Copyright

by

Andrew James Alverson

2006 The Dissertation Committee for Andrew James Alverson certifies that this is the

approved version of the following dissertation:

Phylogeny and evolutionary ecology of thalassiosiroid

Committee:

______Edward C. Theriot, Supervisor

______David M. Hillis

______Robert K. Jansen

______John W. La Claire II

______C. Randal Linder

Phylogeny and evolutionary ecology of thalassiosiroid diatoms

by

Andrew James Alverson, B.S.; M.S.

Dissertation

Presented to the Faculty of the Graduate School of

The University of Texas at Austin

in Partial Fulfillment

of the Requirements

for the Degree of

Doctor of Philosophy

The University of Texas at Austin

August 2006

Phylogeny and evolutionary ecology of thalassiosiroid diatoms

Publication No. ______

Andrew James Alverson, Ph.D.

The University of Texas at Austin, 2006

Supervisor: Edward C. Theriot

Salinity is a significant barrier to the distribution of diatoms, and though it is

generally understood that diatoms are ancestrally marine, the number of times diatoms

independently colonized fresh waters and the adaptations that facilitated these

colonizations remain outstanding questions in evolution. Resolving the exact number of freshwater colonizations will require large-scale phylogenetic reconstruction

with dense sampling of marine and freshwater taxa. A more tractable approach to

understanding the marine–freshwater barrier is to study a group of diatoms with high

diversity in each habitat. The "centric" diatom order affords an excellent

opportunity to study the origin and evolution of diatoms in fresh waters. Thalassiosirales

is a well-supported monophyletic group common in marine, brackish, and freshwater

habitats. Thalassiosirales species historically are classified into the marine

Thalassiosiraceae or freshwater Stephanodiscaceae, reflecting the more generally held

hypothesis that diatoms are naturally split along marine–freshwater lines. The fossil

record suggests that Stephanodiscaceae traces to a single colonization of freshwater in the

mid-Miocene, and in addition, Stephanodiscaceae species share a suite of complex cell

iv wall characters, which has been interpreted as corroborating evidence for their monophyly. I reconstructed the phylogeny of Thalassiosirales and used the phylogeny to test these and other hypotheses and to address a number of other problems related to the

marine–freshwater boundary in diatoms. Phylogenetic analyses showed strong evidence

for multiple colonizations of freshwater and reject all previous colonization hypotheses.

Results further show that part of Stephanodiscaceae is an early diverging lineage within

Thalassiosirales, indicating that these two distantly related and separately derived

Stephanodiscaceae lineages independently evolved a similar set of complex

morphological features upon or shortly after the colonization of fresh waters. Finally,

marine and freshwater diatoms, including Thalassiosirales, show several important

differences in silicon physiology. In addition to containing an order of magnitude more

silica in their cell walls, freshwater diatoms have a drastically lower enzymatic affinity

for silicic acid, the dissolved form of silica used by diatoms. I sequenced the silicon

transporter genes from marine and freshwater Thalassiosirales and show that

physiological differences are not due to differences in the coding sequence.

v

Table of Contents

List of tables...... vii List of figures...... viii Chapter 1. Introduction...... 1 Chapter 2. Stampeding the Rubicon: phylogenetic analysis reveals repeated colonizations of marine and fresh waters by thalassiosiroid diatoms...... 8 Introduction...... 9 Materials and Methods...... 13 Results...... 20 Discussion...... 25 Literature Cited...... 32 Chapter 3. Massive convergent evolution in morphology associated with the colonization of fresh waters in diatoms...... 74 Introduction...... 75 Materials and Methods...... 79 Results...... 83 Discussion...... 87 Literature Cited...... 93 Chapter 4. Strong purifying selection in the silicon transporters of marine and freshwater diatoms...... 107 Introduction...... 108 Materials and Methods...... 113 Results...... 117 Discussion...... 121 Literature Cited...... 127 Chapter 5. Cell wall morphology and systematic importance of ritscheri (Hustedt) Hasle, with a description of Shionodiscus gen. nov...... 147 Introduction...... 148 Materials and Methods...... 149 Results...... 151 Discussion...... 153 Literature Cited...... 162 Appendix...... 172 Bibliography...... 176 Vita...... 198

vi

List of Tables

Table 2.1. Taxa, culture strains, and GenBank accession numbers for chloroplast psbC and rbcL sequences used in Chapter 2...... 56 Table 2.2. Taxa, culture strains, and GenBank accession numbers for nuclear SSU and partial LSU rDNA sequences used in Chapter 2...... 64 Table 2.3. Oligonucleotide primers used to amplify and sequence SSU rDNA, partial LSU rDNA, psbC, and rbcL fragments from Thalassiosirales...... 72 Table 2.4. Data partitions, models of DNA sequence evolution, and results of Bayesian analyses, including Bayes factor comparisons of alternative models...... 73 Table 3.1. Maximum likelihood assessment of correlated evolution between habitat type and each of five morphological characters traditionally used to diagnose the predominantly freshwater Stephanodiscaceae...... 106 Table 4.1. Taxon, culture strain, habitat, and GenBank accession number for silicon transporter sequences analyzed in Chapter 4...... 139 Table 4.2. Number of free parameters (p), log-likelihood (l), and parameter estimates for random-sites and clade models...... 143 Table 4.3. Likelihood ratio test statistics (2Δ l), degrees of freedom (df), and p-values for comparisons of random-sites and clade models...... 144 Table 4.4. Number of free parameters (p, including parameters for codon frequencies), log-likelihood (l), and parameter estimates for fixed-sites models that distinguished internal and external segments (partition 1) from transmembrane segments (partition 2)...... 145 Table 4.5. Likelihood ratio test statistics (2Δ l), degrees of freedom (df), and p-values for comparisons of fixed-sites models...... 146

vii

List of Figures

Figure 2.1. Tree topologies representing alternative hypotheses about the relationship between the predominantly marine Thalassiosiraceae and predominantly freshwater Stephanodiscaceae...... 41 Figure 2.2. Phylogram from maximum parsimony analysis of the combined chloroplast (psbC + rbcL) dataset...... 43 Figure 2.3. Majority-rule consensus tree from Bayesian analysis of the combined chloroplast (psbC + rbcL) dataset...... 45 Figure 2.4. Phylogram from maximum parsimony analysis of the combined nuclear (SSU + partial LSU rDNA) dataset...... 47 Figure 2.5. Majority-rule consensus tree from Bayesian analysis of the combined nuclear (SSU + partial LSU rDNA) dataset...... 49 Figure 2.6. Phylogram from maximum parsimony analysis of the total combined (nuclear + chloroplast) dataset...... 51 Figure 2.7. Majority-rule consensus tree from Bayesian analysis of the total combined (nuclear + chloroplast) dataset...... 53 Figure 2.8. Maximum likelihood (ML) character mapping and ancestral state reconstruction of habitat type in Thalassiosirales...... 55 Figure 3.1. Maximum likelihood mapping of habitat type and binary codings for five convergent morphological characters in the two independently derived Stephanodiscaceae lineages...... 99 Figure 3.2. Scanning electron micrographs of radial branches in Thalassiosirales, highlighting the heavily silicified "costae" in Stephanodiscaceae...... 101 Figure 3.3. Scanning electron micrographs of marginal chambers in Thalassiosirales, highlighting the central and marginal laminae in Stephanodiscaceae...... 103 Figure 3.4. Scanning electron micrographs showing ultrastructure of strutted process "cowlings" in Thalassiosirales, highlighting the prominent cowlings in Stephanodiscaceae...... 104 Figure 3.5. Scanning electron micrographs showing ultrastructure of strutted process satellite pore covers in Thalassiosirales, highlighting the broad, ridgelike cowlings in Stephanodiscaceae...... 105 Figure 4.1. Gene phylogeny of silicon transporter genes (SITs) from Thalassiosirales...... 137 Figure 4.2. Cartoon representing the secondary structure of the last seven transmembrane segments of silicon transporters (SITs) in Thalassiosirales...... 138

viii

List of Figures (continued)

Figures 5.1–5.6 Cell wall morphology of Thalassiosira ritscheri with light microscopy (5.1 5.2) and field-emission scanning electron microscopy (5.3–5.6)...... 168 Figure 5.7–5.12 Scanning electron micrographs of internal (5.7, 5.9, 5.11) and external (5.8, 5.10, 5.12) features Thalassiosira ritscheri...... 170 Figures 5.13, 5.14 Scanning electron micrographs of Thalassiosira ritscheri, showing location of the labiate process (5.13) and its relation to the first girdle band opening (5.14)...... 171 Appendix 1 Strict consensus of nine most parsimonious trees from maximum parsimony analysis of the combined chloroplast dataset...... 173 Appendix 2 Strict consensus of 26 most parsimonious trees from maximum parsimony analysis of the combined nuclear dataset...... 174 Appendix 3 Strict consensus of two most parsimonious trees from maximum parsimony analysis of the total combined (chloroplast+nuclear) dataset.. 175

ix

Chapter 1. Introduction

Diatoms are a lineage of unicellular photoautotrophs within algae,

known most notably for their resilient and beautifully ornamented cell walls of opaline

silica. Given sufficient moisture and sunlight, diatoms can thrive in virtually any habitat, though they are principally found in association with submerged surfaces (benthic) or suspended in the water column of lakes and oceans (planktonic). Salinity imposes a significant barrier to the distribution of many organisms, including diatoms (Mann, 1999;

Round and Sims, 1980). Nevertheless, diatoms are a principle component of both marine and freshwater ecosystems. Unlike marine diatoms, freshwater diatoms must manage the steep difference in water potential between the inside of the cell and the surrounding medium, which presumably requires substantial energy expenditure. Reports of contractile vacuoles in diatoms are sparse and unsubstantiated (Fritsch, 1935), so it is

unclear how they manage this osmotic stress. It is generally understood that diatoms are ancestrally marine, so the number of times diatoms have independently colonized freshwater and the physiological adaptations that facilitated these transitions remain outstanding questions in diatom evolution (Mann, 1999; Round and Sims, 1980;

Strelnikova and Lastivka, 1999). Resolving the exact number of transitions between

marine and freshwater habitats will require a large-scale reconstruction of the diatom

phylogeny with dense sampling of marine and freshwater taxa. An alternative and

1

perhaps more tractable approach to understanding the marine–freshwater barrier is to study a group of diatoms with high diversity in each habitat.

The centric diatom order Thalassiosirales affords an excellent opportunity to study the origin and evolution of diatoms in freshwater. Thalassiosirales is a well- supported monophyletic group common in marine, brackish, and freshwater habitats. The roughly 500 described species are divided between the marine Thalassiosiraceae and freshwater Stephanodiscaceae (Glezer and Makarova, 1986). Citing the paucity of genera spanning both habitats, Round and Sims (1980) proposed a general evolutionary scenario consistent with the Thalassiosiraceae–Stephanodiscaceae classification. The research presented in this dissertation uses Thalassiosirales as a model system to test such hypotheses, and to address a number of other problems related to the marine–freshwater boundary in diatoms.

A robust, densely sampled phylogeny is a prerequisite to testing freshwater colonization hypotheses. Thus, the phylogeny and systematics of extant Thalassiosirales are the subjects of Chapter 2. DNA sequence data were collected for 83 taxa, including

79 Thalassiosirales and four outgroup taxa. Taxon sampling was limited primarily by culture availability: 37 cultures came from public culture collections, and the remaining

46 cultures were isolated specifically for this study. Most of the culture strains isolated for this study were species that had never been cultured, and many will be made publicly available through the Provasoli–Guillard National Center for Culture of Marine

Phytoplankton. Every effort was made to sample the range of extant morphological diversity evenly and adequately, and the 79 taxa used for phylogeny reconstruction represent a fair cross-section of this diversity. Four DNA markers were used to

2

reconstruct the phylogeny: nuclear small subunit rDNA (~1750 nt), the first and second divergent domains of nuclear large subunit rDNA (~600 nt), chloroplast rbcL (1473 nt), and chloroplast psbC (~1250 nt). The phylogeny was used to test previous freshwater colonization hypotheses and to determine the total number and directions of marine– freshwater transitions. This project represents the most ambitious attempt to date to reconstruct the phylogeny of any group of diatoms, in terms of both number of characters and number of ingroup taxa. Phylogenetic hypotheses from Chapter 2 provided the basis for much of the research presented in subsequent chapters.

Phylogenetic hypotheses from Chapter 2 revealed at least three independent colonizations of freshwater, one of which was unexpected and is the subject of Chapter 3.

The freshwater Stephanodiscaceae (≥ 200 species) share a number of morphological characteristics, the complexity of which is such that no one has considered that they could have evolved more than once. The lacustrine fossil record suggests that

Stephanodiscaceae traces back to a single successful colonization of freshwater in the middle Miocene (Krebs, 1990). Phylogenetic analyses showed, however, that part of the

Stephanodiscaceae is an early diverging lineage within the order Thalassiosirales that colonized and diversified in freshwater much earlier than the independent, Miocene colonization by the rest of the family. Moreover, species in these two distantly related

Stephanodiscaceae lineages independently evolved a similar suite of complex morphological characters upon or shortly after colonizing freshwater. The purpose of this chapter is to thoroughly document the convergent morphological evolution in these two freshwater lineages, and to use comparative methods to determine if these characters are significantly correlated with the freshwater habitat.

3

The diatom cell wall is composed of hydrated amorphous silica, and at least one

potential advantage of freshwater habitats is the surplus of silicic acid compared to

oceans (silicic acid is the dissolved form of silicon used by diatoms). On average, freshwater diatoms contain an order of magnitude more silica in their cell walls than marine diatoms (Conley et al., 1989), and culture experiments have shown that freshwater diatoms exposed to high salinity media grow faster (cell divisions/day) and contain less silica in their cell walls than normal (Tuchman et al., 1984). In addition, silicon-dependent growth has been measured for several diatoms, and results suggest that marine taxa have a substantially greater affinity for silicic acid than freshwater taxa

(Martin-Jezequel et al., 2000; Paasche, 1980). A small gene family that encodes a novel silicic-acid transporter (SIT) is responsible for importing silicic-acid from the environment into the diatom cell (Hildebrand et al., 1998; Hildebrand et al., 1997). SIT fragments were cloned and sequenced from 59 species of Thalassiosirales, and the phylogeny and molecular evolution of the SIT gene family in Thalassiosirales are the subjects of Chapter 4.

Chapter 5 is the product of a major effort to collect and culture Thalassiosira

species from the Antarctic for the phylogenetic analysis. Although somewhat of a

departure from the previous three chapters, as DNA sequencing technology advances, the

descriptive work in this chapter will most likely outlast the rest of this dissertation.

Within Thalassiosirales, Thalassiosira is arguably the most taxonomically and

morphologically diverse genus. At least two distinct morphological groups have been

recognized within Thalassiosira (Hasle, 1968; Hasle and Syvertsen, 1997). Group "A"

Thalassiosira species, which include the type species, T. nordenskioeldii, have short

4

inward and long outward extensions of the strutted processes, and a labiate process on the

valve mantle. Group "B" species have exceptionally long inward and reduced outward

extensions of the strutted processes, and a labiate process on the valve face. I collected

and cultured Thalassiosira ritscheri, which has a combination of group A and B characters. It has a labiate process on the valve face and reduced outward extensions of the strutted processes. I show for the first time that T. ritscheri has short inward, A-type

extensions of the strutted processes. A phylogenetic interpretation of these conditions

suggests a close relationship between T. ritscheri and the traditionally held group B

species. Species diagnosed by the autapomorphic condition of a labiate process away

from the valve mantle, including many group "B" Thalassiosira species, are transferred into Shionodiscus gen. nov.

Literature cited

Conley, D. J., S. S. Kilham, and E. Theriot. 1989. Differences in silica content between

marine and freshwater diatoms. Limnology & Oceanography 34:205–213.

Fritsch, F. E. 1935. The Structure and Reproduction of the Algae. Cambridge University

Press, London.

Glezer, Z. I., and I. V. Makarova. 1986. New order and family of diatoms

(Bacillariophyta). Botanicheskii Zhurnal (Leningrad) 71:673–676.

Hasle, G. R. 1968. The valve processes of the centric diatom genus Thalassiosira. Nytt

Magasin for Botanikk 15:193–201.

Hasle, G. R., and E. E. Syvertsen. 1997. Marine Diatoms. Pages 5–386 in Identifying

Marine (C. R. Tomas, ed.) Academic Press, San Diego.

5

Hildebrand, M., K. Dahlin, and B. E. Volcani. 1998. Characterization of a silicon

transporter gene family in Cylindrotheca fusiformis: sequences, expression

analysis, and identification of homologs in other diatoms. Molecular and General

Genetics 260:480–486.

Hildebrand, M., B. E. Volcani, W. Gassmann, and J. I. Schroeder. 1997. A gene family of

silicon transporters. Nature 385:688–689.

Krebs, W. N. 1990. The biochronology of freshwater planktonic diatom communities in

western North America. Pages 485–499 in Proceedings of the 11th International

Diatom Symposium (J. P. Kociolek, ed.) California Academy of Sciences, San

Francisco.

Mann, D. G. 1999. Crossing the Rubicon: the effectiveness of the marine/freshwater

interface as a barrier to the migration of diatom germplasm. Pages 1–21 in

Proceedings of the 14th International Diatom Symposium (S. Mayama, M. Idei,

and I. Koizumi, eds.). Koeltz Scientific Books, Koenigstein.

Martin-Jezequel, V., M. Hildebrand, and M. A. Brzezinski. 2000. Silicon metabolism in

diatoms: Implications for growth. Journal of Phycology 36:821–840.

Paasche, E. 1980. Silicon. Pages 259–284 in The Physiological Ecology of

Phytoplankton (I. Morris, ed.) University of California Press, Berkeley.

Round, F. E., and P. A. Sims. 1980. The distribution of diatom genera in marine and

freshwater environments and some evolutionary considerations. Pages 301–320 in

Proceedings of the Sixth Symposium on Recent and Fossil Diatoms (R. Ross, ed.)

Otto Koeltz Science Publishers, Hirschberg.

6

Strelnikova, N. I., and T. V. Lastivka. 1999. The problem of the origin of marine and

freshwater diatoms. Pages 113–123 in Proceedings of the 14th International

Diatom Symposium (S. Mayama, M. Idei, and I. Koizumi, eds.). Koeltz Scientific

Books, Koenigstein.

Tuchman, M. L., E. C. Theriot, and E. F. Stoermer. 1984. Effects of low level salinity

concentrations on the growth of meneghiniana Kütz. (Bacillariophyta).

Archiv Fuer Protistenkunde 128:319–326.

7

Chapter 2: Stampeding the Rubicon: phylogenetic analysis reveals repeated

colonizations of marine and fresh waters by thalassiosiroid diatoms

Abstract. Salinity imposes a significant barrier to the distribution of many organisms, including diatoms. Diatoms are ancestrally marine, and the number of times they have independently colonized fresh waters and the physiological adaptations that facilitated these transitions remain outstanding questions in diatom evolution. The colonization of fresh waters has been compared to "crossing the Rubicon," suggesting that successful colonization events are rare, irreversible, and lead to substantial species diversification.

To test these hypotheses, we reconstructed the phylogeny of Thalassiosirales, a diatom lineage with high diversity in both marine and fresh waters. We collected a total of ~5.3 kb of DNA sequence data from the nuclear (SSU and partial LSU rDNA) and chloroplast genomes (psbC and rbcL) and reconstructed the phylogeny using parsimony and

Bayesian methods. Alternative topology tests strongly reject all previous colonization hypotheses, including monophyly of the predominantly freshwater Stephanodiscaceae.

Results show at least three independent colonizations of fresh waters, and whereas previous accounts of freshwater-to-marine transitions are usually discounted, our results provide compelling evidence for as many as three independent re-colonizations of the marine habitat, two of which led to subsequent speciation events. This study adds valuable phylogenetic context to previous debate about the nature of the salinity barrier in

8 diatoms and provides compelling evidence that, at least for Thalassiosirales, the salinity barrier might be less formidable than previously thought.

Introduction

Diatoms are a lineage of unicellular photoautotrophs within heterokont algae, known most notably for their intricately ornamented cell walls of opaline silica

(Andersen, 2004; Round et al., 1990). Given sufficient moisture and sunlight, diatoms can thrive in virtually any habitat, though they are principally found in association with submerged surfaces (benthic) or suspended in the water column of rivers, lakes, and oceans (planktonic). The importance of marine planktonic diatoms in the biogeochemical cycling of many elements, particularly carbon and silica, cannot be overstated. As a whole, marine phytoplankton account for >46% of global net primary production, despite representing <1% of global primary producer biomass (Behrenfeld and Falkowski, 1997;

Falkowski et al., 1998; Field et al., 1998). Diatoms alone account for ≥40% of marine primary production, which translates to roughly one-quarter of global net primary production (Nelson et al., 1995; Werner, 1977). Silica comprises ~25% (by weight) of the earth’s crust (Iler, 1979), and diatoms are the primary biological mediators of the silica cycle in oceans, where the element is relatively scarce and has been since the marked diversification and ecological expansion of diatoms beginning in the early Cenozoic

(Katz et al., 2004; Martin-Jezequel et al., 2000; Nelson et al., 1995; Treguer et al., 1995).

Addition of salts causes significant changes to the chemical–physical properties of water, including increased density and osmotic pressure and decreased freezing point and vapor pressure (Dawes, 1998). As a result, salinity imposes a significant barrier to the

9 distribution of many organisms, including diatoms (Mann, 1999; Round and Sims, 1980).

Unlike marine diatoms, freshwater diatoms must manage the steep difference in water potential between the inside of the cell and the surrounding medium, which presumably requires substantial energy expenditure. Reports of contractile vacuoles in diatoms are sparse and unsubstantiated (Fritsch, 1935), so it is unclear how they overcome this osmotic stress. It is generally understood, based on the fossil record, that diatoms are ancestrally marine, so the number of times diatoms have independently colonized fresh

waters and the physiological adaptations that facilitated these transitions remain

outstanding questions in diatom evolution (Mann, 1999; Round and Sims, 1980;

Strelnikova and Lastivka, 1999). Previous efforts to understand the relationship between marine and freshwater diatoms did not fully incorporate phylogeny, relying more on patterns inferred from an explicitly phenetic diatom classification system (Mann, 1999;

Round et al., 1990; Round and Sims, 1980). For example, one generally held hypothesis is that diatoms are naturally split along marine–freshwater lines, based on the observation that most genera are restricted to either marine or fresh waters—few genera have representatives in both habitats (Round and Sims, 1980). The colonization of fresh waters by diatoms has been compared to "crossing the Rubicon," implying that successful colonization events are—in addition to their historic proportion—rare and irreversible

(Mann, 1999). Examples of "leakage" (Round and Sims, 1980) of marine taxa into fresh waters are common and indisputable (e.g. Edlund et al., 2000; Weber, 1970), whereas accounts of freshwater taxa colonizing marine waters are usually either discounted

(Mann, 1999) or dismissed outright (Round and Sims, 1980). Thus, the salinity barrier in diatoms is thought to be crossed from one direction only (Mann, 1999; Round and Sims,

10 1980). Resolving the exact number and directions of marine–freshwater transitions will

require a large-scale reconstruction of the diatom phylogeny with dense sampling of

marine and freshwater taxa. An alternative, more tractable approach is to focus on a single lineage with high diversity in each habitat. Among these, the thalassiosiroid lineage provides an excellent opportunity to study the origin and evolution of diatoms in

fresh waters.

The family Thalassiosiraceae originally was described by Lebour (1930) and later

emended by Hasle (1973) to include all diatoms that possess a fultoportula (strutted

process), a feature now recognized as a synapomorphy for the lineage (Theriot and

Serieyssol, 1994). The strutted process consists of a central tube surrounded by 2–6 pores

(satellite pores) inwardly, and its outward structure varies from a simple pore to long,

elaborate extensions that connect sibling cells. Strutted processes are the site of β-chitin

synthesis (Herth, 1978; Herth, 1979; Herth and Barthlott, 1979), which among diatoms, is

unique to thalassiosiroids. β-chitin threads extruded through strutted processes on the

valve face facilitate chain formation, whereas threads extruded through strutted processes

on the valve margin are thought to enhance buoyancy in the water column (Walsby and

Xypolyta, 1977). Thalassiosiroid diatoms are common in the plankton of large rivers and

reservoirs, freshwater lakes, brackish and estuarine waters, coastal marine and open

ocean habitats (Hasle and Syvertsen, 1997; Stoermer and Julius, 2003). Citing this

distribution, Thalassiosiraceae was raised to ordinal status (Thalassiosirales) and genera were classified into marine Thalassiosiraceae or freshwater Stephanodiscaceae (Glezer and Makarova, 1986). Thalassiosirales dates to at least the late Cretaceous (Hasle and

Syvertsen, 1985; Medlin et al., 1996), and while the marine Thalassiosiraceae is probably

11 paraphyletic (Glezer and Makarova, 1986; Medlin et al., 1996), evidence from the

lacustrine fossil record and morphology suggests that Stephanodiscaceae is monophyletic, tracing to a single, successful colonization of fresh waters in the middle

Miocene (Glezer and Makarova, 1986; Krebs, 1990; Krebs and Bradbury, 1982; Krebs et

al., 1987; Theriot and Serieyssol, 1994).

Thalassiosirales is a large and phylogenetically diverse lineage within centric

diatoms. It is nominally diverse, with more than 1000 names at the rank of species and

below (according to the computerized database of verified diatom names at the California

Academy of Sciences, E. Fourtanier and J.P. Kociolek, pers. comm.). The nominal

diversity undoubtedly exceeds species diversity (Spamer and Theriot, 1997), so 500 fossil

and living species is probably a conservative estimate of the total species diversity.

