species of European primroses in Primula sect. Aleuritia . Journal of Biogeography . 40: 1278 1289 Thompson KA, Husband BC and Maherali H. 2014. Climatic niche differences between diploid and tetraploid cytotypes of Chamerion angustifolium (Onagraceae). American Journal of Botany. 101(11): 1868 1875 . Van Dijk P and Bakx Schotman T. 1997. Chloroplast DNA phylogeography and cytotype geography in autopolyploid Plantago media. Molecular Ecology. 6: 345 352. Wallac e LE. 2003. Molecular Evidence for Allopolyploid Speciation and Recurrent Origins in Platanthera huronensis (Orchidaceae). International Journal of Plant Science . 164(6): 907 916. Warren DL, Glor RE, Turelli M. 2008. Environmental niche equivalency versu s conservatism: quatitative approaches to niche evolution. Evolution . 62(11): 2868 2883. Trends in Ecology and Evolution . 27(9): 497 500. Warwick, S.I., Al Shehbaz, I.A., Sauder, C.A. 2006. Phylogenetic po sition of Arabis arenicola and generic limits of Aphragmus and Eutrema (Brassicaceae) based on sequences of nuclear ribosomal DNA . Canadian Journal of Botany 84: 269 281. Table 1. Locations for Eutrema penlandii Rollins and E. edwardsii R. Br. (Brassicaceae ) populations sampled for chromosome counts 1 , flow cytometry 2 , and allozyme analysis 3 .

PAGE 53

47 Weber WA. 1985. New names and combinations, principally in the Rocky Mountain USA Flora V . Phytologia 58: 382 384. Weber WA. 2003. The Middle Asian Element in the Sout hern Rocky Mountain flora of the western United States: a critical biogeographical review . J. Biogeogr . 30: 649 685. Weber WA, Wittmann RC. 2012. Colorado Flora: Eastern Slope. Afield guide to the vascular plants. Fourth Ed. University Press of Colorado. Boulder, CO. Wyatt R, Stoneburner A, Odrzykoski IJ. 1989. Bryophyte isozymes: systematics and evolutionary implications. In: Soltis DE, Soltis PS (eds) Isozymes in Plant Biology, Dioscorides Press, Portland: 221 240.

PAGE 54

48