Species are distributed among roughly 30 genera, of which ~10 are monotypic and ~10 contain fewer than 10 species. The generic classification is largely phenetic, so most genera are not easily distinguished from one another, or particularly, from the larger genera Cyclotella and Thalassiosira. No single character or combination of characters can be considered derived and therefore indicative of monophyly for either genus.

Illustrative of this, considerable effort has been made to document the range of morphological diversity in both Cyclotella (Loginova, 1990; Lowe, 1975; Serieyssol,

1980) and Thalassiosira (Fryxell and Hasle, 1977; Fryxell and Hasle, 1979a; Fryxell and

Hasle, 1979b; Hasle, 1968; Hasle, 1978a; Hasle, 1978b; Tanimura, 1996), and to identify phenetic clusters of species within them. Although several of these phenetic groups have been split into new genera (e.g. Discostella Houk & Klee), most new names have yet to find wide acceptance among diatomists, at least in practice. Because of this, because

12 many of these groups have no apparent apomorphy, and mostly because it is beyond the scope of this study to address the many issues surrounding Thalassiosirales classification, we retain the names Cyclotella and Thalassiosira as have been historically applied.

The primary goal of this research was to reconstruct the phylogeny of extant

Thalassiosirales with broad sampling across the entire lineage and to use the phylogeny to understand the history and pattern of freshwater colonization. We expanded the taxonomic and character sampling of previous studies (Julius, 2000; Julius and Tanimura,

2001; Medlin et al., 1996) to include 78 accessions, representing 10 genera from marine

Thalassiosiraceae and freshwater Stephanodiscaceae. Two nuclear and two chloroplast

DNA markers were used to reconstruct the phylogeny. Datasets from the two genetic compartments were analyzed individually and in combination using parsimony and

Bayesian methods. Phylogenetic hypotheses were used to determine the number and direction of freshwater colonization events and to test previous hypotheses about the relationship between marine and freshwater taxa. This study adds valuable phylogenetic context to previous debate about the nature of the salinity barrier in diatoms and provides compelling evidence that, at least for Thalassiosirales, the salinity barrier is less formidable than previously thought.

Materials and Methods

Taxon sampling

Two nuclear (SSU rDNA and partial LSU rDNA) and two chloroplast (rbcL and partial psbC) markers were sequenced for 82 accessions, including 78 Thalassiosirales and four outgroup taxa (Tables 2.1, 2.2). Every effort was made to sample the range of extant morphological diversity evenly and adequately, using monographs and previous

13 phylogenetic studies as guides. Forty-four taxa representing seven genera were sampled

from the predominantly marine Thalassiosiraceae, including at least one representative

from each of the major subgroups within Thalassiosira (Fryxell and Hasle, 1977; Fryxell and Hasle, 1979b; Hasle, 1968; Hasle, 1978b; Hasle and Fryxell, 1977; Hasle and Lange,

1989). A previous phylogenetic analysis of Stephanodiscaceae found extant species distributed among six crown clades (Julius, 2000), and our analyses include at least one representative from each of these six clades. The four outgroup taxa were chosen because they consistently comprise the sister clade to Thalassiosirales in SSU rDNA phylogenies of diatoms (Alverson and Theriot, 2005 for review). Three sequences are missing from the final matrix: Cyclotella stylorum SSU rDNA, Thalassiosira mediterranea SSU rDNA, and Helicotheca tamesis LSU rDNA; SSU rDNA sequences downloaded from GenBank were used for Helicotheca tamesis (X85385) and Ditylum brightwellii (X85386) (Table

2).

Cell culture and DNA methods

Cells were grown in batch culture at 14o C on a 12:12 light-dark cycle. Marine

species were grown in f/2 medium (Guillard, 1975; Guillard and Ryther, 1962), and

freshwater species were grown in COMBO medium (Kilham et al., 1998). Cells were

harvested early in stationary phase by concentrating them into a pellet through a series of centrifugations. Cell pellets were stored in 1.5-mL tubes at –80oC until DNA extraction

commenced. Frozen cells were broken with stainless steel beads as tubes were shaken for

2 minutes at 30 Hz in a Brinkman Retsch Mixer Mill MM300 (Westbury, NY, USA).

Total nucleic acids were extracted with the DNeasy® Plant Mini Kit (Qiagen, Hilden,

Germany). Cyclotella distinguenda did not survive in culture, so total DNA was extracted

14 from an environmental sample where it represented >95% relative abundance and where

no other thalassiosiroid species were observed; gene fragments were subsequently PCR

amplified and cloned from this environmental DNA preparation. Partial fragments of SSU

rDNA (~1750 nt), partial LSU (~600 nt), rbcL (~1470 nt), and psbC (~1100 nt) were

amplified by PCR (see Table 3 for primer sequences). The volume of each PCR was

50µL: 1.0 µL purified DNA; 5 µL 10× buffer; 2.5mM MgCl2 (5µL of 25mM stock); 0.2 mM each dATP, dCTP, dGTP, dTTP (1 µL each of 10 mM stocks); 0.1 µM each primer

(0.25 µL each of 20 µM stocks); 0.2 units Taq polymerase; and ddH2O to a final volume

of 50 µL. PCR conditions for SSU followed Alverson and Kolnick (2005). PCR conditions for LSU d1–d2 and partial rbcL (primers nd6+ and dp7) were as follows: 94o

C for 3:30, 35 cycles of (94o C for 50 s, 53o C for 50 s, 72o C for 60 s), and final extension

at 72o C for 7 m. PCR conditions for psbC and partial rbcL (primers rbcL66+ and

rbcL1255–) were the same but with 1:10 m extension time. When necessary, PCR

products were cloned as described in Alverson and Kolnick (2005). PCR products were

purified with QIAquick® PCR Purification columns (QIAGEN, Hilden, Germany).

Forward and reverse strands were cycle sequenced with BigDye (Applied Biosystems,

Foster City, CA, USA) using a combination of nested primers (Table 2). Sequences were

resolved with a MJ Research BaseStation DNA Fragment Analyzer (MJ Research, Inc.,

San Francisco, CA, USA) or ABI 3700 DNA Analyzer (Applied Biosystems, Foster City,

CA, USA). Sequence chromatograms were edited and assembled into contigs with

Sequencher ver. 3.0 (Gene Codes Corporation, Ann Arbor, MI, USA).

Multiple sequence alignment

15 Sequences for the protein coding genes psbC and rbcL lacked indels and were

aligned manually with MacClade ver. 4.08 (Maddison and Maddison, 2003). SSU rDNA

sequences were aligned to maximize the juxtapositioning of similar primary and

secondary structure of the 18S rRNA molecule, using a template of 181 aligned diatom

SSU rDNA sequences (Alverson et al., 2006). This alignment allowed us to distinguish

between sites that were paired or unpaired in the 18S rRNA molecule for use in

phylogenetic analyses. Partial LSU rDNA sequences initially were aligned with Clustal X using the default settings for gap-opening and gap-extension. Highly variable regions

were re-aligned with Clustal X using a variety of gap-opening/gap-extension penalties,

followed by manual adjustments to minimize the number of variable sites.

Phylogeny estimation

Parsimony analyses—Maximum parsimony analyses of the chloroplast encoded psbC

and rbcL datasets were run individually and in combination. Tree searches were done

with the "new technology" search algorithm implemented in TNT (Goloboff et al., 2004).

Five hundred random addition sequence replicates were performed, with all cycles,

rounds, and repetitions increased by an order of magnitude beyond default values.

Nonparametric bootstrap analyses (10,000 pseudoreplicates) were done to assess branch support. The "traditional" search algorithm in TNT was used for each pseudoreplicate of the bootstrap analysis. These analyses were repeated for the nuclear encoded LSU and

SSU datasets, run individually and in combination, as well as the total (chloroplast + nuclear) combined datasets.

Bayesian analyses—MrModeltest (Nylander, J.A.A. 2004. MrModeltest v2. Program distributed by the author. Evolutionary Biology Centre, Uppsala University) was used to

16 determine the most appropriate model of DNA sequence evolution for each gene

partition. The posterior probability distribution was estimated using Metropolis-Coupled

Markov Chain Monte Carlo (MCMCMC) as implemented in MrBayes (ver. 3.1—

Ronquist and Huelsenbeck, 2003). Default priors were used for all analyses. Each run

used four chains, one cold and three heated, with temperature and proposal parameters adjusted as necessary to facilitate convergence and mixing of the chains. We assessed stationarity by tracking posterior probabilities of tree splits using the cump and slide commands in AWTY (Wilgenbusch, J.C., Warren, D.L. & Swofford, D.L. 2004. AWTY:

A system for graphical exploration of MCMC convergence in Bayesian phylogenetic

inference. http://king2.csit.fsu.edu:16080/CEBProjects/awty/awty_start.php). Results of sliding window analyses (slide) were generally more erratic than point estimates of cumulative split posterior probabilities (cump), so stationarity was assumed when all

splits appeared to have stabilized by one (cump) or both of these measures. We

performed at least two independent runs per dataset and confirmed that the independent

runs had sampled the same posterior distribution by comparing split posterior

probabilities with the compare command in AWTY before samples from the stationary

phase of independent runs were pooled. The posterior distribution of tree topologies and

branch lengths were summarized with the sumt command in MrBayes. Details for

specific analyses are summarized in Table 4.

Several model partitioning strategies were applied to the individual and combined

datasets (Table 4). Two models were applied to the chloroplast dataset. We first applied

the GTR + G + I to the combined chloroplast dataset, essentially treating the two genes as

a single gene partition. Site-specific rate models that consider codon position often

17 provide a better fit to protein coding datasets than do standard nucleotide models

(Shapiro et al., 2006), so we also applied a single GTR + G + I to combined first and second codon positions and a single GTR + G + I model to third positions ("CP112" sensu Shapiro et al., 2006). For the nuclear SSU rDNA dataset, we first applied a single

GTR + G + I to all positions. One strategy to overcome nonindependence of nucleotide substitutions in rRNA helices is use of "doublet" models that consider pairs of sites rather than considering them individually (Schoniger and Von Haeseler, 1994), so a second SSU

rDNA model applied Doublet + G + I to paired sites and GTR + G + I to unpaired sites.

For the combined nuclear dataset, we first applied a single GTR + G + I model, treating

the two genes as a single gene partition. A second model applied Doublet + G + I to paired SSU rDNA sites, GTR + G + I to unpaired SSU rDNA sites, and GTR + G + I to the LSU rDNA sites. The harmonic mean of the likelihood values sampled during the

MCMC analysis was calculated for each partitioning scheme with the sump command in

MrBayes, and relative support for the different models was assessed using Bayes factor comparisons (Kass and Raftery, 1995; Nylander et al., 2004 for phylogenetic applications). Model choice for the combined chloroplast + nuclear analysis was based on the results of these Bayes factor comparisons.

Hypothesis testing

Two hypotheses were tested, the first of which constrained Stephanodiscaceae to be monophyletic, consistent with the hypothesis of a single, primary colonization of freshwater (Fig. 2.1a—Krebs, 1990; Krebs et al., 1987; Theriot and Serieyssol, 1994).

Two variations of this hypothesis were tested, one of which included T. pseudonana

(synonym = Cyclotella nana Hustedt) in Stephanodiscaceae and one of which did not.

18 The second hypothesis enforced reciprocal monophyly of Thalassiosiraceae and

Stephanodiscaceae (+ T. pseudonana), consistent with the relationships proposed by

Round and Sims (1980) and Glezer and Makarova (1986) (Fig. 2.1b). Hypothesis testing

was carried out on the combined chloroplast, combined nuclear, and total combined

(chloroplast + nuclear) datasets. For Bayesian hypothesis testing, we constructed a 95%

credible set of trees from those sampled from the MCMC analyses. If tree topologies

compatible with a particular freshwater colonization hypothesis were absent from the

95% credible set of trees, the hypothesis could be rejected, and if present, the cumulative

posterior probability of that topology was calculated (Buckley, 2002; Reeder, 2003).

Second, we used the SOWH parametric bootstrap test (Goldman et al., 2000; Hillis et al.,

1996; Huelsenbeck et al., 1996), with 1000 data matrices simulated on the constrained

(null) tree topology according to a single null model (GTR + G + I in all cases, Table 4)

and parameter values derived from the original dataset. The SOWH test suffers from high

Type I error rate when the simulation model deviates too strongly from the true underlying model (Buckley, 2002; Huelsenbeck et al., 1996). Although no formal

correction exists, use of the parsimony criterion to generate the test statistic and null

distribution might offset the high Type I error rate by decoupling the inference model

(parsimony) from the simulation model (GTR + G + I) (Buckley, 2002), so all parametric

bootstrap tests were based on the parsimony criterion. Data simulations and analysis of results were done using the "Batch Architect" feature in the Mesquite software package

(Maddison and Maddison, 2006).

Reconstructing the history of freshwater colonization

19 All taxa were coded as marine (0) or freshwater (1), and ancestral states were

estimated using maximum parsimony and maximum likelihood. Maximum likelihood reconstructions (Pagel, 1999; Schluter et al., 1997) were based on the Asymmetrical

Markov k-state 2 parameter model, which allows for different forward (0Æ1) and

backward (1Æ0) rates, representing the rates of freshwater and marine colonization

events, respectively. All model parameters were estimated from the data, and the

ancestral reconstruction at any given node was considered significantly better than the

alternative if its likelihood score exceeded the alternative reconstruction by two log-

likelihood units (Pagel, 1999). Maximum likelihood reconstruction was carried out using

the combined Bayesian topology and branch lengths but with 18 taxa on zero or near-zero

branches pruned from the tree. All ancestral state reconstructions were done using the

Mesquite software package (Maddison and Maddison, 2006).

Results

The combined chloroplast (psbC + rbcL) dataset contained 2725 columns, of

which 1252 columns were psbC and 1473 were rbcL (Table 4). Of the 704 parsimony informative sites in the chloroplast dataset, 358 of these were in the psbC gene and 346 in the rbcL gene (Table 4). The combined nuclear (SSU + LSU rDNA) dataset contained

2615 columns, of which 2034 columns were SSU rDNA and 581 were LSU rDNA. Of the

598 parsimony informative sites in the nuclear dataset, 372 of these were in the SSU rDNA gene and 226 in the partial LSU rDNA gene. A total of 27 columns from the 5' and

3' ends of the psbC fragments were excluded from phylogenetic analyses, reflecting differences in the total sequence length recovered from different taxa, not natural length variation. A total of 99 columns from the SSU rDNA alignment were excluded from

20 phylogenetic analyses, most of which were empty and kept only to maintain the base pairings identified in a larger diatom alignment (Alverson et al., 2006). A total of 30 ambiguously aligned columns from the LSU rDNA alignment also were excluded from phylogenetic analyses.

Bayesian analyses and model choice

Bayesian MCMC analyses were run until all bipartition posterior probability

(BPP) values across independent runs differed by less than 0.05, which required 10–80 million generations (Table 4). Comparisons of BPP values across runs indicated that, for the combined chloroplast (single GTR + G + I model) and the total combined analyses, one or more of four independent runs had not reached stationarity after 80 and 40 million generations, respectively, and so were discarded (Table 4).

The Akaike Information Criterion favored the GTR + G + I model of sequence evolution for all gene partitions, and all genes were run individually under this model

(not shown). The protein-coding chloroplast genes psbC and rbcL were combined into a single partition and analyzed under a single GTR + G + I nucleotide model (Table 4).

Bayes factor comparison, however, strongly favored the alternative, codon-based CP112 model, which improved the fit over the single GTR + G + I nucleotide model by nearly

1100 log-likelihood units (Table 4). The nuclear SSU rDNA gene was analyzed with a single GTR + G + I model as well as a partitioned model that applied a Doublet + G + I model to paired nucleotides in stems and a standard GTR + G + I to unpaired nucleotides in loops and bulges. The partitioned model improved the fit by 843 log-likelihood units and was strongly favored by the Bayes factor comparison (Table 4). The combined nuclear dataset (SSU + LSU rDNA) was analyzed with the favored SSU model and a

21 standard GTR + G + I model for the LSU partition; this partitioned model provided an improvement of 339 log-liklihood units over a single GTR + G + I model and was also strongly favored by the Bayes factor comparison (Table 4). Based on these results, the total combined (nuclear + chloroplast) dataset was analyzed with the CP112 model applied to the chloroplast partition, Doublet + G + I to SSU rRNA stems, GTR + G + I to

SSU rRNA loops and bulges, and GTR + G + I to the LSU rDNA region (Table 4).

Phylogenetic results

Parsimony analysis of the combined chloroplast dataset recovered nine most parsimonious trees of 4130 steps (Table 4). Although relationships at the tips of the tree were strongly supported, the majority of deeper nodes were weakly supported (Fig. 2.2) and collapsed in the strict consensus (Appendix 1). Results from the Bayesian CP112 analysis of the combined chloroplast data were similar to parsimony results (Figs. 2.2,

2.3). Topological conflicts between parsimony and Bayesian topologies concerned weakly supported nodes in both analyses, with the notable exception of the placement of

T. oceanica, which was an early diverging branch in the parsimony tree and in a more derived position among and other Thalassiosira species in the Bayesian topology (Figs. 2.2, 2.3). Chloroplast data support monophyly of Bacterosira,

Skeletonema, Cyclostephanos, and Stephanodiscus. Porosira was non-monophyletic, and multiple, sometimes distantly related, lineages within Thalassiosira were recovered, some of which included the smaller genera Bacterosira, Detonula, Minidiscus, and

Skeletonema. The basal position of Clade A (Lauderia + Porosira) was well-supported in each of the two analyses, and Stephanodiscaceae was divided into two distinct, well- supported clades (Figs. 2.2, 2.3; Clades B and C).

22 Unlike the chloroplast dataset, the combined nuclear dataset showed more

divergence and better support for deeper nodes than for the tips of the tree (Figs. 2.4,

2.5). Parsimony analysis resulted in 26 most parsimonious trees of 2621 steps. Parsimony

and Bayesian topologies were nearly identical, with minor conflicts lacking support in

one or both of the trees (Figs. 2.4, 2.5) or collapsing in the parsimony strict consensus

tree (Appendix 2). Use of a doublet model for paired sites in the SSU rDNA dataset

effectively reduced the number of characters by nearly 500, which might account for the

lower overall divergence observed in the Bayesian tree. Both analyses support monophyly of Bacterosira, Skeletonema, Cyclostephanos, and Stephanodiscus. In addition, nuclear data strongly support a basal position for Clade A and separation of

Stephanodicaceae into two distinct, distantly related clades, B and C (Figs. 2.4, 2.5;

Clades B and C).

Chloroplast and nuclear data were complementary in their resolution and support

for different parts of the tree, with chloroplast data resolving closer, species relationships

and nuclear data resolving the overall backbone of the tree (Figs. 2.2–2.5). The two

datasets were combined for parsimony and Bayesian analysis, and as might be expected,

resolution and support for deeper nodes reflected the nuclear topologies, whereas

shallower nodes tended toward chloroplast topologies (Figs. 2.6, 2.7). Parsimony and

Bayesian analyses of the combined data resulted in similar topologies, with identical bipartitions tending to have higher BPP than bootstrap support (Figs. 2.6, 2.7). With only

one node collapsed, the parsimony strict consensus tree was nearly fully resolved

(Appendix 3). The three notable conflicts between parsimony and Bayesian trees

concerned placement of Thalassiosira sp. (CCMP1065), T. angulata, and Bacterosira,

23 the positions of which were unsupported in one or both of the analyses (Figs. 2.6, 2.7).

The remaining discussion will focus solely on the Bayesian topology (Fig. 2.7). The

combined data support monophyly of Bacterosira, Skeletonema, Cyclostephanos, and

Stephanodiscus (BPP ≥ 0.95). Clade A is the earliest diverging lineage within

Thalassiosirales (BPP ≥ 0.95), and Stephanodiscaceae consists of two well-supported

clades (Clades B and C, each BPP ≥ 0.95) separated by nine bipartitions, seven of which had BPP ≥ 0.95 (Fig. 2.7). Clade B includes the type species of Cyclotella, Cy. distinguenda, and an assemblage of taxa informally referred to as the "Cy. menegheniana complex" (Beszteri et al., 2005). The morphologically simple, nondescript species,

Thalassiosira pseudonana (synonym = Cyclotella nana) is sister to Clade B (BPP ≥

0.95), a relationship that is supported by at least two strutted process characters (not shown). The type species of Thalassiosira, T. nordenskioeldii, is more closely related to

Detonula and Minidiscus than to most other species of Thalassiosira included in these analyses (Fig. 2.7). Finally, T. oestrupii var. venrickae is on a long branch, and most aspects of its morphology—areolar structure, internal and external strutted process ultrastructure, and position of the labiate process—indicate that this species is also highly derived morphologically (Fig. 2.7).

Hypothesis testing and habitat mapping

Tree topologies consistent with a priori freshwater colonization hypotheses (Fig.

2.1) were absent from all 95% credible sets of trees and so were rejected. Likewise,

SOWH parametric bootstrap tests rejected all a priori hypotheses (p < 0.001).

Parsimony mapping required seven steps to reconcile the distribution of marine and freshwater taxa on each of the parsimony and Bayesian tree topologies (Figs. 2.2–

24 2.7). In all cases, the most recent common ancestor of Thalassiosirales was unequivocally

marine, there were at least four independent colonizations of fresh waters, two of which

resulted in substantial diversification (Clades B and C). Parsimony mapping was

generally equivocal about the habitat type of the most recent common ancestor of Clade

B, whereas the ancestral state for Clade C was unequivocally freshwater (Figs. 2.2–2.7).

Parsimony mapping showed unequivocal evidence for marine-to-freshwater transitions,

as well as some evidence within Clade B for re-colonization of the marine habitat by

freshwater taxa (Figs. 2.2–2.7).

Habitat was also mapped onto the combined Bayesian tree (topology + branch

lengths) with maximum likelihood, allowing for different rates of forward

(marineÆfreshwater) and backward (freshwaterÆmarine) transitions (Fig. 2.8).

Likelihood reconstructions indicated that the ancestral habitat type of Thalassiosirales was marine (Fig. 2.8) and provided greater resolution over parsimony about the number and directions of marine–freshwater transitions. Likelihood mapping favors three independent colonizations of fresh waters in the lineages leading to T. gessneri, Clade B

+ T. pseudonana, and Clade C (Fig. 2.8). In addition, likelihood favored three independent re-colonizations of the marine habitat from freshwater ancestors, two of

which were statistically significant and led to subsequent speciation (Fig. 2.8; [Cy. striata

+ Cy. stylorum], [Cyclotella sp. L1844 + Cyclotella sp. MC01]).

Discussion

Phylogeny of Thalassiosirales

The character and taxonomic sampling underlying these phylogenies greatly expanded upon previous investigations of Thalassiosirales phylogeny based on single

25 genes or focused primarily on relationships within Thalassiosiraceae or

Stephanodiscaceae (Julius, 2000; Kaczmarska et al., 2006; Medlin et al., 1996), neither of

which was monophyletic in our analyses (Figs. 2.2–2.7). Nearly every phylogenetic

analysis of diatoms to date is based on a single nuclear rDNA marker (typically SSU or partial LSU rDNA—see Alverson and Theriot, 2005 for review). Together, these two markers provided resolution and support for deeper nodes in our trees but were less informative at lower levels (Figs. 2.4, 2.5), whereas the chloroplast data—rarely used in diatom studies—were more useful in resolving lower, species-level relationships (Figs.

2.2, 2.3). Topological conflicts between nuclear and chloroplast phylogenies were generally minor and unsupported in one or both of the competing trees. Analysis of the combined (nuclear + chloroplast) dataset resulted in a fully resolved and strongly supported phylogeny, with higher- and lower-level relationships tending toward nuclear and chloroplast results, respectively (Figs. 2.6, 2.7).

Although these analyses included a representative cross-section of the total extant morphological diversity in Thalassiosirales (Hasle and Syvertsen, 1997; Julius, 2000;

Theriot and Serieyssol, 1994), sampling was limited for some groups, such as

Thalassiosira species with a "plicated" valve face (Hasle and Lange, 1989; Julius and

Tanimura, 2001) or linear areolar array (Hasle and Fryxell, 1977). In addition, these analyses did not consider the many extinct genera in Thalassiosirales (e.g., Mesodictyon

Theriot & Bradbury, Tertiarius H. Håkannson & G.C. Khursevich). Ongoing efforts to reconstruct Thalassiosirales phylogeny from both morphological and DNA sequence data can better accommodate this diversity, and in doing so, provide a more complete representation of the phylogeny. These more inclusive phylogenetic hypotheses will more

26 adequately address the numerous taxonomic uncertainties documented in Thalassiosirales literature and underscored by our phylogenetic results. Our results, for example, offer little toward resolving the taxonomic uncertainty surrounding the large and vaguely defined genus, Thalassiosira. In contrast, the most important taxonomic implication from this study concerned the strongly supported phylogenetic position of C. distinguenda, the type species for Cyclotella, which now allows for clear delimitation of the taxonomic scope and content of Cyclotella sensu stricto (Fig. 2.7; Clade B). Finally, alternative topology tests strongly rejected the higher-level Thalassiosiraceae–Stephanodiscaceae classification, which was based heavily on habitat preference (Glezer and Makarova,

1986).

The colonization of fresh waters

Salinity is thought to be a significant barrier to the distribution of diatoms, though the exact nature of the salinity barrier is uncertain because the problem has not been approached, nor have existing hypotheses been tested, in a phylogenetic, hypothesis- driven framework. Rather, previous inferences were based on patterns inferred from a diatom classification system that is largely phenetic (Mann, 1999; Round et al., 1990;

Round and Sims, 1980). The initial hypothesis of Round and Sims (1980) posited an unlikely evolutionary scenario in which all major diatom lineages would be naturally split along marine–freshwater lines (Fig. 2.1a), and whereas secondary "leakage" of freshwater species into the marine habitat was irrefutable, the possibility of movement in the opposite direction was dismissed outright. The concept of a uni-directional salinity barrier is implicit in the analogy that likens the colonization of fresh waters to Julius

Caesar's "crossing the Rubicon" (Mann, 1999). The analogy is a useful one, however,

27 because it makes several testable predictions about the nature of the salinity barrier in

diatoms. First, salinity is a uni-directional barrier, and observations suggest that the most

probable colonization events proceed from marine to fresh waters (Mann, 1999; Round

and Sims, 1980). Second, marine–freshwater transitions are major, landmark events in

the evolution of diatoms, and as such, successful colonization events should result in substantial species diversification. Third, as major historical events, successful transitions between marine and fresh waters are rare. Finally, colonization events are irreversible, meaning that descendants of a derived, freshwater ancestor are not expected to re-

colonize the marine environment.

Thalassiosirales is one of the predominant diatom lineages, in both cell abundance

and species diversity, in the plankton of marine and fresh waters and so provides an

excellent opportunity to test these hypotheses. With at least six transitions between

marine and fresh waters, our results showed evidence for considerably more movement

between the two habitats than expected (Figs. 2.2–2.8). One freshwater colonization

involved T. gessneri, the sole representative of a presumably larger clade of marine,

brackish, and freshwater Thalassiosira species with a tangentially undulated (or

"plicated") valve face (Hasle and Lange, 1989; Julius and Tanimura, 2001; Tanimura,

1996). We also found surprisingly strong evidence that the predominantly freshwater

Stephanodiscaceae is polyphyletic, consisting of two distantly related lineages that

represent independent colonizations of fresh waters (Figs. 2.2–2.7; Clades B and C).

Together, these two lineages comprise the vast majority of freshwater diversity in the

Thalassiosirales. One Stephanodiscaceae lineage (Figs 2.2–2.7; Clade B) includes

obligate marine (e.g. Cy. stylorum), obligate freshwater (e.g. Cy. distinguenda), and

28 "euryhaline" species with populations found in one or both habitats (e.g. Cy.

menegheniana) (Mann, 1999; Round and Sims, 1980). The second Stephanodiscaceae

lineage (Figs. 2.2–2.7; Clade C) can be considered obligate freshwater, since

Cyclostephanos, Stephanodiscus, and the remaining (misclassified) Cyclotella have no

known representatives in brackish or marine waters. As implied by their classification,

Stephanodiscaceae is traditionally regarded as monophyletic, based on the freshwater

stratigraphic sequence (e.g. Krebs, 1990; Krebs et al., 1987) and several shared

morphological characters (Theriot and Serieyssol, 1994), which our analyses suggest are

convergent. The morphological characters are compelling and need to be formally coded

and included in phylogenetic analyses, whereas the absence of Clade B species from the

fossil record prior to the Miocene is negative evidence and therefore not in direct conflict

with our findings.

Colonizations of the marine habitat by freshwater taxa are thought to be rare

(Mann, 1999), if not physiologically impossible (Round and Sims, 1980). Mann (1999)

classified Cyclotella as "euryhaline" and predicted that it might represent one of the few

exceptions to this rule. We found compelling evidence for three independent re-

colonizations of the marine habitat, two of which involved Cyclotella (Fig. 2.8). Each of

two marine re-colonizations within Cyclotella were statistically significant, and each led

to subsequent, albeit limited, species diversification (Fig. 2.8). A full one-half of the

habitat shifts uncovered by our analyses were freshwater-to-marine, thus providing the

first evidence for a plausible—if not frequent—freshwater-to-marine colonization route

by diatoms.

29 These analyses did not include many of the Thalassiosira species known to have

secondarily colonized fresh waters (see Edlund et al., 2000), so although our results

captured the two colonization events that led to the greatest species diversification (Figs.

2.2–2.7; Clades B and C), the actual number of "leakages" between the two habitats is far

greater than the few we uncovered. We fully expect that these intraspecific transitions

outnumber those that lead to substantial, if any, species diversification. These small-scale transitions, discounted as "leakages" and trivialized in the Rubicon analogy, probably

account for the vast majority of marine–freshwater transitions in Thalassiosirales and

across diatoms as a whole. Thus, these cases present an excellent opportunity for

comparative studies aimed at understanding the physiological basis of salinity tolerance

and euryhalinity in diatoms. Our treatment of habitat as a binary variable, though a fair

representation of their natural distribution, was slightly oversimplified. A comparative

assessment of salinity optima and salinity tolerance might provide further insights in the

nature of the salinity barrier in diatoms. For example, the physiological differences

between euryhaline species (e.g. T. pseudonana) and strictly marine or strictly freshwater

species are unclear. Ease of culture and the availability of a well-supported phylogeny

make Thalassiosirales an excellent system for studying these types of questions.

These results uncovered substantially more marine–freshwater transitions than

expected, particularly in our finding of two distinct freshwater lineages within

Stephanodiscaceae. We also found compelling evidence for a freshwater-to-marine

colonization route by diatoms. Nevertheless, marine and freshwater species were not distributed randomly across the phylogeny, which indicates that salinity does present a

barrier to the spatial distribution of some diatoms. These results underscore that the

30 pattern and sequence of habitat shifts in Thalassiosirales—inferred from a densely sampled species phylogeny—could not have been gleaned from any level of the classification system, which offers little if anything to evolutionary studies of diatoms.

For example, phylogenetic analysis showed that one genus, Cyclotella, is polyphyletic.

Dense taxon sampling of marine and freshwater species within Cyclotella revealed the strongest evidence for a freshwater-to-marine colonization pathway, which was found in just one of the three Cyclotella lineages (Fig. 2.8). Like salinity, pH constitutes a significant barrier to the distribution of freshwater diatoms, particularly freshwater pennates. A phylogenetic approach like the one taken here will provide necessary phylogenetic context to the distribution of diatoms along pH gradients (Pither and

Aarssen, 2005a; Pither and Aarssen, 2005b; Pither and Aarssen, 2006; Telford et al.,

2006) and other important environmental variables.

Acknowledgements

We thank Greta Fryxell and Matt Julius for insights about Thalassiosirales and phylogeny. We thank David Czarnecki, Mark Edlund, Matt Julius, Sung-Ho Kang, James

Nienow, Jan Rines, Elizabeth Ruck, and many others for sharing collecting localities or directly providing water samples or cultures. This paper is dedicated to David Czarnecki, who, until his recent death, maintained the world's largest culture collection of freshwater diatoms and who was of great help to us in establishing our own diatom culture collection. This research was supported by NSF PEET grant (DEB-0118883) and NSF

Dissertation Improvement Grant (DEB-0407815).

31 Literature cited

Alverson, A. J., J. J. Cannone, R. R. Gutell, and E. C. Theriot. 2006. The evolution of

elongate shape in diatoms. Journal of Phycology 42:655–668.

Alverson, A. J., and L. Kolnick. 2005. Intragenomic nucleotide polymorphism among

small subunit (18S) rDNA paralogs in the diatom genus Skeletonema

(Bacillariophyta). Journal of Phycology 41:1248–1257.

Alverson, A. J., and E. C. Theriot. 2005. Comments on recent progress toward

reconstructing the diatom phylogeny. Journal of Nanoscience and

Nanotechnology 5:57–62.

Andersen, R. A. 2004. Biology and systematics of heterokont and haptophyte algae.

American Journal of Botany 91:1508–1522.

Behrenfeld, M. J., and P. G. Falkowski. 1997. Photosynthetic rates derived from satellite-

based chlorophyll concentration. Limnology & Oceanography 42:1–20.

Beszteri, B., E. Acs, and L. K. Medlin. 2005. Ribosomal DNA sequence variation among

sympatric strains of the Cyclotella meneghiniana complex (Bacillariophyceae)

reveals cryptic diversity. Protist 156:317–333.

Buckley, T. R. 2002. Model misspecification and probabilistic tests of topology:

Evidence from empirical data sets. Systematic Biology 51:509–523.

Dawes, C. J. 1998. Abiotic factors. Pages 16–44 in Marine Botany. John Wiley & Sons,

Inc., New York.

Edlund, M. B., C. M. Taylor, C. L. Schelske, and E. F. Stoermer. 2000. Thalassiosira

baltica (Grunow) Ostenfeld (Bacillariophyta), a new exotic species in the Great

Lakes. Canadian Journal of Fisheries & Aquatic Sciences 57:610–615.

32 Falkowski, P. G., R. T. Barber, and V. Smetacek. 1998. Biogeochemical controls and

feedbacks on ocean primary production. Science 281:200–206.

Field, C. B., M. J. Behrenfeld, J. T. Randerson, and P. Falkowski. 1998. Primary

production of the biosphere: integrating terrestrial and oceanic components.

Science 281:237–240.

Fritsch, F. E. 1935. The Structure and Reproduction of the AlgaeCambridge University

Press, London.

Fryxell, G. A., and G. R. Hasle. 1977. The genus Thalassiosira: some species with a

modified ring of central strutted processes. Beiheft zur Nova Hedwigia 54:67–98.

Fryxell, G. A., and G. R. Hasle. 1979a. The genus Thalassiosira: species with internal

extensions of the strutted processes. Phycologia 18:378–393.

Fryxell, G. A., and G. R. Hasle. 1979b. The genus Thalassiosira: T. trifulta sp. nova and

other species with tricolumnar supports on strutted processes. Beihefte zur Nova

Hedwigia 64:13–40.

Glezer, Z. I., and I. V. Makarova. 1986. New order and family of diatoms

(Bacillariophyta). Botanicheskii Zhurnal (Leningrad) 71:673–676.

Goldman, N., J. P. Anderson, and A. G. Rodrigo. 2000. Likelihood-based tests of

topologies in phylogenetics. Systematic Biology 49:652–670.

Goloboff, P. A., J. S. Farris, and K. C. Nixon. 2004. TNT. Cladistics 20:84–84.

Guillard, R. R. L. 1975. Culture of phytoplankton for feeding marine invertebrates. Pages

26–60 in Culture of Marine Invertebrate Animals (W. L. Smith, and M. H.

Chanley, eds.). Plenum Press, New York.

33 Guillard, R. R. L., and J. H. Ryther. 1962. Studies of marine planktonic diatoms. I.

Cyclotella nana Hustedt and Detonula confervacea (Cleve) Gran. Canadian

Journal of Microbiology 8:229–239.

Hasle, G. R. 1968. The valve processes of the centric diatom genus Thalassiosira. Nytt

Magasin for Botanikk 15:193–201.

Hasle, G. R. 1973. Thalassiosiraceae, a new diatom family. Norwegian Journal of Botany

20:67–69.

Hasle, G. R. 1978a. Some freshwater and brackish water species of the diatom genus

Thalassiosira Cleve. Phycologia 17:263–292.

Hasle, G. R. 1978b. Some Thalassiosira species with one central process. Norwegian

Journal of Botany 25:77–110.

Hasle, G. R., and G. A. Fryxell. 1977. The genus Thalassiosira: some species with a

linear areola array. Nova Hedwigia Beiheft 54:14–60.

Hasle, G. R., and C. B. Lange. 1989. Freshwater and brackish water Thalassiosira

(Bacillariophyceae) taxa with tangentially undulated valves. Phycologia 28:120–

135.

Hasle, G. R., and E. E. Syvertsen. 1985. Thalassiosiropsis, a new diatom genus from the

fossil records. Micropaleontology (New York) 31:82–91.

Hasle, G. R., and E. E. Syvertsen. 1997. Marine Diatoms. Pages 5–386 in Identifying

Marine Phytoplankton (C. R. Tomas, ed.) Academic Press, San Diego.

Herth, W. 1978. Special chitin-fibril-synthesizing apparatus in centric diatom Cyclotella.

Naturwissenschaften 65:260–261.

34 Herth, W. 1979. Site of beta-chitin fibril formation in centric diatoms. 2. Chitin-forming

cytoplasmic structures. Journal of Ultrastructure Research 68:16–27.

Herth, W., and W. Barthlott. 1979. Site of beta-chitin fibril formation in centric diatoms.

1. Pores and fibril formation. Journal of Ultrastructure Research 68:6–15.

Hillis, D. M., B. K. Mable, and C. Moritz. 1996. Applications of molecular systematics:

the state of the field and a look into the futurein Molecular Systematics (D. M.

Hillis, C. Moritz, and B. K. Mable, eds.). Sinauer Associates, Sunderland,

Massachusetts.

Huelsenbeck, J. P., D. M. Hillis, and R. Jones. 1996. Parametric bootstrapping in

molecular phylogenetics: applications and performance. Pages 19–45 in

Molecular Zoology: Advances, Strategies and Protocols (J. D. Ferraris, and S. R.

Palumbi, eds.). Wiley-Liss, New York.

Iler, R. K. 1979. The Chemistry of Silica: Solubility, Polymerization, Colloid and Surface

Properties, and BiochemistryJohn Wiley & Sons, New York.

Julius, M. L. 2000. Phylogeny of the Cyclostephanoid Diatoms: An Investigation of their

Morphology and Stratigraphy. PhD thesis. University of Michigan, Natural

Resources and Environment, 209 pages.

Julius, M. L., and Y. Tanimura. 2001. Cladistic analysis of plicated Thalassiosira

(Bacillariophyceae). Phycologia 40:111–122.

Kaczmarska, I., M. Beaton, A. C. Benoit, and L. K. Medlin. 2006. Molecular phylogeny

of selected members of the order Thalassiosirales (Bacillariophyta) and evolution

of the fultoportula. Journal of Phycology 42:121–138.

35 Kass, R. E., and A. E. Raftery. 1995. Bayes Factors. Journal of the American Statistical

Association 90:773–795.

Katz, M. E., Z. V. Finkel, D. Grzebyk, A. H. Knoll, and P. G. Falkowski. 2004.

Evolutionary trajectories and biogeochemical impacts of marine eukaryotic

phytoplankton. Annual Review of Ecology Evolution and Systematics 35:523–

556.

Kilham, S. S., D. A. Kreeger, C. E. Goulden, and S. Lynn. 1998. COMBO: A defined

freshwater culture medium for algae and zooplankton. Hydrobiologia 377:147–

159.

Krebs, W. N. 1990. The biochronology of freshwater planktonic diatom communities in

western North America. Pages 485–499 in Proceedings of the 11th International

Diatom Symposium (J. P. Kociolek, ed.) California Academy of Sciences, San

Francisco.

Krebs, W. N., and J. P. Bradbury. 1982. Neogene Lacustrine Diatom Biostratigraphy of

Western Snake River Basin, Idaho and Oregon. AAPG Bulletin 66:1693–1693.

Krebs, W. N., J. P. Bradbury, and E. Theriot. 1987. Neogene and Quaternary lacustrine

diatom biochronology western USA. Palaios 2:505–513.

Lebour, M. V. 1930. The planktonic diatoms of northern seas. Royal Society Publication

no. 116:1–244.

Loginova, L. P. 1990. Classification of the diatom genus Cyclotella. Pages 37–46 in

Proceedings of the 10th International Symposium on Recent and Fossil Diatoms

(H. Simola, ed.) Otto Koeltz, Koenigstein.

36 Lowe, R. L. 1975. Comparative ultrastructure of the valves of some Cyclotella species

(Bacillariophyceae). Journal of Phycology 11:415–424.

Maddison, W. P., and D. R. Maddison. 2003. MacClade. Sinauer Associates, Inc.

Maddison, W. P., and D. R. Maddison. 2006. Mesquite: a modular system for

evolutionary analysis, version 1.1.

Mann, D. G. 1999. Crossing the Rubicon: the effectiveness of the marine/freshwater

interface as a barrier to the migration of diatom germplasm. Pages 1–21 in

Proceedings of the 14th International Diatom Symposium (S. Mayama, M. Idei,

and I. Koizumi, eds.). Koeltz Scientific Books, Koenigstein.

Martin-Jezequel, V., M. Hildebrand, and M. A. Brzezinski. 2000. Silicon metabolism in

diatoms: Implications for growth. Journal of Phycology 36:821–840.

Medlin, L. K., W. H. C. F. Kooistra, R. Gersonde, and U. Wellbrock. 1996. Evolution of

the diatoms (Bacillariophyta): III. Molecular evidence for the origin of the

Thalassiosirales. Nova Hedwigia Beiheft 112:221–234.

Nelson, D. M., P. Treguer, M. A. Brzezinski, A. Leynaert, and B. Queguiner. 1995.

Production and dissolution of biogenic silica in the ocean: revised global

estimates, comparison with regional data and relationship to biogenic

sedimentation. Global Biogeochemical Cycles 9:359–372.

Nylander, J. A. A., F. Ronquist, J. P. Huelsenbeck, and J. L. Nieves-Aldrey. 2004.

Bayesian phylogenetic analysis of combined data. Systematic Biology 53:47–67.

Pagel, M. 1999. The maximum likelihood approach to reconstructing ancestral character

states of discrete characters on phylogenies. Systematic Biology 48:612–622.

37 Pither, J., and L. W. Aarssen. 2005a. Environmental specialists: their prevalence and their

influence on community-similarity analyses. Ecology Letters 8:261–271.

Pither, J., and L. W. Aarssen. 2005b. The evolutionary species pool hypothesis and

patterns of freshwater diatom diversity along a pH gradient. Journal of

Biogeography 32:503–513.

Pither, J., and L. W. Aarssen. 2006. How prevalent are pH-specialists diatoms? A reply to

Telford et al. (2006). Ecology Letters 9:E6–E12.

Reeder, T. W. 2003. A phylogeny of the Australian Sphenomorphus group (Scincidae :

Squamata) and the phylogenetic placement of the crocodile skinks (Tribolonotus):

Bayesian approaches to assessing congruence and obtaining confidence in

maximum likelihood inferred relationships. Molecular Phylogenetics and

Evolution 27:384–397.

Ronquist, F., and J. P. Huelsenbeck. 2003. MrBayes 3: Bayesian phylogenetic inference

under mixed models. Bioinformatics 19:1572–1574.

Round, F. E., R. M. Crawford, and D. G. Mann. 1990. The Diatoms: Biology &

Morphology of the GeneraCambridge University Press, Cambridge.

Round, F. E., and P. A. Sims. 1980. The distribution of diatom genera in marine and

freshwater environments and some evolutionary considerations. Pages 301–320 in

Proceedings of the Sixth Symposium on Recent and Fossil Diatoms (R. Ross, ed.)

Otto Koeltz Science Publishers, Hirschberg.

Schluter, D., T. Price, A. O. Mooers, and D. Ludwig. 1997. Likelihood of ancestor states

in adaptive radiation. Evolution 51:1699–1711.

38 Schoniger, M., and A. Von Haeseler. 1994. A Stochastic model for the evolution of

autocorrelated DNA sequences. Molecular Phylogenetics and Evolution 3:240–

247.

Serieyssol, K. K. 1980. Cyclotella species of late Miocene age from St. Bauzile, France.

Pages 27–42 in Proceedings of the Sixth Symposium on Recent and Fossil

Diatoms (R. Ross, ed.) Otto Koeltz Science Publishers, Koenigstein.

Shapiro, B., A. Rambaut, and A. J. Drummond. 2006. Choosing appropriate substitution

models for the phylogenetic analysis of protein-coding sequences. Mol Biol Evol

23:7–9.

Spamer, E. E., and E. C. Theriot. 1997. "Stephanodiscus minutulus", "S. minutus", and

similar epithets in taxonomic, ecological, and evolutionary studies of modern and

fossil diatoms (Bacillariophyceae: Thalassiosiraceae): A century and a half of

uncertain taxonomy and nomenclatural hearsay. Proceedings of the Academy of

Natural Sciences of Philadelphia 148:231–272.

Stoermer, E. F., and M. L. Julius. 2003. Centric Diatoms. Pages 559–594 in Freshwater

Algae of North America (J. D. Wehr, and R. G. Sheath, eds.). Academic Press,

Amsterdam.

Strelnikova, N. I., and T. V. Lastivka. 1999. The problem of the origin of marine and

freshwater diatoms. Pages 113–123 in Proceedings of the 14th International

Diatom Symposium (S. Mayama, M. Idei, and I. Koizumi, eds.). Koeltz Scientific

Books, Koenigstein.

Tanimura, Y. 1996. Fossil marine plicated Thalassiosira: Taxonomy and an idea on

phylogeny. Diatom Research 11:165–202.

39 Telford, R. J., V. Vandvik, and H. J. B. Birks. 2006. How many freshwater diatoms are

pH specialists? A response to Pither & Aarssen (2005). Ecology Letters 9:E1–E5.

Theriot, E., and K. Serieyssol. 1994. Phylogenetic systematics as a guide to

understanding features and potential morphological characters of the centric

diatom family Thalassiosiraceae. Diatom Research 9:429–450.

Treguer, P., D. M. Nelson, A. J. Vanbennekom, D. J. Demaster, A. Leynaert, and B.

Queguiner. 1995. The silica balance in the world ocean: a reestimate. Science

268:375–379.

Walsby, A. E., and A. Xypolyta. 1977. Form resistance of chitan fibers attached to cells

of Thalassiosira fluviatilis Hustedt. British Phycological Journal 12:215–223.

Weber, C. I. 1970. A new freshwater centric diatom Microsiphona potamos gen. et sp.

nov. Journal of Phycology 6:149–153.

Werner, D. 1977. Introduction with a note on taxonomy. Pages 1–17 in The Biology of

Diatoms (D. Werner, ed.) University of California Press, Berkeley.

40

Figure 2.1. Tree topologies representing alternative hypotheses about the relationship between the marine Thalassiosiraceae and freshwater Stephanodiscaceae.

41 Figure 2.2. Phylogram of one of nine most parsimonious trees from maximum parsimony analysis of the combined chloroplast (psbC + rbcL) dataset (tree length = 4130 steps, CI

= 0.313, RI = 0.633). Non-parametric bootstrap values greater than 95% are shown with

an asterisk (*). Terminal taxa are identified as either marine (black) or freshwater (blue),

and the history of marine–freshwater transitions was reconstructed with maximum

parsimony (7 steps). Three clades (A–C) are highlighted to facilitate discussion in the text. Generic abbreviations are: Bacterosira (Ba.), Bellerochea (Be.), Cyclostephanos

(Cs.), Cyclotella (Cy.), Detonula (De.), Ditylum (Di.), Helicotheca (H.), Lauderia (La.),

Lithodesmium (Li.), Minidiscus (M.), Porosira (P.), Skeletonema (Sk.), Stephanodiscus

(St.), Thalassiosira (T.).

42 Figure 2.2

43 Figure 2.3. Majority-rule consensus tree from Bayesian analysis of the combined

chloroplast (psbC + rbcL) dataset. See Table 4 for model details. Bayesian posterior

probability values greater than 0.95 are shown with an asterisk (*). Terminal taxa are

identified as either marine (black) or freshwater (blue), and the history of marine–

freshwater transitions was reconstructed with maximum parsimony (7 steps). Three

clades (A–C) are highlighted to facilitate discussion in the text. Generic abbreviations

are: Bacterosira (Ba.), Bellerochea (Be.), Cyclostephanos (Cs.), Cyclotella (Cy.),

Detonula (De.), Ditylum (Di.), Helicotheca (H.), Lauderia (La.), Lithodesmium (Li.),

Minidiscus (M.), Porosira (P.), Skeletonema (Sk.), Stephanodiscus (St.), Thalassiosira

(T.).

44 Figure 2.3

45 Figure 2.4. One of 26 most parsimonious trees from maximum parsimony analysis of the

combined nuclear (SSU + partial LSU rDNA) dataset (tree length = 2621 steps, CI =

0.487, RI = 0.761). Non-parametric bootstrap values greater than 95% are shown with an

asterisk (*). Terminal taxa are identified as either marine (black) or freshwater (blue), and

the history of marine–freshwater transitions was reconstructed with maximum parsimony

(7 steps). Three clades (A–C) are highlighted to facilitate discussion in the text. Generic

abbreviations are: Bacterosira (Ba.), Bellerochea (Be.), Cyclostephanos (Cs.), Cyclotella

(Cy.), Detonula (De.), Ditylum (Di.), Helicotheca (H.), Lauderia (La.), Lithodesmium

(Li.), Minidiscus (M.), Porosira (P.), Skeletonema (Sk.), Stephanodiscus (St.),

Thalassiosira (T.).

46 Figure 2.4

47 Figure 2.5. Majority-rule consensus tree from Bayesian analysis of the combined nuclear

(SSU + partial LSU rDNA) dataset. See Table 4 for model details. Bayesian posterior

probability values greater than 0.95 are shown with an asterisk (*). Terminal taxa are

identified as either marine (black) or freshwater (blue), and the history of marine–

freshwater transitions was reconstructed with maximum parsimony (7 steps). Three

clades (A–C) are highlighted to facilitate discussion in the text. Generic abbreviations

are: Bacterosira (Ba.), Bellerochea (Be.), Cyclostephanos (Cs.), Cyclotella (Cy.),

Detonula (De.), Ditylum (Di.), Helicotheca (H.), Lauderia (La.), Lithodesmium (Li.),

Minidiscus (M.), Porosira (P.), Skeletonema (Sk.), Stephanodiscus (St.), Thalassiosira

(T.).

48 Figure 2.5

49 Figure 2.6. One of two most parsimonious trees from maximum parsimony analysis of the total combined (nuclear + chloroplast) dataset (tree length = 6858 steps, CI = 0.381,

RI = 0.682). Non-parametric bootstrap values greater than 95% are shown with an asterisk (*). Terminal taxa are identified as either marine (black) or freshwater (blue), and the history of marine–freshwater transitions was reconstructed with maximum parsimony

(7 steps). Three clades (A–C) are highlighted to facilitate discussion in the text. Generic abbreviations are: Bacterosira (Ba.), Bellerochea (Be.), Cyclostephanos (Cs.), Cyclotella

(Cy.), Detonula (De.), Ditylum (Di.), Helicotheca (H.), Lauderia (La.), Lithodesmium

(Li.), Minidiscus (M.), Porosira (P.), Skeletonema (Sk.), Stephanodiscus (St.),

Thalassiosira (T.).

50 Figure 2.6

51 Figure 2.7. Majority-rule consensus tree from Bayesian analysis of the total combined

(nuclear + chloroplast) dataset. See Table 4 for model details. Bayesian posterior

probability values greater than 0.95 are shown with an asterisk (*). Terminal taxa are

identified as either marine (black) or freshwater (blue), and the history of marine–

freshwater transitions was reconstructed with maximum parsimony (7 steps). Three

clades (A–C) are highlighted to facilitate discussion in the text. Generic abbreviations

are: Bacterosira (Ba.), Bellerochea (Be.), Cyclostephanos (Cs.), Cyclotella (Cy.),

Detonula (De.), Ditylum (Di.), Helicotheca (H.), Lauderia (La.), Lithodesmium (Li.),

Minidiscus (M.), Porosira (P.), Skeletonema (Sk.), Stephanodiscus (St.), Thalassiosira

(T.). Scanning electron micrographs of the cell exterior (left) and interior strutted process ultrastructure (right) are shown for seven exemplar taxa.

52 Figure 2.7

53 Figure 2.8. Maximum likelihood (ML) character mapping and ancestral state reconstruction of habitat type in Thalassiosirales. This analysis used the combined

Bayesian topology and branch lengths (Fig. 2.7), with several zero and near-zero length branches pruned from the tree. Pie graphs at internal nodes show the relative ML support for marine (white) and freshwater (black) ancestral states. Asterisks (*) indicate that the more strongly supported ancestral state is significantly better than the alternative state.

Generic abbreviations are: Bacterosira (Ba.), Cyclostephanos (Cs.), Cyclotella (Cy.),

Detonula (De.), Lauderia (La.), Minidiscus (M.), Porosira (P.), Skeletonema (Sk.),

Stephanodiscus (St.), Thalassiosira (T.).

54 Figure 2.8

55 Table 2.1. Taxon, culture strain, locality, family classification, and GenBank accessions of chloroplast markers for species examined in this study.

Species Culture Locality Family psbC rbcL strain Bacterosira bathyomphala (Cleve) NB04-B6 Narragansett Bay, RI, USA Thalassiosiraceae DQ514734 DQ514816 Syvertsen & Hasle Bacterosira sp. CCMP991 Chase Creek, MA, USA Thalassiosiraceae DQ514716 DQ514798 Detonula pumila (Castracane) Gran NB48 Narragansett Bay, RI, USA Thalassiosiraceae DQ514732 DQ514814 Lauderia annulata Cleve CS30 Pacific Ocean, La Jolla, Thalassiosiraceae DQ514687 DQ514769 CA, USA

Minidiscus trioculatus (F.J.R. CCMP495 Gulf of Maine, ME, USA Thalassiosiraceae DQ514711 DQ514793 56 Taylor) Hasle Porosira glacialis (Grunow) CCMP1099 Southern Ocean, Thalassiosiraceae DQ514685 DQ514767 Jørgensen Antarctica Porosira pseudodelicatula CCMP1433 McMurdo Sound, Thalassiosiraceae DQ514686 DQ514768 (Hustedt) Jousé Antarctica Zingone & CCAP1077/3 Narragansett Bay, RI, USA Thalassiosiraceae DQ514735 DQ514817 Sarno Skeletonema grethae Zingone & CCAP1077/4 Strait of Georgia, British Thalassiosiraceae DQ514736 DQ514818 Sarno Columbia, Canada Zingone & NB02-45 Narragansett Bay, RI, USA Thalassiosiraceae DQ514740 DQ514822 Sarno Species Culture Locality Family psbC rbcL strain Skeletonema menzellii Guillard CCMP787 Sargasso Sea, Atlantic Thalassiosiraceae DQ514739 DQ514821 Ocean Skeletonema pseudocostatum CCAP1077/7 Mediterranean Sea, Thalassiosiraceae DQ514737 DQ514819 Medlin Alexandria, Egypt Skeletonema subsalsum (Cleve- CCAP1077/8 Lower Lough Erne, Ireland Thalassiosiraceae DQ514738 DQ514820 Euler) Bethge Thalassiosira aestivalis Gran & CCMP976 Vancouver, British Thalassiosiraceae DQ514712 DQ514794 Angst Columbia, Canada Thalassiosira angulata (Gregory) BEN02-35 San Joaquin River, Thalassiosiraceae DQ514706 DQ514788 Hasle Benicia, CA, USA

Thalassiosira anguste-lineata (A. BEN02-30 San Joaquin River, Thalassiosiraceae DQ514704 DQ514786 57 Schmidt) G. Fryxell & Hasle Benicia, CA, USA Thalassiosira antarctica Comber CCMP982 Oslo Fjord, Norway Thalassiosiraceae DQ514713 DQ514795 Thalassiosira cf. pacifica FB02-35 San Francisco Bay, CA, Thalassiosiraceae DQ514728 DQ514810 USA Thalassiosira eccentrica BER02-9 San Francisco Bay, CA, Thalassiosiraceae DQ514707 DQ514789 (Ehrenberg) Cleve emend G. USA Fryxell & Hasle Thalassiosira gessneri (Grunow) G. AN02-08 San Joaquin River, Thalassiosiraceae DQ514703 DQ514785 Fryxell & Hasle Antioch, CA, USA Thalassiosira guillardii Hasle CC03-04 Clam Creek, GA, USA Thalassiosiraceae DQ514708 DQ514790 Species Culture Locality Family psbC rbcL strain Thalassiosira guillardii Hasle CCMP988 North Atlantic Ocean Thalassiosiraceae DQ514714 DQ514796 Thalassiosira mediterranea CS16 Port Phillip Bay, Australia Thalassiosiraceae DQ514724 DQ514806 (Schroder) Hasle Thalassiosira minima Gaarder CCMP990 unknown Thalassiosiraceae DQ514715 DQ514797 emend Hasle Thalassiosira minuscula Krasske CCMP1093 Pacific Ocean, La Jolla, Thalassiosiraceae DQ514721 DQ514803 CA, USA Thalassiosira minuscula Krasske FB02-31 San Francisco Bay, CA, Thalassiosiraceae DQ514727 DQ514809 USA Thalassiosira nodulolineata BEN02-33 San Joaquin River, Thalassiosiraceae DQ514705 DQ514787

(Hendey) Hasle & G. Fryxell Benicia, CA, USA 58 Thalassiosira nordenskioeldii Cleve FB02-19 San Francisco Bay, CA, Thalassiosiraceae DQ514726 DQ514808 USA Thalassiosira oceanica Hasle CCMP1001 North Atlantic Ocean Thalassiosiraceae DQ514717 DQ514799 Thalassiosira oestrupii var. CC03-15 Clam Creek, GA, USA Thalassiosiraceae DQ514709 DQ514791 venrickae G. Fryxell & Hasle Hasle & ETC1 Lake Erie, MI, USA Thalassiosiraceae DQ514701 DQ514783 Heimdal Thalassiosira pseudonana Hasle & NEPC709 Hood Canal, WA, USA Thalassiosiraceae DQ514702 DQ514784 Heimdal Thalassiosira pseudonana Hasle & CCMP1057 Dabob Bay, Washington, Thalassiosiraceae DQ514719 DQ514801 Heimdal USA Species Culture Locality Family psbC rbcL strain Thalassiosira punctigera FB02-06 San Francisco Bay, CA, Thalassiosiraceae DQ514725 DQ514807 (Castracane) Hasle USA Thalassiosira punctigera NB02-22 Narragansett Bay, RI, USA Thalassiosiraceae DQ514733 DQ514815 (Castracane) Hasle Thalassiosira ritscheri (Hustedt) LC01-12 Drake Passage, Southern Thalassiosiraceae DQ514731 DQ514813 Hasle Ocean Thalassiosira rotula Meunier CCMP1812 East Sound, Orcas Island, Thalassiosiraceae DQ514723 DQ514805 Washington, USA Thalassiosira sp. CCMP1065 Baffin Bay, Arctic Ocean Thalassiosiraceae DQ514720 DQ514802 Thalassiosira sp. CCMP353 Narragansett Bay, RI, USA Thalassiosiraceae DQ514710 DQ514792 59 Thalassiosira tumida (Janisch) CCMP1469 McMurdo Sound, Southern Thalassiosiraceae DQ514722 DQ514804 Hasle in Hasle, Heimdal & G. Ocean Fryxell Thalassiosira tumida (Janisch) LA09-20 Drake Passage, Southern Thalassiosiraceae DQ514730 DQ514812 Hasle in Hasle, Heimdal & G. Ocean Fryxell Thalassiosira weissflogii (Grunow) CCMP1010 Gulf Stream, Atlantic Thalassiosiraceae DQ514718 DQ514800 G. Fryxell & Hasle Ocean Thalassiosira weissflogii (Grunow) L1296 unknown Thalassiosiraceae DQ514729DQ514811 G. Fryxell & Hasle Species Culture Locality Family psbC rbcL strain Cyclostephanos invisitatus (Hohn & FHTC26 Fairport Harbor, Lake Erie, Stephanodiscaceae DQ514745 DQ514827 Hellerman) Theriot, Stoermer & OH, USA Håkansson Cyclostephanos sp. WTC16 West Lake Okoboji, IA, Stephanodiscaceae DQ514758 DQ514840 USA Cyclostephanos sp. WTC18 West Lake Okoboji, IA, Stephanodiscaceae DQ514759 DQ514841 USA Cyclostephanos tholiformis E.F. FHTC15 Fairport Harbor, Lake Erie, Stephanodiscaceae DQ514744 DQ514826 Stoermer, Håkansson & E.C. OH, USA Theriot Cyclotella atomus Hustedt ROR01-4 Raccourci Old River, LA, Stephanodiscaceae DQ514697 DQ514779 USA 60 Cyclotella bodanica Grunow J98-1 Jackson Lake, WY, USA Stephanodiscaceae DQ514747 DQ514829 Cyclotella cryptica Reimann, Lewin CCMP331 unknown Stephanodiscaceae DQ514688 DQ514770 & Guillard Cyclotella distinguenda Hustedt none Tiplady Bog, MI, USA Stephanodiscaceae DQ514698 DQ514780 Cyclotella gamma Sovereign Cygamma Lake Itasca, MN, USA Stephanodiscaceae DQ514690 DQ514772 Cyclotella cf. cryptica WC03-1 Waller Creek, TX, USA Stephanodiscaceae DQ514700 DQ514782 Cyclotella menegheniana Kützing TI1 Lake Titicaca, Peru Stephanodiscaceae DQ514699 DQ514781 Cyclotella menegheniana Kützing LS03-1 Lake Superior, MI, USA Stephanodiscaceae DQ514695 DQ514777 Species Culture Locality Family psbC rbcL strain Cyclotella menegheniana Kützing F8 Fugleso Lake, Denmark Stephanodiscaceae DQ514692 DQ514774 Cyclotella ocellata Pantocsek LB8 Lake Buchanan, TX, USA Stephanodiscaceae DQ514750 DQ514832 Cyclotella cf. menegheniana L1263 Stump Lake, ND, USA Stephanodiscaceae DQ514693 DQ514775 Cyclotella sp. MC01 Mango Creek, Belize Stephanodiscaceae DQ514696 DQ514778 Cyclotella sp. L1844 Salton Sea, Imperial Stephanodiscaceae DQ514694 DQ514776 County, CA, USA Cyclotella striata (Kützing) Grunow CCMP1586 Jakarta Harbor, Indonesia Stephanodiscaceae DQ514689 DQ514771 in Cleve & Grunow Cyclotella stylorum Brightwell DA04-06 Golfo de Nicoya, Dona Stephanodiscaceae DQ514691 DQ514773

Ana, Costa Rica 61 Cyclotella cf. pseudostelligera L435 Montezuma Well National Stephanodiscaceae DQ514748 DQ514830 Monument, Yavapai Co., AZ, USA Cyclotella pseudostelligera ROR01-1 Raccourci Old River, LA, Stephanodiscaceae DQ514751 DQ514833 (Hustedt) Houk & Klee USA Cyclotella stelligera (Cleve & L1360 Big Pond, Cedar Hill, TX, Stephanodiscaceae DQ514749 DQ514831 Grunow) Houk & Klee USA Stephanodiscus agassizensis H. CHTC1 Cleveland Harbor, Lake Stephanodiscaceae DQ514741 DQ514823 Håkansson & H. Kling Erie, OH, USA Stephanodiscus binderanus ESB2 Lake Erie, MI, USA Stephanodiscaceae DQ514742 DQ514824 (Kützing) Krieger Species Culture Locality Family psbC rbcL strain Stephanodiscus hantzschii f. tenuis WTC21 West Lake Okoboji, IA, Stephanodiscaceae DQ514760 DQ514842 (Hustedt) Håkansson & Stoermer USA Stephanodiscus minutulus (Kützing) J95-21 Jackson Lake, WY, USA Stephanodiscaceae DQ514746 DQ514828 Cleve & Möller Stephanodiscus minutulus (Kützing) WTC1 West Lake Okoboji, IA, Stephanodiscaceae DQ514757 DQ514839 Cleve & Möller USA Stephanodiscus minutulus (Kützing) FHTC11 Fairport Harbor, Lake Erie, Stephanodiscaceae DQ514743 DQ514825 Cleve & Möller OH, USA Stephanodiscus minutulus (Kützing) Y95-6 Yellowstone Lake, WY, Stephanodiscaceae DQ514761 DQ514843 Cleve & Möller USA

Stephanodiscus minutulus (Kützing) Y98-1 Yellowstone Lake, WY, Stephanodiscaceae DQ514762 DQ514844 62 Cleve & Möller USA Stephanodiscus neoastraea H. Sneo4 Grober Müggelsee, Berlin, Stephanodiscaceae DQ514752 DQ514834 Håkansson & B. Hickel Germany Stephanodiscus niagarae Ehrenberg J95-16 Jackson Lake, WY, USA Stephanodiscaceae DQ514753 DQ514835 Stephanodiscus niagarae Ehrenberg OKA-A Okamanpeedan Lake, MN, Stephanodiscaceae DQ514754 DQ514836 USA Stephanodiscus reimerii Theriot WLO11 West Lake Okoboji, IA, Stephanodiscaceae DQ514755 DQ514837 USA Stephanodiscus yellowstonensis Y7 Yellowstone Lake, WY, Stephanodiscaceae DQ514756 DQ514838 Theriot USA Species Culture Locality Family psbC rbcL strain Bellerochea malleus (Brightwell) CCMP143 North Pacific Ocean, outgroup DQ514681 DQ514763 Van Heurck Muroran, Japan Ditylum brightwellii (T. West) CCMP1810 Willapa Bay, Washington outgroup DQ514684 DQ514766 Grunow USA Helicotheca tamesis (Shrubsole) M. CCMP1760 Darien, Panama, near outgroup DQ514682 DQ514764 Ricard Samba River Lithodesmium undulatum Ehrenberg CCMP1806 outgroup DQ514683DQ514765

63 Table 2.2. Taxon, culture strain, locality, family classification, and GenBank accessions of nuclear markers for species examined in this study.

Species Culture Locality Family SSU LSU strain Bacterosira bathyomphala (Cleve) NB04-B6 Narragansett Bay, RI, USA Thalassiosiraceae DQ514894 DQ512444 Syvertsen & Hasle Bacterosira sp. CCMP991 Chase Creek, MA, USA Thalassiosiraceae DQ514877 DQ512426 Detonula pumila (Castracane) Gran NB48 Narragansett Bay, RI, USA Thalassiosiraceae DQ514892 DQ512442 Lauderia annulata Cleve CS30 Pacific Ocean, La Jolla, Thalassiosiraceae DQ514849 DQ512397 CA, USA Minidiscus trioculatus (F.J.R. CCMP495 Gulf of Maine, ME, USA Thalassiosiraceae DQ514872 DQ512421

Taylor) Hasle 64 Porosira glacialis (Grunow) CCMP1099 Southern Ocean, Thalassiosiraceae DQ514847 DQ512395 Jørgensen Antarctica Porosira pseudodelicatula CCMP1433 McMurdo Sound, Thalassiosiraceae DQ514848 DQ512396 (Hustedt) Jousé Antarctica Skeletonema grethae Zingone & CCAP1077/3 Narragansett Bay, RI, USA Thalassiosiraceae DQ512445 Sarno AY684941 Skeletonema grethae Zingone & CCAP1077/4 Strait of Georgia, British Thalassiosiraceae DQ512446 Sarno Columbia, Canada AY684942 Skeletonema japonicum Zingone & NB02-45 Narragansett Bay, RI, USA Thalassiosiraceae DQ512450 Sarno AY684968 Species Culture Locality Family SSU LSU strain Skeletonema menzellii Guillard CCMP787 Sargasso Sea, Atlantic Thalassiosiraceae DQ512449 Ocean DQ011161 Skeletonema pseudocostatum CCAP1077/7 Mediterranean Sea, Thalassiosiraceae DQ512447 Medlin Alexandria, Egypt AY684952 Skeletonema subsalsum (Cleve- CCAP1077/8 Lower Lough Erne, Ireland Thalassiosiraceae DQ512448 Euler) Bethge AY684962 Thalassiosira aestivalis Gran & CCMP976 Vancouver, British Thalassiosiraceae DQ514873 DQ512422 Angst Columbia, Canada Thalassiosira angulata (Gregory) BEN02-35 San Joaquin River, Thalassiosiraceae DQ514867 DQ512416 Hasle Benicia, CA, USA

Thalassiosira anguste-lineata (A. BEN02-30 San Joaquin River, Thalassiosiraceae DQ514865 DQ512414 65 Schmidt) G. Fryxell & Hasle Benicia, CA, USA Thalassiosira antarctica Comber CCMP982 Oslo Fjord, Norway Thalassiosiraceae DQ514874 DQ512423 Thalassiosira cf. pacifica FB02-35 San Francisco Bay, CA, Thalassiosiraceae DQ514888 DQ512438 USA Thalassiosira eccentrica BER02-9 San Francisco Bay, CA, Thalassiosiraceae DQ514868 DQ512417 (Ehrenberg) Cleve emend G. USA Fryxell & Hasle Thalassiosira gessneri (Grunow) G. AN02-08 San Joaquin River, Thalassiosiraceae DQ514864 DQ512413 Fryxell & Hasle Antioch, CA, USA Thalassiosira guillardii Hasle CC03-04 Clam Creek, GA, USA Thalassiosiraceae DQ514869 DQ512418 Species Culture Locality Family SSU LSU strain Thalassiosira guillardii Hasle CCMP988 North Atlantic Ocean Thalassiosiraceae DQ514875 DQ512424 Thalassiosira mediterranea CS16 Port Phillip Bay, Australia Thalassiosiraceae DQ512434 (Schroder) Hasle missing Thalassiosira minima Gaarder CCMP990 unknown Thalassiosiraceae DQ514876 DQ512425 emend Hasle Thalassiosira minuscula Krasske CCMP1093 Pacific Ocean, La Jolla, Thalassiosiraceae DQ514882 DQ512431 CA, USA Thalassiosira minuscula Krasske FB02-31 San Francisco Bay, CA, Thalassiosiraceae DQ514887 DQ512437 USA Thalassiosira nodulolineata BEN02-33 San Joaquin River, Thalassiosiraceae DQ514866 DQ512415

(Hendey) Hasle & G. Fryxell Benicia, CA, USA 66 Thalassiosira nordenskioeldii Cleve FB02-19 San Francisco Bay, CA, Thalassiosiraceae DQ514886 DQ512436 USA Thalassiosira oceanica Hasle CCMP1001 North Atlantic Ocean Thalassiosiraceae DQ514878 DQ512427 Thalassiosira oestrupii var. CC03-15 Clam Creek, GA, USA Thalassiosiraceae DQ514870 DQ512419 venrickae G. Fryxell & Hasle Thalassiosira pseudonana Hasle & ETC1 Lake Erie, MI, USA Thalassiosiraceae DQ514862 DQ512411 Heimdal Thalassiosira pseudonana Hasle & NEPC709 Hood Canal, WA, USA Thalassiosiraceae DQ514863 DQ512412 Heimdal Thalassiosira pseudonana Hasle & CCMP1057 Dabob Bay, Washington, Thalassiosiraceae DQ514880 DQ512429 Heimdal USA Species Culture Locality Family SSU LSU strain Thalassiosira punctigera FB02-06 San Francisco Bay, CA, Thalassiosiraceae DQ514885 DQ512435 (Castracane) Hasle USA Thalassiosira punctigera NB02-22 Narragansett Bay, RI, USA Thalassiosiraceae DQ514893 DQ512443 (Castracane) Hasle Thalassiosira ritscheri (Hustedt) LC01-12 Drake Passage, Southern Thalassiosiraceae DQ514891 DQ512441 Hasle Ocean Thalassiosira rotula Meunier CCMP1812 East Sound, Orcas Island, Thalassiosiraceae DQ514884 DQ512433 Washington, USA Thalassiosira sp. CCMP1065 Baffin Bay, Arctic Ocean Thalassiosiraceae DQ514881 DQ512430 Thalassiosira sp. CCMP353 Narragansett Bay, RI, USA Thalassiosiraceae DQ514871 DQ512420 67 Thalassiosira tumida (Janisch) CCMP1469 McMurdo Sound, Southern Thalassiosiraceae DQ514883 DQ512432 Hasle in Hasle, Heimdal & G. Ocean Fryxell Thalassiosira tumida (Janisch) LA09-20 Drake Passage, Southern Thalassiosiraceae DQ514890 DQ512440 Hasle in Hasle, Heimdal & G. Ocean Fryxell Thalassiosira weissflogii (Grunow) CCMP1010 Gulf Stream, Atlantic Thalassiosiraceae DQ514879 DQ512428 G. Fryxell & Hasle Ocean Thalassiosira weissflogii (Grunow) L1296 unknown Thalassiosiraceae DQ514889DQ512439 G. Fryxell & Hasle Species Culture Locality Family SSU LSU strain Cyclostephanos invisitatus (Hohn & FHTC26 Fairport Harbor, Lake Erie, Stephanodiscaceae DQ514899 DQ512455 Hellerman) Theriot, Stoermer & OH, USA Håkansson Cyclostephanos sp. WTC16 West Lake Okoboji, IA, Stephanodiscaceae DQ514912 DQ512468 USA Cyclostephanos sp. WTC18 West Lake Okoboji, IA, Stephanodiscaceae DQ514913 DQ512469 USA Cyclostephanos tholiformis E.F. FHTC15 Fairport Harbor, Lake Erie, Stephanodiscaceae DQ514898 DQ512454 Stoermer, Håkansson & E.C. OH, USA Theriot Cyclotella atomus Hustedt ROR01-4 Raccourci Old River, LA, Stephanodiscaceae DQ514858 DQ512407 USA 68 Cyclotella bodanica Grunow J98-1 Jackson Lake, WY, USA Stephanodiscaceae DQ514901 DQ512457 Cyclotella cryptica Reimann, Lewin CCMP331 unknown Stephanodiscaceae DQ514850 DQ512398 & Guillard Cyclotella distinguenda Hustedt none Tiplady Bog, MI, USA Stephanodiscaceae DQ514859 DQ512408 Cyclotella gamma Sovereign Cygamma Lake Itasca, MN, USA Stephanodiscaceae DQ514852 DQ512400 Cyclotella cf. cryptica WC03-1 Waller Creek, TX, USA Stephanodiscaceae DQ514861 DQ512410 Cyclotella menegheniana Kützing TI1 Lake Titicaca, Peru Stephanodiscaceae DQ514860 DQ512409 Cyclotella menegheniana Kützing LS03-1 Lake Superior, MI, USA Stephanodiscaceae DQ514856 DQ512405 Species Culture Locality Family SSU LSU strain Cyclotella menegheniana Kützing F8 Fugleso Lake, Denmark Stephanodiscaceae DQ514853 DQ512402 Cyclotella ocellata Pantocsek LB8 Lake Buchanan, TX, USA Stephanodiscaceae DQ514904 DQ512460 Cyclotella cf. menegheniana L1263 Stump Lake, ND, USA Stephanodiscaceae DQ514854 DQ512403 Cyclotella sp. MC01 Mango Creek, Belize Stephanodiscaceae DQ514857 DQ512406 Cyclotella sp. L1844 Salton Sea, Imperial Stephanodiscaceae DQ514855 DQ512404 County, CA, USA Cyclotella striata (Kützing) Grunow CCMP1586 Jakarta Harbor, Indonesia Stephanodiscaceae DQ514851 DQ512399 in Cleve & Grunow Cyclotella stylorum Brightwell DA04-06 Golfo de Nicoya, Dona Stephanodiscaceae DQ512401

Ana, Costa Rica missing 69 Cyclotella cf. pseudostelligera L435 Montezuma Well National Stephanodiscaceae DQ514902 DQ512458 Monument, Yavapai Co., AZ, USA Cyclotella pseudostelligera ROR01-1 Raccourci Old River, LA, Stephanodiscaceae DQ514905 DQ512461 (Hustedt) Houk & Klee USA Cyclotella stelligera (Cleve & L1360 Big Pond, Cedar Hill, TX, Stephanodiscaceae DQ514903 DQ512459 Grunow) Houk & Klee USA Stephanodiscus agassizensis H. CHTC1 Cleveland Harbor, Lake Stephanodiscaceae DQ514895 DQ512451 Håkansson & H. Kling Erie, OH, USA Stephanodiscus binderanus ESB2 Lake Erie, MI, USA Stephanodiscaceae DQ514896 DQ512452 (Kützing) Krieger Species Culture Locality Family SSU LSU strain Stephanodiscus hantzschii f. tenuis WTC21 West Lake Okoboji, IA, Stephanodiscaceae DQ514914 DQ512470 (Hustedt) Håkansson & Stoermer USA Stephanodiscus minutulus (Kützing) J95-21 Jackson Lake, WY, USA Stephanodiscaceae DQ514900 DQ512456 Cleve & Möller Stephanodiscus minutulus (Kützing) WTC1 West Lake Okoboji, IA, Stephanodiscaceae DQ514911 DQ512467 Cleve & Möller USA Stephanodiscus minutulus (Kützing) FHTC11 Fairport Harbor, Lake Erie, Stephanodiscaceae DQ514897 DQ512453 Cleve & Möller OH, USA Stephanodiscus minutulus (Kützing) Y95-6 Yellowstone Lake, WY, Stephanodiscaceae DQ514915 DQ512471 Cleve & Möller USA

Stephanodiscus minutulus (Kützing) Y98-1 Yellowstone Lake, WY, Stephanodiscaceae DQ514916 DQ512472 70 Cleve & Möller USA Stephanodiscus neoastraea H. Sneo4 Grober Müggelsee, Berlin, Stephanodiscaceae DQ514906 DQ512462 Håkansson & B. Hickel Germany Stephanodiscus niagarae Ehrenberg J95-16 Jackson Lake, WY, USA Stephanodiscaceae DQ514907 DQ512463 Stephanodiscus niagarae Ehrenberg OKA-A Okamanpeedan Lake, MN, Stephanodiscaceae DQ514908 DQ512464 USA Stephanodiscus reimerii Theriot WLO11 West Lake Okoboji, IA, Stephanodiscaceae DQ514909 DQ512465 USA Stephanodiscus yellowstonensis Y7 Yellowstone Lake, WY, Stephanodiscaceae DQ514910 DQ512466 Theriot USA Species Culture Locality Family SSU LSU strain Bellerochea malleus (Brightwell) CCMP143 North Pacific Ocean, outgroup DQ514845 DQ512392 Van Heurck Muroran, Japan Ditylum brightwellii (T. West) CCMP1810 Willapa Bay, Washington outgroup DQ512394 Grunow USA X85386 Helicotheca tamesis (Shrubsole) M. CCMP1760 Darien, Panama, near outgroup Ricard Samba River X85385 missing Lithodesmium undulatum Ehrenberg CCMP1806 outgroup DQ514683DQ514846 71 Table 2.3. Oligonucleotide primers used to amplify and sequence SSU rDNA, partial LSU

rDNA, psbC, and rbcL fragments from Thalassiosirales.

Name Marker Sequence (5' to 3') SSU11 SSU AAC CTG GTT GAT CCT GCC AGT SSU11+ SSU TGA TCC TGC CAG TAG TCA TAC GCT SSU301+ SSU ATC ATT CAA GTT TCT GCC C SSU515+ SSU TGG AAT GAG AAC AAT TTA A SSU1004+ SSU CGA AGA TGA TTA GAT ACC ATC G SSU1451+ SSU TGT GAT GCC CTT AGA TGT CCT GG ITS1DR2 SSU CCT TGT TAC GAC TTC ACC TTC C SSU1147– SSU AGT TTC AGC CTT GCG ACC ATA C SSU568– SSU CAG ACT TGC CCT CCA ATT GA

D1R1 LSU d1–d2 ACC CGC TGA ATT TAA GCA TA D2C2 LSU d1–d2 CCT TGG TCC GTG TTT CAA GA

psbC+1 psbC CAC GAC CWG AAT GCC ACC AAT psbC221+ psbC ACG CAT TGT TTC ACC ACC psbC499+ psbC ACG TGC CCA AGA GAA TGG TTT TG psbC–2 psbC ACA GGM TTY GCT TGG TGG AGT GG psbC587– psbC ATC TTG TTG GTG GTC ATA TTT GG psbC857– psbC CTT TGG TTA TGA CTG GCG TG

rbcL66+1 rbcL TTA AGG AGA AAT AAA TGT CTC AAT CTG rbcL527+ rbcL AAA ACA TTC CAA GGT CCT GCT rbcL1255–2 rbcL TTG GTG CAT TTG ACC ACA GT rbcL587– rbcL GTC TAA ACC ACC TTT TAA MCC TTC nd6+1 rbcL GTA AAT GGA TGC GTA TG dp72 rbcL AAA SHD CCT TGT GTW AGT YTC 1 Forward PCR amplification primer

2 Reverse PCR amplification primer

72 Table 2.4. Data partitions, models of DNA sequence evolution, and results of Bayesian analyses, including Bayes factor comparisons of alternative models.

Dataset Rate matrix Rate variation total runs generations harmonic Bayes factor (runs used) (burnin)mean ln B10 2ln B10 SSU GTR G+I 2 (2) 30 (15) –10,150 843* 1686* SSU stems / SSU loops Doublet / GTR G+I / G+I 2 (2) 10 (4) –9,307 SSU + LSU GTR G+I 2 (2) 20 (10) –16875 SSU stems / SSU loops / Doublet / GTR / GTR G+I / G+I / G+I 2 (2) 10 (5) –16536 339* 678* LSU chloroplast1 GTR G+I 4 (2) 80 (64) –24057 1 1092* 2184* chloroplast × codon GTR11- + GTR--2 (G+I)11- + (G+I)--2 2 (2) 40 (20) –22965 SSU stems / SSU loops / Doublet / GTR / GTR G+I / G+I / G+I / 4 (3) 40 (16) –39327 1 LSU / chloroplast × / GTR11- + GTR--2 (G+I)11- + (G+I)--2

codon 73 1 "chloroplast" refers to combined psbC + rbcL

* Bayes factor comparison favors the alternative, more complex model, based on Kass and Raftery (1995)

Chapter 3: Massive convergent evolution in morphology associated with the

colonization of fresh waters in diatoms

Abstract. Despite countless observations, little is known about the functional and adaptive importance underlying the renowned patterns and ornamentations in the siliceous cell walls of diatoms. Thus, an ultimate goal in diatom biology is to determine the adaptive value, if any, of their remarkable cell wall morphologies and to identify the selective forces driving morphological change. The order Thalassiosirales has high species diversity in marine and fresh waters, and phylogenetic analysis showed

unexpectedly strong evidence for multiple colonizations of freshwaters, two of which led

to substantial species diversification. We used the phylogeny of Thalassiosirales as the

basis for comparative analyses of several morphological characters previously thought to

support a single, primary origin of Thalassiosirales in fresh waters. We present

compelling evidence for convergent evolution in these characters upon, or in some cases,

shortly after the colonization of fresh waters. Two of the convergent traits involve the

strutted process, a structure involved in the production of β-chitin fibers that are thought

to enhance cell buoyancy in the water column, whereas other features are involved in the

formation of internal siliceous chambers. With little data on diatom functional

morphology, the adaptive advantages conferred by these traits remain uncertain.

Nevertheless, the strikingly high level of gross morphological convergence—sometimes

via recurrent evolutionary trajectories—and the correlated evolution of convergent traits

74 with freshwater colonization provide the first evidence of its kind for morphological

adaptation in diatoms.

Introduction

Diatoms are unicellular photoautotrophs within heterokont algae known most notably for their highly ornamented cell walls of opaline silica, called "frustules"

(Andersen, 2004). The intricacy and diversity of cell wall patterns in diatoms have long fascinated taxonomists, and more recently, have received attention from scientists interested in using diatoms for nanotechnology applications (Gordon et al., 2005).

Despite more than a century of observations with light microscopy and several decades of observations with scanning electron microscopy, little is known about the functional or adaptive importance underlying the spectacular morphological variation in diatom cell

wall morphology (Round et al., 1990). Their bewildering, yet seemingly inexplicable,

array of features might have been what led Charles Darwin to conclude the following:

"Few objects are more beautiful than the minute siliceous cases of the diatomaceae: were these created that they might be examined and admired under the higher powers of the microscope?" (Darwin, 1872). Thus, the "holy grail" of diatom biology remains to determine the adaptive value, if any, underlying this morphological variation, to identify the selective forces driving morphological change, and ultimately, to understand the genetic basis of the siliceous diatom cell wall.

The order Thalassiosirales is an excellent candidate for studies of morphological adaptation in diatoms. With ≥500 fossil and living species, it is speciose, morphologically diverse, and well characterized (e.g. Håkansson, 2002; Hasle, 1968; Hasle and Syvertsen,

75 1997; Julius, 2000; Round et al., 1990; Theriot and Serieyssol, 1994). For example, it is

the only order of diatoms that has had much of its morphology rendered into characters in

phylogenetic terms (Theriot and Serieyssol, 1994), and there are available phylogenetic

hypotheses for Thalassiosirales inferred from both morphology (Julius, 2000; Theriot,

1992) and DNA sequences (Alverson, 2006). Thalassiosirales has high species diversity in marine and freshwater habitats—a division of considerable interest in terms of total silica content of the cell wall because, across phylogenetic lines, marine diatoms contain an order of magnitude less silica than freshwater diatoms (Conley et al., 1989). The consequences of this disparity for detailed studies of morphological evolution have yet to be explored.

Within Thalassiosirales, genera are traditionally classified into the marine

Thalassiosiraceae or freshwater Stephanodiscaceae (Glezer and Makarova, 1986), recognizing that within each there has been some "leakage" (Round and Sims, 1980) of species between habitats. This classification was eclectically based, and ecological distribution was considered a primary distinguishing characteristic (Glezer and

Makarova, 1986). Although there was no explicit attempt to base this classification on phylogeny, there have been suggestions that the Stephanodiscaceae, in particular, is a natural evolutionary lineage. For example, the lacustrine stratigraphic sequence of

Stephanodiscaceae genera suggests that the entire Stephanodiscaceae traces to a single, successful colonization of fresh waters in the middle Miocene (Julius, 2000; Krebs, 1990;

Krebs and Bradbury, 1982; Krebs et al., 1987). Morphological and molecular data exist to test for monophyly of both Thalassiosiraceae and Stephanodiscaceae, and both data

76 agree that Thalassiosiraceae is non-monophyletic but strongly disagree about the

monophyly of Stephanodiscaceae.

Diatom taxonomy and classification are based almost exclusively on

morphological features of the siliceous cell wall, and monophyly of Stephanodiscaceae is

suggested by several such characters (Glezer and Makarova, 1986; Julius, 2000; Theriot

and Serieyssol, 1994). For example, most species in Stephanodiscaceae have prominent,

internally thickened branches of silica radiating from the valve center. Thickened

branches are, in turn, a necessary component for the network of chambers on the interior

cell wall of many Cyclotella and Cyclostephanos species (Håkansson, 2002; Julius, 2000;

Theriot and Serieyssol, 1994). These chambers are partially covered and can be complex,

three-dimensional structures. In contrast, these features are either diminutive or entirely

absent in all species of Thalassiosiraceae. Monophyly of Stephanodiscaceae is further

suggested by ultrastructural features of the fultoportula (or "strutted process"), a cell wall

structure unique to Thalassiosirales (Hasle, 1968; Theriot and Serieyssol, 1994). Most species have at least one ring of strutted processes around the valve margin and one or

more strutted processes on the valve face (Hasle and Syvertsen, 1997; Stoermer and

Julius, 2003). Strutted processes are involved in the production of β-chitin fibers: the

exceptionally long fibers extruded through marginal strutted processes are thought to

enhance cell buoyancy in the water column, whereas those from central-area strutted

processes facilitate colony formation in some species (Herth, 1978; Herth, 1979; Herth

and Barthlott, 1979; Walsby and Xypolyta, 1977). The number, arrangement, and

ultrastructure of strutted processes are important taxonomic characters, particularly for

distinguishing between Stephanodiscaceae and Thalassiosiraceae (Julius, 2000; Theriot

77 and Serieyssol, 1994). At least two diagnostic, and presumably derived, ultrastructural features of the strutted process are unique to Stephanodiscaceae and therefore indicative of monophyly (Julius, 2000; Theriot and Serieyssol, 1994, and see below). In short, the

complexity and apparent phylogenetic congruence of all these features—internally thickened branches, a chambered cell wall, and strutted process ultrastructure—is such that no one has considered that they could have evolved more than once (Julius, 2000;

Theriot and Serieyssol, 1994). Somewhat surprisingly then, the phylogenetic hypotheses of Alverson (2006) showed strong evidence for multiple colonizations of fresh waters by

Thalassiosirales, two of which led to substantial species diversification and separated the

Stephanodiscaceae into two distinct and distantly related lineages. These results suggest that the set of characters previously thought to support monophyly of Stephanodiscaceae were independently derived in the two freshwater lineages.

Thus, Thalassiosirales is well suited for studies of adaptation in that it has a densely sampled and well-supported phylogenetic hypothesis (Alverson, 2006), it spans a significant ecological boundary, and has a set of complex, well-characterized morphological traits. The comparative method provides a powerful tool for identifying the origin and evolution of novel traits and for recognizing correlative patterns between traits and extrinsic variables—a process that ultimately can lead to the discovery of adaptations (Felsenstein, 1985; Harvey and Pagel, 1991). We use the phylogeny of

Thalassiosirales inferred from nuclear and chloroplast DNA sequences (Alverson, 2006)

as the basis for comparative analyses of five morphological characters previously thought

to support monophyly of the Stephanodiscaceae (Julius, 2000; Theriot and Serieyssol,

1994). We present compelling evidence for convergent evolution in these traits, some via

78 strikingly similar evolutionary trajectories, and show that most of these characters are evolving in correlation with habitat type.

Materials and methods

Species phylogeny

The Thalassiosirales phylogeny included 83 taxa (79 ingroup, 4 outgroup) chosen to represent the full range of extant morphological diversity in the lineage (Alverson,

2006). The phylogeny was inferred from a combined Bayesian analysis of 5184 nt, which included chloroplast-encoded psbC and rbcL (2698 nt) and nuclear-encoded SSU and partial LSU rDNA (2486 nt) (Alverson, 2006). Parsimony and Bayesian analyses gave similar tree topologies, with Bayesian posterior probability values tending to be higher than parsimony nonparametric bootstrap values for the same node (Alverson, 2006).

Nevertheless, topological conflicts were unsupported in one or both analyses (Alverson,

2006), so we chose the fully-resolved Bayesian majority rule consensus tree (with branch lengths) for ancestral state reconstructions and character correlation analyses.

Morphological character observations

Diatom cultures were isolated as individual cells and grown in batch culture at 14o

C on a 12:12 light-dark cycle. Marine species were grown in f/2 medium (Guillard, 1975;

Guillard and Ryther, 1962), and freshwater species were grown in COMBO medium

(Kilham et al., 1998). Cells were boiled in 30% hydrogen peroxide for one hour, rinsed

several times, then dried onto filtered onto 0.22 µm nitrocellulose filters for observation

with a Hitachi S-4500 field emission scanning electron microscope (FE-SEM).

Five frustule characters traditionally used to diagnose the freshwater

Stephanodiscaceae (Julius, 2000; Theriot and Serieyssol, 1994) initially were coded as

79 multistate characters and subsequently converted to binary codings for correlation analyses. The five morphological characters include the following:

1. Radial costae (character 6b in Theriot and Serieyssol, 1994): internal feature,

the branches of silica that radiate outwards from the valve center. These

branches are either poorly developed and undifferentiated or only slightly

thickened in Thalassiosiraceae (code = 0) (Fig. 3.2a, e), whereas these

branches are heavily silicified and internally expanded—both horizontally

and vertically—in most Stephanodiscaceae (code = 1) (Fig. 3.2b–d, f, g

[arrows]). Thickened radial branches (or "costae") are a necessary component

of central and marginal chambers (see below).

2. Central chamber (character 6d in Theriot and Serieyssol, 1994): internal

feature, due to presence of internally thickened radial branches (character 1)

and central laminae, which are thin, siliceous membranes that extend distally

from center of the cell outward toward the valve margin (Fig. 3.3b, c, f, g

[arrows]). Well-developed central laminae extend laterally to join adjacent,

thickened costae, forming hollow chambers beneath them. Within

Thalassiosirales, this feature is known only from Stephanodiscaceae genera

that are either exclusively freshwater (e.g. Tertiarius, Håkansson and

Khursevich, 1997) or mixed marine–freshwater (e.g. Cyclotella). Central

chambers were coded as present if a central lamina was present in any form

(code = 1) (Fig. 3.3b, c, f, g [arrows]), or absent (code = 0) (Fig. 3.3a, e

[arrows]).

80 3. Marginal chamber (character 6e in Theriot and Serieyssol, 1994): internal

feature similar to central chamber but occurring on the margin of the valve,

formed by intersection of internally thickened radial branches (character 1)

and marginal laminae, which are thin, siliceous membranes that extend

medially from the valve margin toward the valve center (Fig. 3.3—b–d, f, g

[arrows]). Like the central lamina, well-developed marginal laminae extend

laterally to join adjacent, thickened costae, forming hollow chambers beneath

them. Some species or populations have a marginal lamina but no central

lamina (e.g. some populations of Cy. menegheniana), or vice-versa,

suggesting that these two characters are developmentally independent. Within

Thalassiosirales, this feature is known only from genera that are either

exclusively freshwater (e.g. Tertiarius and Pliocaenicus) or mixed marine–

freshwater (e.g. Cyclotella). Marginal chambers were coded as present, if a

marginal lamina was present in any form (code = 1) (Fig. 3.3—b–d, f, g

[arrows]), or absent (code = 0) (Fig. 3.3a, e [arrows]).

4. Strutted process cowling (character 10d in Theriot and Serieyssol, 1994):

internal feature, a ridge of silica surrounding the satellite pore, away from the

central tube of the strutted process. This structure is either completely absent

(Fig. 3.4—Thalassiosira minuscula [arrow]) or appearing as a slight siliceous

rim in most Thalassiosiraceae (Fig. 3.4—Porosira glacialis [arrow]) (code =

0). In contrast, strutted process cowlings in Stephanodiscaceae are

substantially more developed, often with raised rims that appear as

"shoulders," which in some species almost completely obscure the satellite

81 pores (code = 1) (Fig. 3.4—Cyclotella atomus, Stephanodiscus minutulus

[arrows]).

5. Satellite pore cover (character 10b in Theriot and Serieyssol, 1994): internal

feature, a siliceous outgrowth from the central tube of the strutted process that

partially occludes the satellite pores. The junction of the central tube and

satellite pore cover (SPC) is narrow in Thalassiosiraceae (Fig. 3.5—P.

glacialis, Skeletonema japonicum [arrows]) and much broader in

Stephanodiscaceae (Fig. 3.5—Cyclotella stylorum, Cyclostephanos invisitatus

[arrows]). Thus, satellite pore covers vary in form from absent to small,

narrow tabs (code = 0) to broad ridges (code = 1).

Character mapping

Maximum likelihood reconstructions (Pagel, 1999; Schluter et al., 1997) of habitat type were based on the Asymmetrical Markov k-state 2 parameter model, which allows for different forward (0Æ1) and backward (1Æ0) rates, representing the rates of freshwater (1) and marine (0) colonization events, respectively. All model parameters were estimated from the data, and the ancestral state reconstruction at any given node was considered significantly better than the alternative if its likelihood score exceeded the alternative reconstruction by two log-likelihood units (Pagel, 1999). Ancestral reconstructions were done with the Mesquite software package (Maddison and Maddison,

2006).

The maximum likelihood method described by Pagel (1994) and implemented in the software program Discrete (Pagel, 1998) was used to test for correlated evolution between habitat type and each of the five morphological characters. Briefly, two models

82 are fit to the data and compared using the likelihood ratio test (LRT); the LRT statistic is

compared to a chi-square distribution (df = 4) to determine whether the eight-parameter

dependent model provides a significantly better fit than the more restricted, four-

parameter independent model (Pagel, 1994). Correlated evolution between the two traits

is inferred when the more complex, dependent model provides a significantly better fit

than the simpler, hierarchically nested independent model (Pagel, 1994).

Results

Maximum likelihood mapping showed that the common ancestor of modern

Thalassiosirales was marine, and there have been three independent colonizations of fresh

waters (Fig. 3.1). Two of these freshwater colonizations led to substantial species diversification as represented by the two Stephanodiscaceae lineages (Fig. 3.1, Clades A and B), which together account for the vast majority of freshwater species diversity in

Thalassiosirales (Mann, 1999; Round et al., 1990; Round and Sims, 1980). These two lineages are separated by six bipartitions with BPP ≥ 0.95 (Fig. 3.1, Clades A and B), and

Bayesian and parametric bootstrap alternative topology tests both rejected monophyly of

Stephanodiscaceae (i.e. the hypothesis of a single, primary origin of Thalassiosirales in fresh waters) (Alverson, 2006). Stephanodiscaceae Clade A is ancestrally freshwater but includes a mixture of obligate marine (e.g. Cy. stylorum), obligate freshwater (Cyclotella distinguenda), and euryhaline species (e.g. Thalassiosira pseudonana) species, whereas all species in the younger Clade B are obligate freshwater (Fig. 3.1) (Hasle and

Syvertsen, 1997; Stoermer and Julius, 2003). Diatom classification is based primarily on cell wall morphology (e.g. Round et al., 1990), and as implied by the classification of

Clade A and B species into the genus Cyclotella and the family Stephanodiscaceae,

83 species in the two clades share several independently derived morphological features

(Figs. 3.1–3.5).

Cell wall morphogenesis in Thalassiosirales proceeds outward from a central ring,

resulting in cells with near-radial symmetry (Round et al., 1990). A series of radial

branches are formed initially (Round et al., 1990), and among Thalassiosiraceae, these

primary branches remain largely undifferentiated internally (Fig. 3.2a, e). Radial

branches in Stephanodiscaceae, by contrast, are expanded into robust, heavily silicified

ribs (or "costae"). Within Clade A, only T. pseudonana lacks this feature, indicating that

the derived, heavily silicified internal costae evolved shortly after the colonization of

fresh waters. (Figs. 3.1, 3.2b–d). Internally thickened costae evolved a second time in the

common ancestor of Clade B, this time simultaneous with the colonization of fresh

waters (Figs. 3.1, 3.2f, g). The derived costa morphology was subsequently lost in

Stephanodiscus (Figs. 3.1, 3.2h). Presence of internally thickened costae was

significantly correlated with freshwater habitat (p < 0.05) (Table 1).

One feature traditionally used to diagnose Cyclotella and Cyclostephanos is the presence of marginal chambers (Håkansson, 2002; Stoermer and Julius, 2003). A chamber is formed when the gap between adjacent costae is bridged by a thin siliceous canopy (or "lamina"), creating a hollow chamber beneath it (Figs. 3.2b, c, g, 3.3b–d, f, g).

Chambers can occur on the valve margin, near the valve center, or in both places (Fig.

3.1), and though internally thickened costae are a prerequisite for chamber formation, many species that have this feature lack chambers (Figs. 3.2f, 3.3h). All species of

Thalassiosiraceae—in basal branches (e.g. Porosira) and in branches intermediate between Clades A and B (e.g. Skeletonema and Bacterosira)—lack marginal chambers,

84 central chambers, and both of the constituent chamber parts (Figs. 3.1, 3.2a, e). The cell

wall interior appears as a smooth, undifferentiated surface in all of the Thalassiosiraceae

(Figs. 3.2a, e, 3.3a, e), which contrasts the highly differentiated and sometimes elaborately chambered internal morphology of Stephanodiscaceae (Figs. 3.2b–d, f, g,

3.3b–d, f–h). Presence of marginal and central chambers in Cy. distinguenda (Figs. 3.2b,

3.3b) indicates that these characters evolved in Clade A soon after the colonization of

fresh waters (Fig. 3.1). Several species in Clade A retained internally thickened costae

but lost one (e.g. Cy. atomus) or both chamber types (e.g. Cyclotella cryptica) (Fig. 3.1),

and both chamber types were fully retained in two of the secondarily marine lineages

within Clade A (Fig. 3.1). In Clade B, marginal and central chambers evolved in the

lineage containing Cyclotella bodanica and Cyclotella ocellata (Figs. 3.1, 3.3f, g), indicating that like Clade A, this feature evolved shortly after the colonization of fresh

waters (Fig. 3.1). In this tree, both chamber types were lost in subsequent lineages with

(Cyclostephanos) and without (Stephanodiscus) internally thickened costae (Figs. 3.1,

3.3f–h). Notably, many Cyclostephanos species not included in these analyses possess

chambers (Håkansson, 2002; Theriot and Serieyssol, 1994). Presence of marginal

chambers was weakly correlated with freshwater habitat (p < 0.10), whereas the

correlation between presence of central chambers and freshwater habitat was statistically

significant (p < 0.05) (Table 1).

Strutted process ultrastructure differs markedly between Thalassiosiraceae and

Stephanodiscaceae (Figs. 3.4, 3.5). Two of the more conspicuous differences concern the

2–6 holes (or "satellite pores") that flank the central tube of the strutted process. Strutted

process "cowlings" are accessory structures that surround satellite pores at their base

85 (Theriot and Serieyssol, 1994). Without exception, cowlings occur as either poorly

developed rims (Fig. 3.4—P. glacialis, Thalassiosira punctigera, and Bacterosira

bathyomphala) or are completely absent (Fig. 3.4—Lauderia annulata, Thalassiosira sp.,

T. minuscula) across the entire Thalassiosiraceae. In contrast, the strutted process

cowlings are highly derived and much more prominent in the two Stephanodiscaceae

lineages (Figs. 3.1, 3.4, Clades A and B). The strutted processes of T. pseudonana have

three satellite pores, each surrounded by a raised, well-developed cowling with a weakly

developed shoulder (Fig. 3.4), indicating that evolution of the derived cowling

morphology in Clade A was coincident with the colonization of fresh waters (Fig. 3.1). In

the remaining Clade A species, cowlings have an even more inflated shoulder, which

forms an enclosed chamber around the satellite pore (Fig. 3.4— Cyclotella gamma, Cy.

distinguenda, Cy. atomus, and Cy. menegheniana). A strikingly similar evolutionary

trajectory is seen in Clade B. With a well-developed rim and slightly inflated shoulder,

the cowling morphology of Cyclotella pseudostelligera (Clade B) resembles that of T.

pseudonana (Clade A) (Figs. 3.1, 3.4). Thus, a similarly derived cowling structure independently evolved in Clades A and B upon colonization of fresh waters (Fig. 3.4),

and more striking yet is the recurrent evolution of the inflated shoulder, which in pictures

presented here, is most visible in the comparison of Cy. menegheniana (Clade A) and St. minutulus (Clade B) (Fig. 3.4). Again, the similarly derived strutted process ultrastructure

in Clades A and B—as illustrated by these two species—is unmatched outside of the

Stephanodiscaceae.

In most species, the central tube of the strutted process bears an ornamentation directly above each satellite pore that covers the satellite pore to a greater (e.g. L.

86 annulata) or lesser (e.g. B. bathyomphala) degree (Fig. 3.4). Satellite pore covers appear

as small—usually pointed or rounded—tabs in most Thalassiosiraceae (Fig. 3.5), though

the structure is greatly diminished (Fig. 3.5—Thalassiosira sp. and Sk. japonicum) or entirely absent (Fig. 3.4—B. bathyomphala) in some species. In contrast, the satellite

pore covers have nearly uniform morphology across the Stephanodiscaceae, appearing as

broadly attached, ridgelike structures (Fig. 3.5, Clades A and B). Stated differently, an

individual satellite pore cover spans a much greater proportion of the central tube

diameter for species within Stephanodiscaceae (Fig. 3.5, Clades A and B). Both T.

pseudonana (Fig. 3.4, Clade A) and Cyclotella stelligera (Fig. 3.5, Clade B) have

derived, broadly attached satellite pore covers, indicating that this feature independently

evolved in Clades A and B upon colonization of fresh waters (Fig. 3.1). The

ultrastructure remains fairly uniform across Clades A and B, except in the Cy.

bodanica+Cy. ocellata clade, which despite having a highly autapomorphic satellite pore

cover, retains the plesiomorphic condition of a broad attachment point to the central tube

(Fig. 3.5). Both strutted process characters were significantly correlated with freshwater

habitat (p < 0.05) (Table 1).

Discussion

Species in the genus Cyclotella and family Stephanodiscaceae were classified

together based on their occurrence in fresh waters (Glezer and Makarova, 1986) and

shared presence of numerous morphological characters of the siliceous cell wall (Glezer

and Makarova, 1986; Håkansson, 2002; Julius, 2000; Theriot and Serieyssol, 1994). A

densely sampled, multi-gene phylogenetic analysis of Thalassiosirales strongly supported

the separation of Stephanodiscaceae into two distinct and distantly related lineages,

87 which represent the two major, independent diversifications of thalassiosiroids into fresh waters (Alverson, 2006). This result has important implications for our understanding of the pattern and timing of freshwater colonization by thalassiosiroids. Independent evidence (e.g. from a molecular clock study) would provide valuable insight into the absolute ages of Clades A and B by showing whether there is a large temporal gap between the origins of Clades A and B that is thus far undiscovered in the fossil record.

Alternatively, a molecular clock study might support a more punctuated evolutionary history whereby major lineages within Thalassiosirales—including Clades A and B— arose nearly simultaneously, resulting in a stratigraphic sequence irresolvable by the fossil record. The latter seems unlikely given the fully resolved, strongly supported phylogeny (Fig. 3.1, Alverson, 2006) and high resolving capacity of the freshwater stratigraphic record (Theriot et al., 2006). The morphological implications of non- monophyly of Stephanodiscaceae are more striking because of the direct conflict between the inferred phylogeny (Fig. 3.1) and the presumed phylogenetic signal in the suite of characters indicative of monophyly for Stephanodiscaceae (Figs. 3.1–3.5, Julius, 2000;

Theriot and Serieyssol, 1994). The distribution of these morphological characters across the phylogeny and their correlation to habitat type, however, show that they are independently derived in the two Stephanodiscaceae lineages (Fig. 3.1).

Species in the two Stephanodiscaceae lineages independently evolved heavily silicified internal branches (costae) upon (Clade B) or shortly after (Clade A) the colonization of fresh waters (Figs. 3.1, 3.2). The presence of this character in Clades A and B and its correlated evolution with fresh waters does not reflect the general trend of higher silica content in freshwater diatoms (Conley et al., 1989). For example, costae are

88 internally thickened throughout Stephanodiscaceae except in Stephanodiscus, which has heavily silicified cell walls but lacks internally thickened costae (Figs. 3.1, 3.2).

Stephanodiscus species differentially allocate excess silica elsewhere, for example, into prominent external spines or areolar occlusions (Genkal and Håkansson, 1990; Theriot,

1987). Thus, the allocation of silica into radial branches appears to be a genetically

controlled trait, not ecophenotypic variation. The convergence of this character in Clades

A and B and its significant correlation with their occurrence in fresh waters suggests that it does confer some selective advantage. However, any potential adaptive value of

internal thickenings alone is unclear.

Internally thickened costae are a prerequisite for the formation of internal

chambers, which are formed by the development of a siliceous membrane (lamina) across

adjacent costae (Figs. 3.2, 3.3). Many heavily silicified species that possess costae also

lack laminae (e.g. Cy. stelligera and Cyclostephanos tholiformis) and vice-versa (e.g.

some populations of Cy. cryptica), which suggests that lamina development is genetically

controlled and not simply an artifact of the overall higher silica content in freshwater

diatoms (Conley et al., 1989). The two origins of internal chambers presented here are the

only two we know of among centric diatoms. In fact, outside of Stephanodiscaceae,

chambered cell walls are only found in genera within the distantly related raphid pennate

lineage (e.g. Caloneis, Pinnularia, Mastogloia, and Surirella) (Alverson et al., 2006 for

diatom phylogeny; Round et al., 1990). Although internal chambers occur in both marine

and freshwater raphid pennates (Round et al., 1990; Ruck and Kociolek, 2004), the

phylogenetic pattern of chamber gain and loss and any possible correlation to habitat type

are unknown in this lineage. Many chamber-bearing raphid pennates have been shown to

89 localize mitochondria to their marginal chambers, such that each marginal chamber bears either its own mitochondrion or a portion of a larger, branched mitochondrion (Edgar,

1980; Stoermer et al., 1964). Diatoms might be localizing mitochondria to chambers on the cell periphery to satisfy increased metabolic demands in this area of the cell (Edgar,

1980). It is unknown whether the marginal chambers of Stephanodiscaceae contain mitochondria, but if present—especially in marginal chambers—mitochondria would be immediately adjacent to strutted processes, which lie on the heavily silicified costae in most species (Figs. 3.2, 3.3).

Thalassiosirales is the predominant diatom lineage, in both cell abundance and species diversity, in the plankton of fresh waters (Hasle and Syvertsen, 1997; Mann,

1999; Round and Sims, 1980; Stoermer and Julius, 2003). Among freshwater planktonic diatoms, thalassiosiroids are uniquely endowed with strutted processes, siliceous structures involved in the production of β-chitin fibers (Herth, 1978; Herth, 1979; Herth and Barthlott, 1979). The β-chitin fibers extruded from marginal strutted processes should increase drag and therefore decrease sinking rate (Round et al., 1990; Walsby and

Xypolyta, 1977). Thus, the strutted process is frequently cited as the most likely factor that led to the predominance of thalassiosiroids in modern freshwater plankton

assemblages (Round and Sims, 1980). We showed compelling evidence for convergent

evolution of two strutted process ultrastructural features in the two Stephanodiscaceae

lineages upon colonization of fresh waters (Figs. 3.1, 3.4, 3.5). One feature, the strutted

process cowling, evolved along strikingly similar evolutionary trajectories in the two

lineages. More specifically, basal species evolved a similarly derived, raised cowling

upon colonization of fresh waters (Figs. 3.1, 3.4—T. pseudonana and Cy.

90 pseudostelligera), followed by the subsequent evolution of a nearly identical inflated

"shoulders" in descendant lineages (Fig. 3.4). This shoulder creates a cavity that fully encloses the satellite pore in some species (Fig. 3.4). The near identity of this character in

Clades A and B is best illustrated in comparisons of Cy. menegheniana (Clade A) and St. minutulus (Clade B) (Figs. 3.4, 3.5). A second strutted process character—a broadly attached, ridgelike satellite pore cover (Fig. 3.5)—independently evolved in Clades A and

B upon colonization of fresh waters (Fig. 3.1). The gross morphological convergence of strutted process characters, their significant correlation with fresh waters, and the remarkably similar evolutionary trajectories of cowling convergence in Clades A and B strongly suggest that these characters were under strong directional selection upon colonization of fresh waters. Each of these structures is intimately associated with the strutted process satellite pores (Figs. 3.4, 3.5), suggesting a possible overlap or redundancy in function. Unfortunately, the functional roles of strutted process features

(e.g. satellite pores and central tube morphology) are entirely uncharacterized, which prohibits any interpretation about the adaptive value of the independently derived features in Stephanodiscaceae.

In light of these findings, the buoyancy hypothesis of Round and Sims (1980) deserves rigorous testing. A controlled set of culture experiments across a range of salinity treatments will determine whether freshwater thalassiosiroids are, in fact, more buoyant or better able to control their buoyancy in the water column. Tighter control of buoyancy would prevent sinking out of the photic zone or possibly allow them to adjust to the spatially dynamic light and nutrient conditions in the water column. The former could provide a selective advantage in fresh waters, where sinking rates might be faster

91 due to the overall lower density of fresh waters. Although β-chitin production might

confer some selective advantage over other diatom lineages, the characteristics of β-chitin production and excretion appear to be highly similar in marine and freshwater thalassiosiroids (Herth, 1978; Herth and Barthlott, 1979). In light of this, and with little or no data on strutted process functional morphology, we cannot begin to speculate about the potential adaptive value of the convergent strutted process characters found in

Stephanodiscaceae (Figs. 3.4, 3.5). As a starting point, possible differences in β-chitin production and fiber morphology between marine and freshwater taxa should be explored. Finally, if mitochondria are in fact found in the marginal chambers of

Stephanodiscaceae, the possibility of a close association between mitochondria and marginal strutted processes should be explored.

In summary, despite countless observations, little remains known about the functional and adaptive importance underlying the renowned variation in diatom cell wall morphology. We presented strong comparative evidence for massive morphological convergence in Thalassiosirales associated with the colonization of fresh waters. With

little information on diatom functional morphology, the exact advantages conferred by

the convergent traits is uncertain and warrants further study. It is also unclear whether

freshwater, which encompasses a broad range of chemical–physical properties, is the

selective force driving morphological change in these diatoms, or whether freshwater is a

surrogate measure for one or more associated environmental variables. Nevertheless, the

strikingly high level of gross morphological convergence—sometimes via recurrent

evolutionary trajectories—and the correlated evolution of convergent traits with

92 freshwater colonization provide the first evidence of its kind for morphological adaptation in diatoms.

Acknowledgements

This research was supported by NSF PEET grant (DEB-0118883) and NSF Dissertation

Improvement Grant (DEB-0407815).

Literature Cited

Alverson, A. J. 2006. Phylogeny and evolutionary ecology of thalassiosiroid diatoms.

PhD thesis. The University of Texas at Austin, Plant Biology Graduate Program.

Alverson, A. J., J. J. Cannone, R. R. Gutell, and E. C. Theriot. 2006. The evolution of

elongate shape in diatoms. Journal of Phycology 42:655–668.

Andersen, R. A. 2004. Biology and systematics of heterokont and haptophyte algae.

American Journal of Botany 91:1508–1522.

Conley, D. J., S. S. Kilham, and E. Theriot. 1989. Differences in silica content between

marine and freshwater diatoms. Limnology & Oceanography 34:205–213.

Darwin, C. 1872. The Origin of Species, 6th edition. J.M. Dent & Sons, London.

Edgar, L. A. 1980. Fine structure of Caloneis amphisbaena (Bacillariophyceae). Journal

of Phycology 16:62–72.

Felsenstein, J. 1985. Phylogenies and the comparative method. American Naturalist

125:1–15.

Genkal, S. I., and H. Håkansson. 1990. Problem of distinguishing the newly described

diatom genus Pseudostephanodiscus. Diatom Research 5:15–24.

Glezer, Z. I., and I. V. Makarova. 1986. New order and family of diatoms

(Bacillariophyta). Botanicheskii Zhurnal (Leningrad) 71:673–676.

93 Gordon, R., F. Sterrenberg, and K. Sandhage (eds) 2005. A Special Issue on Diatom

Nanotechnology. Journal of Nanoscience and Nanotechnology, 178 pages.

Guillard, R. R. L. 1975. Culture of phytoplankton for feeding marine invertebrates. Pages

26–60 in Culture of Marine Invertebrate Animals (W. L. Smith, and M. H.

Chanley, eds.). Plenum Press, New York.

Guillard, R. R. L., and J. H. Ryther. 1962. Studies of marine planktonic diatoms. I.

Cyclotella nana Hustedt and Detonula confervacea (Cleve) Gran. Canadian

Journal of Microbiology 8:229–239.

Håkansson, H. 2002. A compilation and evaluation of species in the general

Stephanodiscus, Cyclostephanos and Cyclotella with a new genus in the family

Stephanodiscaceae. Diatom Research 17:1–139.

Håkansson, H., and G. Khursevich. 1997. Tertiarius gen. nov., a new genus in the

Bacillariophyceae, the transfer of some cyclotelloid species and a comparison to

closely related genera. Diatom Research 12:19–33.

Harvey, P. H., and M. Pagel. 1991. The Comparative Method in Evolutionary

BiologyOxford University Press, Oxford.

Hasle, G. R. 1968. The valve processes of the centric diatom genus Thalassiosira. Nytt

Magasin for Botanikk 15:193–201.

Hasle, G. R., and E. E. Syvertsen. 1997. Marine Diatoms. Pages 5–386 in Identifying

Marine Phytoplankton (C. R. Tomas, ed.) Academic Press, San Diego.

Herth, W. 1978. Special chitin-fibril-synthesizing apparatus in centric diatom Cyclotella.

Naturwissenschaften 65:260–261.

94 Herth, W. 1979. Site of beta-chitin fibril formation in centric diatoms. 2. Chitin-forming

cytoplasmic structures. Journal of Ultrastructure Research 68:16–27.

Herth, W., and W. Barthlott. 1979. Site of beta-chitin fibril formation in centric diatoms.

1. Pores and fibril formation. Journal of Ultrastructure Research 68:6–15.

Julius, M. L. 2000. Phylogeny of the Cyclostephanoid Diatoms: An Investigation of their

Morphology and Stratigraphy. PhD thesis. University of Michigan, Natural

Resources and Environment, 209 pages.

Kilham, S. S., D. A. Kreeger, C. E. Goulden, and S. Lynn. 1998. COMBO: A defined

freshwater culture medium for algae and zooplankton. Hydrobiologia 377:147–

159.

Krebs, W. N. 1990. The biochronology of freshwater planktonic diatom communities in

western North America. Pages 485–499 in Proceedings of the 11th International

Diatom Symposium (J. P. Kociolek, ed.) California Academy of Sciences, San

Francisco.

Krebs, W. N., and J. P. Bradbury. 1982. Neogene Lacustrine Diatom Biostratigraphy of

Western Snake River Basin, Idaho and Oregon. AAPG Bulletin 66:1693–1693.

Krebs, W. N., J. P. Bradbury, and E. Theriot. 1987. Neogene and Quaternary lacustrine

diatom biochronology western USA. Palaios 2:505–513.

Maddison, W. P., and D. R. Maddison. 2006. Mesquite: a modular system for

evolutionary analysis, version 1.1.

Mann, D. G. 1999. Crossing the Rubicon: the effectiveness of the marine/freshwater

interface as a barrier to the migration of diatom germplasm. Pages 1–21 in

95 Proceedings of the 14th International Diatom Symposium (S. Mayama, M. Idei,

and I. Koizumi, eds.). Koeltz Scientific Books, Koenigstein.

Pagel, M. 1994. Detecting correlated evolution on phylogenies: a general method for the

comparative analysis of discrete characters. Proceedings of the Royal Society of

London Series B-Biological Sciences 255:37–45.

Pagel, M. 1998. Discrete, version 4.0. Program distributed by the author. School of

Animal and Microbial Sciences, University of Reading.

Pagel, M. 1999. The maximum likelihood approach to reconstructing ancestral character

states of discrete characters on phylogenies. Systematic Biology 48:612–622.

Round, F. E., R. M. Crawford, and D. G. Mann. 1990. The Diatoms: Biology &

Morphology of the GeneraCambridge University Press, Cambridge.

Round, F. E., and P. A. Sims. 1980. The distribution of diatom genera in marine and

freshwater environments and some evolutionary considerations. Pages 301–320 in

Proceedings of the Sixth Symposium on Recent and Fossil Diatoms (R. Ross, ed.)

Otto Koeltz Science Publishers, Hirschberg.

Ruck, E. C., and J. P. Kociolek. 2004. Preliminary Phylogeny of the Family Surirellaceae

(Bacillariophyta)J. Cramer, Berlin.

Schluter, D., T. Price, A. O. Mooers, and D. Ludwig. 1997. Likelihood of ancestor states

in adaptive radiation. Evolution 51:1699–1711.

Stoermer, E. F., and M. L. Julius. 2003. Centric Diatoms. Pages 559–594 in Freshwater

Algae of North America (J. D. Wehr, and R. G. Sheath, eds.). Academic Press,

Amsterdam.

96 Stoermer, E. F., H. S. Pankratz, and R. W. Drum. 1964. The fine structure of Mastogloia

grevillei Wm. Smith. Protoplasma 70:1–12.

Theriot, E. 1987. Principal component analysis and taxonomic interpretation of

environmentally related variation in silicification in Stephanodiscus

(Bacillariophyceae). British Phycological Journal 22:359–374.

Theriot, E., and K. Serieyssol. 1994. Phylogenetic systematics as a guide to

understanding features and potential morphological characters of the centric

diatom family Thalassiosiraceae. Diatom Research 9:429–450.

Theriot, E. C. 1992. Clusters, species concepts, and morphological evolution of diatoms.

Systematic Biology 41:141–157.

Theriot, E. C., S. C. Fritz, C. Whitlock, and D. J. Conley. 2006. Late Quaternary rapid

morphological evolution of an endemic diatom in Yellowstone Lake, Wyoming.

Paleobiology 32:38–54.

Walsby, A. E., and A. Xypolyta. 1977. Form resistance of chitan fibers attached to cells

of Thalassiosira fluviatilis Hustedt. British Phycological Journal 12:215–223.

97 Figure 3.1. Phylogeny of Thalassiosirales inferred from combined, mixed-model

Bayesian analysis of DNA sequence data from the chloroplast (psbC and rbcL) and

nuclear (SSU and partial LSU rDNA) genomes (Alverson, 2006). Dashed branches are

supported by Bayesian posterior probability values less than 0.95. Pie graphs at internal

nodes show the relative maximum likelihood support for marine (white) and freshwater

(black) ancestral state reconstructions. Asterisks (*) indicate that the more strongly

supported ancestral state is significantly better than the alternative state. Five binary-

coded characters traditionally thought to diagnose a single, monophyletic freshwater

lineage follow the species name, in the following order: radial costae, central chamber,

marginal chamber, strutted process cowling, and satellite pore cover. The two independently derived Stephanodiscaceae lineages (Clades A and B) are highlighted to

facilitate presentation in text and figures. Generic abbreviations are: Bacterosira (Ba.),

Cyclostephanos (Cs.), Cyclotella (Cy.), Detonula (De.), Lauderia (La.), Minidiscus (M.),

Porosira (P.), Skeletonema (Sk.), Stephanodiscus (St.), Thalassiosira (T.).

98 Figure 3.1

99 Figure 3.2. Internal view showing morphological variation in radial branches in

Thalassiosirales, focusing on the differentiated and heavily silicified radial branches

("costae") in the predominantly freshwater Stephanodiscaceae (Clades A and B—see Fig.

1). Arrows highlight heavily silicified costae, if present. Pictured species include: a)

Thalassiosira sp., b) Cyclotella distinguenda, c) Cyclotella gamma, d) Cyclotella

menegheniana, e) Thalassiosira nodulolineata, f) Cyclotella stelligera, g) Cyclotella

bodanica, and h) Stephanodiscus neoastraea.

100 Figure 3.2

101 Figure 3.3. Presence and absence of internal marginal chambers in Thalassiosirales, focusing on the independently derived chambers present only in species of the

predominantly freshwater Stephanodiscaceae (Clades A and B—see Fig. 1). Upward and

downward arrows highlight marginal and central laminae, respectively. These thin

siliceous membranes span adjacent costae, forming chambers beneath them. Pictured species include: a) Thalassiosira guillardii, b) Cyclotella distinguenda, c) Cyclotella sp., d) Cyclotella menegheniana, e) Thalassiosira antarctica, f) Cyclotella bodanica, g)

Cyclotella ocellata, and h) Cyclostephanos invisitatus.

102 Figure 3.3

103

Figure 3.4. Ultrastructure of the strutted process "cowling" in Thalassiosirales. The central tube of the strutted process is flanked by 2–6 pores. The cowling is the siliceous structure—distal to the central tube—that surrounds each pore. The structure is poorly developed (e.g. P. glacialis and B. bathyomphala) or absent (e.g. Lauderia annulata and

Thalassiosira sp.) in Thalassiosiraceae and prominent in the primarily freshwater

Stephanodiscaceae, concealing the satellite pore in some cases (e.g. Cyclotella gamma and Stephanodiscus neoastraea). Diameter across the entire strutted process is 200–600 nm.

104

Figure 3.5. Ultrastructure of satellite pore covers in Thalassiosirales. Satellite pores are

perforations (pores) adjacent to the central tube of the strutted process, and the satellite

pore covers are siliceous outgrowths from the central tube that cover—to a greater or

lesser degree—the associated satellite pore. Satellite pore covers vary from small, narrow

tabs in Thalassiosiraceae to broad ridges in the predominantly freshwater

Stephanodiscaceae. Diameter across the entire strutted process is 200–600 nm.

105 Table 3.1. Models scores (l) and likelihood-ratio statistics (2Δ l) to determine whether the evolution of five morphological characters in Thalassiosirales is correlated with habitat type (marine versus freshwater).

Independent Dependent model (l) model (l) 2Δ l P-value costae –40.77 –32.38 16.790.002* central chamber –41.97 –36.55 10.84 0.028* marginal chamber –44.17 –39.93 8.49 0.075* cowling –35.14 –27.65 14.980.005* satellite pore cover –35.14 –27.41 15.46 0.004* *The more complex dependent model (8 free parameters) provides a significantly better

fit than the independent model (4 free parameters).

106

Chapter 4: Strong purifying selection in the silicon transporters of marine and

freshwater diatoms

Abstract. Marine and freshwater diatoms show several important differences in silicon physiology. In addition to containing an order of magnitude more silica in their cell walls, freshwater diatoms also have a drastically lower enzymatic affinity for silicon. A novel silicon transporter (SIT), encoded by a small gene family, imports silicon from the environment into the diatom cell. The disparity in uptake efficiency between marine and freshwater diatoms might be caused by differences in SIT genes. Scarcity of silicon in oceans coupled with high demand by diatoms led to the hypothesis that marine diatoms are constrained to maintain a highly efficient uptake system, consistent with observing purifying selection on SITs. This constraint might have been relaxed upon colonization of freshwater, where silicon concentrations are much higher. Alternatively, SITs in freshwater species might be under positive selection to optimize their function in the vastly different nutrient and osmotic environment of fresh waters. Partial SIT genes were cloned and sequenced from 45 species of Thalassiosirales, a diatom lineage with high diversity in marine and fresh waters. Phylogenetically-based codon substitution models were used to test whether SITs from marine and freshwater taxa were under similar selective constraints, and to test whether codons in different structural locations of the protein were under similar functional constraints. Results show that purifying selection is the predominant evolutionary force acting on SITs, irrespective of location in the protein,

107 and that differences in efficiency of silicon uptake between marine and freshwater

diatoms are not attributable to sequence differences in SITs.

Introduction

Diatoms are best known for their resilient and intricately ornamented cell walls of

opaline silica (Si), called "frustules." Silicon comprises ~25% (by weight) of the earth's

crust (Iler, 1979), and it has been estimated that each silica atom, of the 6.1 Tmol

annually input into the world's oceans, is cycled through nearly 40 diatoms before burial

in marine sediments (Treguer et al., 1995). The predominate form of dissolved silicon in

marine and fresh waters is undissociated orthosilicic acid, Si(OH)4, which is also the

form used by most diatoms (Del Amo and Brzezinski, 1999; Paasche, 1980a). Diatoms

are the primary biological mediators of the silica cycle in modern oceans (Conley, 2002;

Siever, 1991), where surface concentrations are extremely low, typically 1–10 μM, and

are thought to have been so since the marked diversification and ecological expansion of

diatoms beginning in the early Cenozoic (Katz et al., 2004; Martin-Jezequel et al., 2000).

Silicic acid concentrations are much greater in freshwater, with values as high as 100 μM

in lakes (Paasche, 1980a) and 150 μM in rivers (Treguer et al., 1995). It is generally

understood that diatoms are ancestrally marine, and that many of the of the major extant

diatom lineages have representatives in fresh waters as the result of independent

colonization events. A large body of evidence drawn from a broad sampling of these

lineages shows several fundamental differences in the ways marine and freshwater

diatoms use silicon.

108 Marine and freshwater diatoms differ both in total silica content and efficiency of silicic acid use. Freshwater diatoms contain one order of magnitude more silica in their frustules than marine diatoms, a difference that supercedes the confounding effects of cell biovolume, habitat, and comparisons of natural populations versus cell cultures (Conley et al., 1989). Culture experiments have shown that diatoms grown under limiting silicic acid concentrations have thinner frustules (Brzezinski et al., 1990; Davis, 1976; Harrison et al., 1977; Paasche, 1975), so the difference in silica content between marine and freshwater diatoms might reflect the disparity in silicon availability between the two habitats. Sponges and radiolarians also become less silicified in response to low silicon availability, a trend that has occurred over geological time and is correlated with depleted oceanic silicon availability coincident with the proliferation of diatoms (Harper and

Knoll, 1974; Moore, 1969; Siever, 1991). The difference in silica content between marine and freshwater diatoms could be an osmotic effect. Data from another set of culture experiments showed that a freshwater strain of Cyclotella menegheniana grew faster (cell divisions/day) and contained less silica in high salinity treatments (Tuchman et al., 1984).

These two responses are probably not independent. Silicon transport is coordinated with cell division in most diatoms, so transport peaks during the G2 and M phases of the cell cycle in concert with valve synthesis (Brzezinski, 1992; Brzezinski et al., 1990;

Hildebrand and Wetherbee, 2003; Martin-Jezequel et al., 2000; Sullivan, 1977). Also, salinity can promote diatom growth (i.e., cell division rate) (Guillard and Ryther, 1962;

Olsen and Paasche, 1986), so the difference in silica content between marine and freshwater diatoms might be a consequence of the hastened cell cycle in marine diatoms, which simply allows less time for silicic acid transport and deposition (Flynn and Martin-

Jezequel, 2000; Martin-Jezequel et al., 2000). Intuitively, a lower overall silica budget

109 should be advantageous for diatoms in the marine environment where silicic acid

concentrations are extremely low and potentially growth limiting. There is also a

potentially greater selective force on marine diatoms to have lightly silicified cells with

possibly reduced sinking rates because, unlike diatoms in fresh waters, marine diatoms

have much less chance of being resuspended after sinking below the well-mixed photic

zone (Conley et al., 1989). Although provocative, there is no direct support for this

hypothesis (Bienfang et al., 1982; Culver and Smith, 1989; Martin-Jezequel et al., 2000).

A second fundamental difference between marine and freshwater diatoms is in

their efficiency of silicic acid utilization. Silicon uptake by diatoms follows the

Michaelis–Menten saturation function, and silicon-limited growth rate—a surrogate

measure for uptake—follows the Monod function (Martin-Jezequel et al., 2000). The

inferred half-saturation constants (Ks for uptake, Kμ for growth) from these experiments

measure a diatom's enzymatic affinity for silicic acid: diatoms with lower Ks and Kμ values are predicted to have a greater enzymatic affinity for silicic acid. These values are commonly used to predict the outcome of interspecific competition for low silicon, whereby the diatom with the lowest Ks or Kμ is predicted to competitively displace other species (Tilman, 1977; Tilman, 1981; Tilman and Kilham, 1976). Half-saturation constants for species of Thalassiosirales—the focus of this study—reflect the variation across diatoms as a whole, with Ks values of 0.2–7.6 μm and Kμ values of 0.02–8.6 μm

(reviewed in Martin-Jezequel et al., 2000). These wide ranges reflect the disparity

between marine and freshwater diatoms, with the clear trend that marine species have

drastically lower Ks and Kμ values than freshwater species (Martin-Jezequel et al., 2000;

Paasche, 1980a). Taken together, these data suggest that marine diatoms might be

constrained to have a lower overall silica budget and substantially greater affinity for

110 silicic acid, possibly due to the scarcity of silicic acid in oceans. In contrast, the higher overall silica budget and increased kinetic parameters of freshwater diatoms suggest that this constraint might be relaxed in fresh waters, which tend to have a surplus of silicic acid.

Silicon uptake by diatoms is controlled by a small gene family of silicon transporters (SITs) originally characterized from the biraphid pennate diatom,

Cylindrotheca fusiformis (Hildebrand et al., 1998; Hildebrand et al., 1993; Hildebrand et al., 1997). Although attempts to identify SIT homologs in other silicified algae have been fruitless, SIT homologs have been found in all diatom species investigated (Armbrust et al., 2004; Grachev et al., 2002; Hildebrand et al., 1998; Sherbakova et al., 2005;

Thamatrakoln et al., 2006). Cylindrotheca fusiformis had five distinct SITs, each predicted to have 10 membrane-spanning segments thought to be the site of silicon transport, and a coiled-coil motif at the carboxy terminus thought to help regulate SIT activity and localization (Hildebrand et al., 1998; Hildebrand and Wetherbee, 2003). SITs from all Thalassiosirales species investigated lack a C-terminal coiled-coil motif

(Thamatrakoln et al., 2006). The regular, coordinated expression of SITs through the cell cycle suggests some degree of sub-functionalization among SIT paralogs, perhaps allowing for fine control of the timing, affinity, and capacity of transport (Hildebrand et al., 1998; Martin-Jezequel et al., 2000; Thamatrakoln et al., 2006). The efficiency of the transport system is such that silicic acid can be imported against a steep, sometimes

~1000-fold concentration gradient across the plasma membrane (Martin-Jezequel et al.,

2000). SITs mediate both influx and efflux of silicic acid, and Thamatrakoln et al. (2006) suggested that influx and efflux might occur separately via the five N-terminal and C- terminal transmembrane segments, respectively. Despite the drastically different silicic

111 acid uptake kinetics of marine and freshwater species, a comparative sequence analysis of

SITs from eight diverse diatom species revealed no clear differences between marine and

freshwater species (Thamatrakoln et al., 2006), though low taxon sampling might have

limited their statistical power to detect such a pattern.

High species diversity in marine and fresh waters and the availability of a densely

sampled phylogenetic hypothesis (Alverson, 2006) make Thalassiosirales an excellent

system for studying the gene phylogeny and molecular evolution of SITs, especially with

regard to well-established differences in silicon physiology between marine and

freshwater diatoms. The scarcity of silicic acid in oceans coupled with high demand by diatoms led to the hypothesis that marine diatoms are constrained to maintain a highly

efficient uptake system, consistent with observing purifying (negative) selection on SITs.

This constraint might have been relaxed upon colonization of freshwater where silicic acid concentrations are much greater. Alternatively, SITs in freshwater taxa might be subject to diversifying (positive) selection in response to the vastly different nutrient and

osmotic environment in fresh waters. Partial SIT genes were PCR-amplified, cloned, and

sequenced from 45 species (26 marine, 19 freshwater) sampled across the

Thalassiosirales phylogeny. The inferred gene phylogeny was used to test a number of

hypotheses with codon substitution models. First, random-sites models (Yang and

Swanson, 2002; Yang et al., 2000a) were used to test for and identify sites under positive

selection without regard to the secondary structure of the protein. Second, exposed (non-

cytosolic), transmembrane, and internal (cytosolic) segments of SITs probably serve

different functions in transporting silicic acid across the plasma membrane, so fixed-sites

models (Yang and Swanson, 2002) were used to test whether sites from these different

segments were under the same selective constraints. Finally, clade models (Bielawski and

112 Yang, 2004; Yang, 1998) were used to test whether SITs from marine and freshwater taxa

were under different selective constraints.

Materials and Methods

Cell culture and laboratory methods

Diatom cell culture techniques and DNA extraction methods followed Alverson et

al. (2006). Degenerate primers were designed from an alignment of full-length SIT

sequences identified in the complete genome sequence of Thalassiosira pseudonana

(Armbrust et al., 2004). Primers SIT22+ (GGI MGK CAG TTC ATG GTI CT) and

SIT1037– (CCR ACG TAI WCT TCG TCG TA) were used to amplify 900–1000

nucleotide fragments under the following conditions: 96o C for 5 m, 37 cycles of (94o C for 50 s, 50o C for 50 s, 72o C for 60 s), and final extension at 72o C for 7 m. Amplicons

were cloned with a TOPO TA Cloning® Kit (Invitrogen, San Diego, CA, USA)

following the instructions of the manufacturer but with one-third reaction size.

Transformant E. coli were spread onto Luria-Bertani (LB) plates containing X-gal (40

mg/L) and kanamycin (50 µg/mL) and incubated for 12–16 hours. Several transformant

colonies per sample were transferred directly to a PCR cocktail and screened for the

insert with provided M13 primers using a "hotstart" protocol: 94o C for 10 m, 72o C for 5

m, 37 cycles of (94o C for 50 s, 48o C for 30 s, 72o C for 1:10 m), and final extension at

72o C for 7 m. Amplicons were purified with QIAquick® PCR Purification columns

(QIAGEN, Hilden, Germany). Forward and reverse strands were cycle-sequenced with

BigDye (Applied Biosystems, Foster City, CA, USA) using the PCR primers (Table 4.1).

Sequences were resolved with an ABI 3700 DNA Analyzer.

Multiple sequence alignment and phylogenetic analysis

113 SIT sequences had very few nucleotide insertions and deletions and were aligned

manually with MacClade (ver. 4.08, Maddison and Maddison, 2003). Nucleotide

sequences were converted to their predicted AA sequences with MacClade for structure

prediction. All phylogenetic analyses were based on the nucleotide alignment.

MrModeltest (Nylander, 2004) was used to determine the most appropriate model

of DNA sequence evolution. Hierarchical likelihood ratio tests and the Akaike

Information Criterion both favored the general time reversible model (GTR) with

parameters for gamma-distributed rate heterogeneity (Γ) and a proportion of invariant

sites (I). The maximum likelihood tree was found using GARLI (Zwickl, 2006), with all

model parameters estimated by the program. Default settings were used, and the

algorithm was run multiple times with different combinations of random number seeds

and starting tree topologies to verify convergence. Final branch lengths and log likelihood scores were calculated with PAUP* (ver. 4.0b10, Swofford, 2001). The

computational efficiency of GARLI made it possible to assess branch support with a full-

optimization maximum likelihood bootstrap analysis (Felsenstein, 1985). Five-hundred pseudoreplicates were performed with default settings, and bootstrap proportions were

calculated by computing a 50% majority-rule consensus tree with PAUP*.

SIT secondary structure

Four tools for prediction of transmembrane domains were used to determine the

SIT topology: HMMTOP 2.0 (Tusnady and Simon, 2001), Phdhtm (Rost et al., 1996),

SOSUI (Hirokawa et al., 1998), and TMAP (Persson and Argos, 1997). There was overall

agreement among the four methods about the core SIT structure, and final

transmembrane boundaries were based on a consensus among the methods, with greater

weight given to HMMTOP 2.0 and Phdhtm, which returned similar predictions and

114 repeatedly have been shown to be accurate predictors of topology (Chen et al., 2002;

Cuthbertson et al., 2005).

Tests for positive and divergent selection

Codon models that consider the ratio of nonsynonymous (amino acid changing,

dn) to synonymous (not amino acid changing, ds) rates of nucleotide substitution are

commonly used to study the molecular evolution of protein coding sequences (Yang,

2002). These models can be used to identify codons evolving under different dn/ds

substitution rate ratios (ω=dn/ds), whereby ω <1, ω=1, and ω>1 indicate purifying

(negative) selection, neutral evolution, and diversifying (positive) selection, respectively

(Yang, 2002). For example, ω>1 indicates that amino-acid-altering substitutions are

being fixed at a higher rate than those that do not change the amino acid, which provides

strong evidence of positive selection. Standard codon models were fit to the dataset using

the PAML software package (Yang, 1997), and likelihood ratio tests (LRTs) were used to

compare the relative fit of the different, hierarchically nested models. Likelihood ratio test statistics were compared to a chi-square distribution to determine if the more complex model provided a significantly better fit than the more restricted model. Model descriptions and model comparison strategies for these analyses are outlined in original articles on the subject (e.g., Bielawski and Yang, 2004; Goldman and Yang, 1994; Muse and Gaut, 1994; Yang and Swanson, 2002; Yang et al., 2000a; Yang et al., 2000b).

Model nomenclature follows Yang and Swanson (2002) and Bielawski and Yang (2004).

Random-sites models M0, M1a, M2a, M3, M7, and M8 were fit to the dataset; these models can identify variation in ω among sites but not among lineages. Likelihood ratio tests to identify positively selected sites compared the following pairs of models: null

115 model M1a(neutral) to alternative model M2a(selection) and null model M7(beta) to alternative model M8(beta&ω).

For fixed-sites models (Yang and Swanson, 2002), nucleotide sequences were divided into one class containing sites in internal and external segments and a second class of sites located in transmembrane segments of the protein. Internal and external segments are directly exposed to ambient internal and external silicon concentrations, which, in part, govern the influx or efflux of silicon via membrane-spanning segments

(Martin-Jezequel et al., 2000). It seems plausible that sites in these two structural classes experience different selection pressures, a hypothesis that can be tested with fixed-sites models (Yang and Swanson, 2002). Codon models A–E (Yang and Swanson, 2002) were fit to the dataset using PAML; LRTs to identify positively selected sites compared the following two pairs of models: null model B (or C) to alternative model D and null model

C to alternative model E.

With no a priori expectation of positive selection or relaxed evolutionary constraint in freshwater lineages, the second set of analyses used "clade" models that provide a more general test for divergent selection in the lineages of interest (Bielawski and Yang, 2004), rather than branch-site models designed for tests of positive selection

(Yang and Nielsen, 2002; Zhang et al., 2005). Clade model D (Bielawski and Yang,

2004) was fit to the dataset and tested against null model M3 to determine if a proportion of the sites in freshwater SITs were under different evolutionary constraints relative to marine SITs. All PAML analyses were run from multiple starting values of ω to verify convergence. In addition to the extraordinarily divergent "SIT3" types recovered from T. pseudonana only (Thamatrakoln et al., 2006), two partial sequences (DQ482555 and

DQ482533) were excluded from selection analyses.

116 Results

Alignment, sequence characterization, and SIT secondary structure

The final alignment contained 97 unique sequences (Table 4.1). Two large indels

were present in the alignment. One highly divergent SIT paralog from T. pseudonana

("TpSIT3" in Thamatrakoln et al., 2006) shared a 4-AA indel (AA residues 136–139)

with the clade of P. glacialis, P. pseudodenticulata, and L. annulata SITs. The latter three species always comprise the earliest diverging lineage in phylogenetic analyses of

Thalassiosirales (Alverson, 2006). Together, the highly divergent sequence of SIT3 in T. pseudonana, its shared indel with basal species of Thalassiosirales, and its sister relationship to these species in unrooted trees (results not shown) support the hypothesis that the similarity of "TpSIT3" to other SIT types traces to a gene duplication that predated the origin of Thalassiosirales. Thus, the SIT gene tree was rooted at this split

(Fig. 4.1). The SIT3 type of T. pseudonana was subsequently removed from selection analyses because its high divergence—even among diatoms as a whole (Thamatrakoln et al., 2006)—suggests that it might not function directly in silicon transport. Finally,

Thalassiosira oceanica had one 72-nt intron, in frame, between transmembrane segments

3 and 4.

Despite minor discrepancies about exact boundaries of the transmembrane segments, there was overall agreement among the four methods on the core structure of the SIT fragments. HMMTOP 2.0, Phdhtm, and SOSUI predicted seven transmembrane segments; SOSUI, which considers single sequences and not alignments, predicted five for some sequences, and TMAP consistently predicted five. The five transmembrane domains predicted by SOSUI and TMAP were always a subset of the seven segments predicted by HMMTOP 2.0, Phdhtm, and SOSUI. The final consensus model had seven

117 transmembrane segments (AA residues 17–35 50–70 106–125 145–167 204–224 250–

267 274–294) separated by four internal (AA residues 1–16, 71–105, 168-203, 268–273) and four external segments (AA residues 36–49 126–144 225–249 295–300) (Fig. 4.2).

These seven transmembrane segments correspond to segments 4–10 predicted by

Thamatrakoln et al. (2006).

SIT phylogeny

Three SIT types (paralogs) were found in the complete genome sequence of

Thalassiosira pseudonana (Armbrust et al., 2004), and likewise multiple SIT types

(typically ≤3) were found in nearly every species examined, all of which appeared orthologous to the SIT1 and SIT2 types from T. pseudonana (Armbrust et al., 2004;

Thamatrakoln et al., 2006). No direct orthologs of the SIT3 type from T. pseudonana were recovered from any other species of Thalassiosirales, which included sampling of very close relatives. In many cases, multiple types recovered from the same species or culture strain grouped together (Fig. 4.1), making it difficult to discern allelic variation from lineage-specific gene duplications. Obvious gene duplication events were identifiable, especially in the densely sampled freshwater lineage (clade D, Fig. 1). For example, distantly related SIT types were found in Cyclotella pseudostelligera,

Stephanodiscus reimerii, and Stephanodiscus yellowstonensis (Fig. 4.1); the relationship between distantly related types within these species presumably traces to an earlier gene duplication (Fig. 4.1). In another example, the expected pattern of orthology was recovered in the two SIT types from the closely related species Stephanodiscus agassizensis and Stephanodiscus neoastraea (Fig. 4.1).

The SIT gene phylogeny was grossly similar to the underlying species phylogeny inferred from nuclear ribosomal DNA and chloroplast encoded psbC and rbcL (Alverson,

118 2006). SITs from the 19 freshwater species sorted into four clades spaced throughout the

gene tree (Fig. 4.1). Two of these clades (Fig. 1, "A" and "B") represented different,

paralogous gene lineages nested within the Cyclotella menegheniana species complex,

and one clade included the three SIT types recovered from the secondarily freshwater species, Thalassiosira gessneri (Fig. 1, "C"). The other freshwater clade included SIT

types from the most species-rich, exclusively freshwater lineage in Thalassiosirales (Fig.

1, "D"). This lineage includes SITs from Cyclostephanos, Stephanodiscus, and some

members of Cyclotella (Fig. 4.1).

Molecular evolution of marine and freshwater SITs

Random-sites models were used to determine whether ω varied among sites

(Nielsen and Yang, 1998; Yang et al., 2000a). Model M0 assumes a single, average ω

across all sites and all branches in the tree, and the estimate ω=0.069 indicated that

purifying selection is the predominant evolutionary force acting on SITs (Table 4.2).

Model M1a, which assumes one class of conserved sites (0<ω<1) and one class of

neutrally evolving sites (ω=1), provided a significantly better fit to the data than M0

(Tables 4.2, 4.3). Model M2a, which adds an additional site class of ω>1 to accommodate

sites evolving under positive selection, did not improve over M1a (M1a = M2a = –

26343.967) (Tables 4.2, 4.3), indicating that no sites are evolving under positive

selection. Model M3 fits k=2 or k=3 ω classes estimated from the data. Both variants of

model M3 provide a significant improvement over M0, indicating significant variation in

ω among sites (Tables 4.2, 4.3). Model M3 (k=3) indicates that 10% of sites are under

substantially reduced functional constraint (ω=0.404) compared to the remaining 90% of

sites (ω<0.112) (Tables 4.2, 4.3). Model M7 fits a beta-distribution of ω to the data that

limited to the interval (0,1); M8 adds an additional site class of ω>1, so the M7–M8

119 comparison provides another test of positive selection. Model M7 and M8 fit the data

equally well (l = –25841.426), providing further evidence that no sites are evolving under

positive selection (Tables 4.2, 4.3). The estimated beta distribution β(0.49, 3.98) from M7 and M8 is an extreme "L" shape, indicating that most sites have ω≈0.

Fixed-sites models were fit to the dataset to determine whether transmembrane sites were under different selective constraints from exposed sites on the cell interior and exterior. The first null model fit to the dataset corresponds to random-sites model M0, which assumes a single, homogenous value of ω across all sites, regardless of position in the protein secondary structure. Model B, which assumes a different rate of evolution between the two site classes (transmembrane versus exposed), actually provided a worse

fit to the data than M0, indicating that sites in the two classes have similar rates of evolution (Tables 4.4, 4.5). Increasingly complex fixed-sites models add additional parameters to account for potential variation in codon frequency (model C), ω (model D),

and transition/transversion rate (model E) (Table 4.4). Although each sequentially more

complex model provided a significantly better fit to the data (e.g., C better than B, D better than C, etc.), their surprising lack of improvement over the homogenous M0 model is the most telling result, which indicates that exposed and transmembrane sites are evolving under similarly strong purifying selection (Tables 4.4, 4.5).

Clade models (Bielawski and Yang, 2004) were used to test for divergent selection between marine and freshwater lineages. For these analyses, all freshwater SITs and subtending ancestral branches were treated as "foreground" lineages and marine branches were treated as "background" lineages (Fig. 4.1). Again, the null model M0

(Tables 4.2–4.4) shows the dominant role of purifying selection in SIT evolution

(ω=0.069), and models M3 (k=2) and M3 (k=3) indicate significant variation in ω among

120 sites (Tables 4.2, 4.3). These models do not account for potential variation in ω among lineages, as we expect to observe between marine and freshwater SITs; therefore, clade model D (Bielawski and Yang, 2004) was used to test for divergent selection between marine and freshwater SITs. Model D (k=2) fits one ω site class that is invariable among marine and freshwater lineages and one class of sites with variable ω between marine and freshwater SITs (Table 4.2). Model D (k=3) fits two ω site classes that are invariable among lineages and one class with variable ω between marine and freshwater lineages.

Both variations of Model D fit the data significantly better than models that do not account for variation in ω among lineages (Tables 4.2, 4.3). Model D (k=3), which provided the best fit, showed that 9% of sites were under different selective constraints in marine and freshwater SITs (Table 4.2). Surprisingly, these divergent sites were under stronger purifying selection in freshwater SITs (ω=0.266) than marine SITs (ω=0.542).

Discussion

A large and longstanding body of evidence shows several important differences in the ways marine and freshwater diatoms use silicon (e.g. Conley et al., 1989; Martin-

Jezequel et al., 2000 for review; Olsen and Paasche, 1986; Paasche, 1980a). The striking disparity in Ks and Kμ values between marine (low values) and freshwater (high values) diatoms was recognized as early as 1980 (Paasche, 1980a). A second fundamental distinction concerns silica content, whereby freshwater diatoms contain an order of magnitude more silica in their cell walls than marine diatoms (Conley et al., 1989). The underlying causes of these two striking differences are largely unknown. The characterization of silicon transporter proteins and the genes that encode them provided initial insights into the molecular basis of silicon use by diatoms (Hildebrand et al., 1998;

Hildebrand et al., 1993; Hildebrand et al., 1997). The complete genome sequence of T.

121 pseudonana made it practical to explore the molecular evolution of SITs in

Thalassiosirales, a lineage that also has high species diversity in both marine and freshwater habitats. In addition, many of the foundational experiments on silicon physiology used species of Thalassiosirales (e.g. Brzezinski, 1992; Guillard et al., 1973;

Paasche, 1973a; Paasche, 1973b; Paasche, 1973c; Paasche, 1975; Paasche, 1980b;

Tilman and Kilham, 1976), further compelling its use in a comparative analysis of marine and freshwater SITs.

Diatom SITs, including those of T. pseudonana, are composed of a sequence of 10 transmembrane segments connected by alternating internal (cytosolic) and external (non- cytosolic) segments (Thamatrakoln et al., 2006). In general, diatoms undergo three different modes of silicic acid uptake: surge, externally controlled, and internally controlled uptake (Conway and Harrison, 1977; Conway et al., 1976; Martin-Jezequel et al., 2000). Externally and internally controlled uptake are governed by external and internal silicic acid concentrations, respectively, so internal and external segments of the protein—presumably exposed to these different silicon environments—might be involved in the detection and control of internally and externally controlled uptake. Exposed segments encounter a broad range of osmotic, silicic acid, and other nutrient conditions, so one hypothesis was that some sites in exposed segments might have undergone positive selection in response to vastly different environmental conditions. In contrast, transmembrane segments are thought to directly mediate influx and efflux and so might be under tighter selective constraints. By identifying sites in these different structural regions, random- and fixed-sites models could be used to test these hypotheses directly.

No sites showed evidence of positive selection, regardless of location in the secondary structure of the protein. Random-sites models showed evidence for some variability in

122 constraint among sites, with 90% of sites under strong purifying selection (0.018 < ω <

0.112) and the remaining 10% under somewhat weaker constraint (ω = 0.404).

Unfortunately, current implementations of these models do not identify the individual

sites under divergent selection when ω < 1, so sites under relaxed constraint could not be

identified and mapped to the protein secondary structure. Although the secondary

structure of SITs was identified as accurately as possible, exact boundaries of

transmembrane segments cannot be predicted algorithmically without some error

(Cuthbertson et al., 2005). Fixed-sites are expected to perform poorly when sites are not

partitioned into the correct structural classes (Yang and Swanson, 2002). Importantly,

however, fixed-sites and random-sites models perform equally well, in terms of

identification of positively selected sites (Yang and Swanson, 2002). In light of this and

the similar finding by both model types of strong purifying selection across nearly all sites in this dataset, potential error in the SIT structure prediction probably had little effect the overall conclusions of this study.

Among the many differences between marine and fresh waters, fresh waters generally are replete with silicic acid (≈100 μM) compared to growth-limiting surface

concentrations of silicic acid over most of the world's oceans (≈10 μM) (Martin-Jezequel

et al., 2000; Paasche, 1980a; Treguer et al., 1995). Low oceanic silicic acid is attributed to high demand by diatoms, which presumably need to maintain a high affinity for silicic acid in order to be successful competitors for this limiting resource. This strong selective pressure is consistent with the extremely low Ks or Kμ of marine diatoms (Martin-

Jezequel et al., 2000; Paasche, 1980a) and led to the hypothesis that SITs of marine

species are under strong purifying selection to maintain their presumably optimal

function. The situation in fresh waters is quite different, where silicic acid concentrations

123 are much greater, and the Ks and Kμ for silicic acid of freshwater diatoms are uniformly

high—representing sluggish, inefficient uptake compared to marine counterparts (Martin-

Jezequel et al., 2000; Paasche, 1980a; Treguer et al., 1995). This evidence, in turn, led to two alternative hypotheses. First, competition for low, growth-limiting silicic acid in oceans might act as constraint on SIT evolution. This constraint might be relaxed in

freshwater, where silicic acid is considerably more abundant. Relaxation of this

constraint might manifest as near-neutral evolution on freshwater SITs (specifically:

ωmarine < ωfreshwater ≤ 1). Alternatively, some residues might be under positive selection to optimize function in a drastically different osmotic and nutrient environment. In fact, the magnitude of purifying selection was fairly uniform across marine and freshwater SITs,

so each of these hypotheses can be rejected. Results actually showed some evidence for selection in the opposite direction of what was expected, with 9% of sites under substantially stronger purifying selection in freshwater SITs.

Taken together, these results indicate that differences in efficiency of silicic acid uptake between marine and freshwater diatoms are not attributable to sequence differences in SITs. One possible explanation is that SIT structure and function had already been optimized in the common ancestor of Thalassiosirales, and purifying selection has suppressed substantive deviations from the ancestral SIT type. Surprising evidence was found that freshwater SITs might be under stronger selective constraints than marine SITs, discounting initial hypotheses that either positive selection had optimized them for function in freshwater, or that high external silicic acid concentrations had relaxed the constraint imposed by growth-limiting silicic acid concentrations in the ancestral marine environment. Analyses that tested specifically for positive selection in the ancestral branches subtending freshwater SIT lineages showed no evidence for

124 episodic changes in SITs upon colonization of freshwater (results not shown), which further suggests that the extremely altered external osmotic and silicon conditions presented by fresh waters do not alter the selective constraints on SITs.

A few caveats on the interpretation of these results are necessary. First, this study examined only seven of the 10 total transmembrane segments. More importantly, the long, external N-terminus was not included in these analyses, so any inferences made here might not apply to this different functional domain. Second, inferences about the gene phylogeny are limited because of potential biases introduced by the experimental protocol. It is clear, for example, that PCR primers preferentially amplified SITs orthologous to the SIT1 and SIT2 types of T. pseudonana. In fact, SIT3-specific primers did not amplify SIT3 orthologs in species other than T. pseudonana. The potentially important role of the T. pseudonana SIT3 type, if present throughout Thalassiosirales, was not addressed in this study. Incomplete sampling of paralogous SITs limited inferences about the actual gene phylogeny because phylogenetic patterns that appear as gene duplications or losses could simply be caused by biased or incomplete sampling of

SITs from different species (e.g., by PCR bias or under-sampling of clones). A broad- scale Southern hybridization study across Thalassiosirales would provide valuable insights into the SIT gene phylogeny by identifying putative gene duplications and losses that could be pursued for follow-up study. A fully sampled gene phylogeny might reveal whether different (paralogous) gene lineages diversified following gene duplication

(Bielawski and Yang, 2004; Ohno, 1970). The regular, coordinated expression of different SITs over the cell cycle suggests some degree of subfunctionalization among

SIT types (Hildebrand, 2000; Hildebrand et al., 1998), a provocative hypothesis that could be tested with a fully sampled gene phylogeny (Bielawski and Yang, 2004).

125 Repeated measurements from a diverse set of diatom species show that freshwater

species have strikingly higher Ks and Kμ for silicic acid than marine species (Martin-

Jezequel et al., 2000; Paasche, 1980a), suggesting a reduced enzymatic affinity or less

efficient uptake mechanism. Results from this study suggest that this difference is most

likely not attributable to differences in the gene sequences of marine and freshwater SITs,

which leaves this longstanding question unanswered. Future investigations might

examine the relative expression levels of SITs in marine and freshwater species, which

would require comparisons of orthologous SITs. To this same end, a comparative analysis

of promoter sequences might also be informative. Finally, silicic acid transport is sodium-

coupled in the marine pennate species Nitzschia alba (Bhattacharyya and Volcani, 1980),

whereas transport appeared sodium- and possibly potassium-coupled in the freshwater

pennate species Navicula pelliculosa (Sullivan, 1976). Some experimental evidence and

comparative sequence data suggest a possible role of zinc in silicic acid transport

(Grachev et al., 2005; Rueter et al., 1981; Rueter and Morel, 1981). If silicic acid

transport is strictly sodium-coupled, the marine environment might simply be more

favorable to uptake. Notably, the euryhaline species Thalassiosira pseudonana had

significantly reduced Ks values for silicic acid in high salinity compared to freshwater

treatments (Olsen and Paasche, 1986). Follow-up studies on freshwater species sampled

across the diatom phylogeny should verify the exact ionic coupling of silicic acid transport in freshwater diatoms. Differences in ionic coupling might underlie some of

disparity in Ks values between marine and freshwater species. If silicic acid transport is,

for example, found to be potassium coupled in a number of distantly related freshwater

lineages, this would provide strong evidence for an adaptive change correlated with the

colonization of freshwater.

126 A central goal in diatom research is to discover the underlying molecular bases of

those physiological traits that make diatoms so important in the structure and function of

aquatic ecosystems, and in the global cycling of biologically important elements, particularly carbon and silica. The complete genome sequence of the diatom

Thalassiosira pseudonana was a milestone in this effort (Armbrust et al., 2004), and it along with a growing number of other studies in the centric diatom order Thalassiosirales

(e.g., Lane et al., 2005; Strzepek and Harrison, 2004) are establishing this group of diatoms as the model system for investigating many important questions in diatom biology. Availability of a densely sampled phylogenetic hypothesis that includes a diversity of marine and freshwater species (Alverson, 2006) make this group particularly suited to powerful comparative studies, especially those focused on understanding physiological differences between marine and freshwater taxa.

Acknowledgements

Mark Hildebrand and Kim Thamatrakoln provided technical advice and valuable insight on SIT structure and function. Gwen Gage provided technical assistance with figures.

This research was supported by NSF PEET grant (DEB-0118883) and NSF Doctoral

Dissertation Improvement Grant (DEB-0407815)

Literature Cited

Alverson, A. J. 2006. Phylogeny and evolutionary ecology of thalassiosiroid diatoms.

PhD thesis. The University of Texas at Austin, Plant Biology Graduate Program.

Armbrust, E. V., J. A. Berges, C. Bowler, B. R. Green, D. Martinez, N. H. Putnam, S.

Zhou, A. E. Allen, K. E. Apt, M. Bechner, M. A. Brzezinski, B. K. Chaal, A.

Chiovitti, A. K. Davis, M. S. Demarest, J. C. Detter, T. Glavina, D. Goodstein, M.

127 Z. Hadi, U. Hellsten, M. Hildebrand, B. D. Jenkins, J. Jurka, V. V. Kapitonov, N.

Kroger, W. W. Y. Lau, T. W. Lane, F. W. Larimer, J. C. Lippmeier, S. Lucas, M.

Medina, A. Montsant, M. Obornik, M. S. Parker, B. Palenik, G. J. Pazour, P. M.

Richardson, T. A. Rynearson, M. A. Saito, D. C. Schwartz, K. Thamatrakoln, K.

Valentin, A. Vardi, F. P. Wilkerson, and D. S. Rokhsar. 2004. The genome of the

diatom Thalassiosira pseudonana: ecology, evolution, and metabolism. Science

306:79–86.

Bhattacharyya, P., and B. E. Volcani. 1980. Sodium-dependent silicate transport in the

apochlorotic marine diatom Nitzschia alba. Proceedings of the National Academy

of Sciences of the United States of America 77:6386–6390.

Bielawski, J. P., and Z. H. Yang. 2004. A maximum likelihood method for detecting

functional divergence at individual codon sites, with application to gene family

evolution. Journal of Molecular Evolution 59:121–132.

Bienfang, P. K., P. J. Harrison, and L. M. Quarmby. 1982. Sinking Rate Response to

Depletion of Nitrate, Phosphate and Silicate in 4 Marine Diatoms. Marine

Biology 67:295–302.

Brzezinski, M. A. 1992. Cell-cycle effects on the kinetics of silicic acid uptake and

resource competition among diatoms. Journal of Plankton Research 14:1511–

1539.

Brzezinski, M. A., R. J. Olson, and S. W. Chisholm. 1990. Silicon availability and cell-

cycle progression in marine diatoms. Marine Ecology Progress Series 67:83–96.

Chen, C. P., A. Kernytsky, and B. Rost. 2002. Transmembrane helix predictions

revisited. Protein Science 11:2774–2791.

128 Conley, D. J. (ed) 2002. Terrestrial ecosystems and the global biogeochemical silica

cycle. Global Biogeochemical Cycles,

Conley, D. J., S. S. Kilham, and E. Theriot. 1989. Differences in silica content between

marine and freshwater diatoms. Limnology & Oceanography 34:205–213.

Conway, H., and P. Harrison. 1977. Marine diatoms grown in chemostats under silicate

or ammonium limitation IV. Transient response of Chaetoceros debilis,

Skeletonema costatum, and Thalassiosira gravida to a single addition of the

limiting nutrient. Marine Biology 43:33–43.

Conway, H., P. Harrison, and C. Davis. 1976. Marine diatoms grown in chemostats under

silicate or ammonium limitation II. Transient response of Skeletonema costatum

to a single addition of the limiting nutrient. Marine Biology 35:187–199.

Culver, M. E., and W. O. Smith. 1989. Effects of environmental variation on sinking

rates of marine phytoplankton. Journal of Phycology 25:262–270.

Cuthbertson, J. M., D. A. Doyle, and M. S. P. Sansom. 2005. Transmembrane helix

prediction: a comparative evaluation and analysis. Protein Engineering Design &

Selection 18:295–308.

Davis, C. O. 1976. Continuous culture of marine diatoms under silicate limitation. 2.

Effect of light intensity on growth and nutrient uptake of Skeletonema costatum.

Journal of Phycology 12:291–300.

Del Amo, Y., and M. A. Brzezinski. 1999. The chemical form of dissolved SI taken up

by marine diatoms. Journal of Phycology 35:1162–1170.

Felsenstein, J. 1985. Confidence limits on phylogenies: an approach using the bootstrap.

Evolution 39:783–791.

129 Flynn, K. J., and V. Martin-Jezequel. 2000. Modelling Si-N-limited growth of diatoms.

Journal of Plankton Research 22:447–472.

Goldman, N., and Z. H. Yang. 1994. Codon-based model of nucleotide substitution for

protein-coding DNA sequences. Molecular Biology and Evolution 11:725–736.

Grachev, M., T. Sherbakova, Y. Masyukova, and Y. Likhoshway. 2005. A potential zinc-

binding motif in silicic acid transport proteins of diatoms. Diatom Research

20:409–411.

Grachev, M. A., N. N. Denikina, S. I. Belikov, E. V. Likhoshvai, M. V. Usol'tseva, I. V.

Tikhonova, R. V. Adel'shin, S. A. Kler, and T. A. Scherbakova. 2002. Elements

of the active center of silicon transporters in diatoms. Molecular Biology 36:534–

536.

Guillard, R. R. L., P. Kilham, and T. A. Jackson. 1973. Kinetics of silicon-limited growth

in the marine diatom Thalassiosira pseudonana Hasle and Heimdal (Cyclotella

nana Hustedt). Journal of Phycology 9:233–237.

Guillard, R. R. L., and J. H. Ryther. 1962. Studies of marine planktonic diatoms. I.

Cyclotella nana Hustedt and Detonula confervacea (Cleve) Gran. Canadian

Journal of Microbiology 8:229–239.

Harper, H. E., and A. H. Knoll. 1974. Silica, diatoms, and Cenozoic radiolarian

evolution. Geology 3:175–177.

Harrison, P. J., H. L. Conway, R. W. Holmes, and C. O. Davis. 1977. Marine diatoms

grown in chemostats under silicate or ammonium limitation. 3. Cellular chemical

composition and morphology of Chaetoceros debilis, Skeletonema costatum, and

Thalassiosira gravida. Marine Biology 43:19–31.

130 Hildebrand, M. 2000. Silicic acid transport and its control during cell wall silicification in

diatoms. Pages 171–188 in Biomineralization: From Biology to Biotechnology

and Medical Applications (E. Bäurlein, ed.) Wiley-VCH, Weinheim.

Hildebrand, M., K. Dahlin, and B. E. Volcani. 1998. Characterization of a silicon

transporter gene family in Cylindrotheca fusiformis: sequences, expression

analysis, and identification of homologs in other diatoms. Molecular and General

Genetics 260:480–486.

Hildebrand, M., D. R. Higgins, K. Busser, and B. E. Volcani. 1993. Silicon-responsive

cDNA clones isolated from the marine diatom Cylindrotheca fusiformis. Gene

132:213–218.

Hildebrand, M., B. E. Volcani, W. Gassmann, and J. I. Schroeder. 1997. A gene family of

silicon transporters. Nature 385:688–689.

Hildebrand, M., and R. Wetherbee. 2003. Components and control of silicification in

diatoms. Progress in Molecular and Subcellular Biology 33:11–57.

Hirokawa, T., S. Boon-Chieng, and S. Mitaku. 1998. SOSUI: classification and

secondary structure prediction system for membrane proteins. Bioinformatics

14:378–379.

Iler, R. K. 1979. The Chemistry of Silica: Solubility, Polymerization, Colloid and Surface

Properties, and BiochemistryJohn Wiley & Sons, New York.

Katz, M. E., Z. V. Finkel, D. Grzebyk, A. H. Knoll, and P. G. Falkowski. 2004.

Evolutionary trajectories and biogeochemical impacts of marine eukaryotic

phytoplankton. Annual Review of Ecology Evolution and Systematics 35:523–

556.

131 Lane, T. W., M. A. Saito, G. N. George, I. J. Pickering, R. C. Prince, and F. M. M.

Morel. 2005. A cadmium enzyme from a marine diatom. Nature 435:42–42.

Maddison, W. P., and D. R. Maddison. 2003. MacClade. Sinauer Associates, Inc.

Martin-Jezequel, V., M. Hildebrand, and M. A. Brzezinski. 2000. Silicon metabolism in

diatoms: Implications for growth. Journal of Phycology 36:821–840.

Moore, T. C. 1969. Radiolaria: change is skeletal weight and resistance to solution.

Bulletin of the Geological Society of America 80:2103–2108.

Muse, S. V., and B. S. Gaut. 1994. A likelihood approach for comparing synonymous and

nonsynonymous nucleotide substitution rates, with application to the chloroplast

genome. Molecular Biology and Evolution 11:715–724.

Nielsen, R., and Z. H. Yang. 1998. Likelihood models for detecting positively selected

amino acid sites and applications to the HIV-1 envelope gene. Genetics 148:929–

936.

Nylander, J. A. A. 2004. MrModeltest v2. Program distributed by the author.

Evolutionary Biology Centre, Uppsala University.

Ohno, S. 1970. Evolution by gene duplicationSpringer-Verlag, Berlin, New York.

Olsen, S., and E. Paasche. 1986. Variable kinetics of silicon limited growth in

Thalassiosira pseudonana (Bacillariophyceae) in response to changed chemical

composition of the growth medium. British Phycological Journal 21:183–190.

Paasche, E. 1973a. The influence of cell size on growth rate, silica content, and some

other properties of four marine diatom species. Norwegian Journal of Botany

20:197–204.

132 Paasche, E. 1973b. Silicon and the ecology of marine plankton diatoms. I. Thalassiosira

pseudonana (Cyclotella nana) grown in a chemostat with silicate as limiting

nutrient. Marine Biology (Berlin) 19:117–126.

Paasche, E. 1973c. Silicon and the ecology of marine plankton diatoms. II. Silicate-

uptake kinetics in five diatom species. Marine Biology (Berlin) 19:262–269.

Paasche, E. 1975. Growth of the plankton diatom Thalassiosira nordenskioldii Cleve at

low silicate concentrations. Journal of Experimental Marine Biology and Ecology

18:173–183.

Paasche, E. 1980a. Silicon. Pages 259–284 in The Physiological Ecology of

Phytoplankton (I. Morris, ed.) University of California Press, Berkeley.

Paasche, E. 1980b. Silicon content of 5 marine plankton diatom species measured with a

rapid filter method. Limnology & Oceanography 25:474–480.

Persson, B., and P. Argos. 1997. Prediction of membrane protein topology utilizing

multiple sequence alignments. Journal of Protein Chemistry 16:453–457.

Rost, B., P. Fariselli, and R. Casadio. 1996. Topology prediction for helical

transmembrane proteins at 86% accuracy. Protein Science 5:1704–1718.

Rueter, J. G., S. W. Chisholm, and F. M. M. Morel. 1981. Effects of copper toxicity on

silicic-acid uptake and growth in Thalassiosira pseudonana. Journal of Phycology

17:270–278.

Rueter, J. G., and F. M. M. Morel. 1981. The Interaction between zinc deficiency and

copper toxicity as it affects the silicic-acid uptake mechanisms in Thalassiosira

pseudonana. Limnology and Oceanography 26:67–73.

133 Sherbakova, T. A., Y. A. Masyukova, T. A. Safonova, D. P. Petrova, A. L. Vereshagin,

T. V. Minaeva, R. V. Adelshin, T. I. Triboy, I. V. Stonik, N. A. Aizdaitcher, M.

V. Kozlov, Y. V. Likhoshway, and M. A. Grachev. 2005. Conserved motif

CMLD in silicic acid transport proteins of diatoms. Molecular Biology 39:269–

280.

Siever, R. 1991. Silica in the oceans: biological-geochemical interplay. Pages 287–295 in

Scientists on Gaia (S. H. Schneider, and P. J. Boston, eds.). MIT Press,

Cambridge.

Strzepek, R. F., and P. J. Harrison. 2004. Photosynthetic architecture differs in coastal

and oceanic diatoms. Nature 431:689–692.

Sullivan, C. W. 1976. Diatom mineralization of silicic acid. I. Si(OH)4 transport

characteristics in Navicula pelliculosa. Journal of Phycology 12:390–396.

Sullivan, C. W. 1977. Diatom mineralization of silicic acid. II. Regulation of Si(OH)4

transport rates during the cell cycle of Navicula pelliculosa. Journal of Phycology

13:86–91.

Swofford, D. L. 2001. PAUP*: Phylogenetic analysis using parsimony (*and other

methods), version 4.0b8. Academic Press.

Thamatrakoln, K., A. J. Alverson, and M. Hildebrand. 2006. Comparative sequence

analysis of diatom silicon transporters: towards a mechanistic model of silicon

transport. Journal of Phycology (in press).

Tilman, D. 1977. Resource competition between planktonic algae: experimental and

theoretical approach. Ecology 58:338–348.

134 Tilman, D. 1981. Tests of resource competition theory using 4 species of Lake Michigan

algae. Ecology 62:802–815.

Tilman, D., and S. S. Kilham. 1976. Phosphate and silicate growth and uptake kinetics of

the diatoms Asterionella formosa and Cyclotella meneghiniana in batch and

semicontinuous culture. Journal of Phycology 12:375–383.

Treguer, P., D. M. Nelson, A. J. Vanbennekom, D. J. Demaster, A. Leynaert, and B.

Queguiner. 1995. The silica balance in the world ocean: a reestimate. Science

268:375–379.

Tuchman, M. L., E. C. Theriot, and E. F. Stoermer. 1984. Effects of low level salinity

concentrations on the growth of Cyclotella meneghiniana Kütz. (Bacillariophyta).

Archiv Fuer Protistenkunde 128:319–326.

Tusnady, G. E., and I. Simon. 2001. The HMMTOP transmembrane topology prediction

server. Bioinformatics 17:849–850.

Yang, Z., and W. J. Swanson. 2002. Codon substitution models to detect adaptive

evolution that account for heterogeneous selective pressures among site classes.

Mol Biol Evol 19:49–57.

Yang, Z. H. 1997. PAML: a program package for phylogenetic analysis by maximum

likelihood. Computer Applications in the Biosciences 13:555–556.

Yang, Z. H. 1998. Likelihood ratio tests for detecting positive selection and application to

primate lysozyme evolution. Molecular Biology and Evolution 15:568–573.

Yang, Z. H. 2002. Inference of selection from multiple species alignments. Current

Opinion in Genetics & Development 12:688–694.

135 Yang, Z. H., and R. Nielsen. 2002. Codon-substitution models for detecting molecular

adaptation at individual sites along specific lineages. Molecular Biology and

Evolution 19:908–917.

Yang, Z. H., R. Nielsen, N. Goldman, and A. M. K. Pedersen. 2000a. Codon-substitution

models for heterogeneous selection pressure at amino acid sites. Genetics

155:431–449.

Yang, Z. H., W. J. Swanson, and V. D. Vacquier. 2000b. Maximum-likelihood analysis

of molecular adaptation in abalone sperm lysin reveals variable selective

pressures among lineages and sites. Molecular Biology and Evolution 17:1446–

1455.

Zhang, J. Z., R. Nielsen, and Z. H. Yang. 2005. Evaluation of an improved branch-site

likelihood method for detecting positive selection at the molecular level.

Molecular Biology and Evolution 22:2472–2479.

Zwickl, D. J. 2006. GARLI, version 0.94. Program distributed by the author. Section of

Integrative Biology, The University of Texas at Austin.

136 Figure 4.1. Gene phylogeny of silicon transporter genes (SITs) from Thalassiosirales. Each SIT is identified by species name and the final three digits of its GenBank accession ID (Table 4.1). The strain ID is noted for conspecific culture strains. SITs from freshwater species (blue) were treated as "foreground" branches in tests for divergent selection (see text). Bootstrap support values greater than 95% are shown with an asterisk (*). The four freshwater SIT lineages (A–D) are labeled to facilitate discussion in the text.

137

Figure 4.2. Cartoon representing the secondary structure of the last seven transmembrane

segments of silicon transporters (SITs) in Thalassiosirales, inferred from an alignment of

partial SIT genes sequences. Amino acid residues are shown by circles, and transmembrane boundaries are demarcated by cylinders embedded within the plasma membrane. The upper segments connecting transmembrane domains are non-cytosolic, and the lower segments represent internal, cytosolic segments.

138 Table 4.1. Taxon, culture strain, habitat, and GenBank accession number for silicon transporter sequences analyzed in this study.

Taxon Culture strain Habitat Locality GenBank ID Bacterosira bathyomphala (Cleve) Syvertsen & NB04-B61 marine Narragansett Bay, RI, USA DQ482537, Hasle DQ482538 Bacterosira sp. CCMP991 marine Chase Creek, MA, USA DQ482508, DQ482509 Cyclotella cryptica Reimann, Lewin & Guillard CCMP331 marine Chase Creek, MA, USA DQ482474 Cyclotella striata (Kützing) Grunow in Cleve & CCMP1586 marine Jakarta Harbor, Indonesia DQ482475, Grunow DQ482476 Detonula pumila (Castracane) Gran NB481 marine Narragansett Bay, RI, USA DQ482535, DQ482536 Lauderia annulata Cleve CS30 marine Pacific Ocean, La Jolla, CA, USA DQ482470 Minidiscus trioculatus (F.J.R. Taylor) Hasle CCMP495 marine Gulf of Maine, ME, USA DQ482500, DQ482501 Porosira glacialis (Grunow) Jørgensen CCMP1099 marine Southern Ocean, Antarctica DQ482466, 139 DQ482467 Porosira pseudodenticulata (Hustedt) Jousé CCMP1433 marine McMurdo Sound, Antarctica DQ482468, DQ482469 Skeletonema grethae Zingone & Sarno CCAP1077/4 marine Strait of Georgia, British DQ482539 Columbia, Canada Skeletonema japonicum Zingone & Sarno NB02-451 marine Narragansett Bay, RI, USA DQ482542 Skeletonema menzelii Guillard CCMP787 marine Sargasso Sea, Atlantic Ocean DQ482541 Skeletonema subsalsum (Cleve-Euler) Bethge CCAP1077/8 marine Lower Lough Erne, Ireland DQ482540 Thalassiosira antarctica Comber CCMP982 marine Oslo Fjord, Norway DQ482504, DQ482505 Thalassiosira cf. pacifica FB02-351 marine San Francisco Bay, CA, USA DQ482523, DQ482524, DQ482525 Taxon Culture strain Habitat Locality GenBank ID Thalassiosira guillardii Hasle CC03-041 marine Clam Creek, GA, USA DQ482531, DQ482532 Thalassiosira minima Gaarder emend Hasle CCMP990 marine unknown DQ482506, DQ482507 Thalassiosira nodulolineata (Hendey) Hasle & G. BEN02-331 marine San Joaquin River, Benicia, CA, DQ482494, Fryxell USA DQ482495, DQ482496 Thalassiosira oceanica Hasle CCMP1001 marine North Atlantic Ocean DQ482510 Thalassiosira pseudonana Hasle & Heimdal CCMP1335 marine Forge River, Long Island, New DQ256066, York USA DQ256067, DQ256068 Thalassiosira pseudonana NEPC709 marine Hood Canal, WA, USA DQ482463, DQ482473 Thalassiosira punctigera (Castracane) Hasle NB02-22 marine Narragansett Bay, RI, USA DQ482471, DQ482472 140 Thalassiosira ritscheri (Hustedt) Hasle LC01-121 marine Drake Passage, Southern Ocean DQ482533, DQ482534 Thalassiosira rotula Meunier CCMP1812 marine East Sound, Orcas Island, DQ482518, Washington, USA DQ482519, DQ482520 Thalassiosira sp. CCMP1065 marine Baffin Bay, Arctic Ocean DQ482514 Thalassiosira sp. CCMP353 marine Narragansett Bay, RI, USA DQ482498, DQ482499 Thalassiosira weissflogii (Grunow) G. Fryxell & CCMP1010 marine Gulf Stream, Atlantic Ocean DQ482511, Hasle DQ482512 Thalassiosira weissflogii CCMP1336 marine Gardiners Island, Long Island, DQ256069, New York USA DQ256070 Thalassiosira weissflogii L1296 marine unknown DQ482528, DQ482529 Taxon Culture strain Habitat Locality GenBank ID Cyclostephanos invisitatus (Hohn & Hellerman) FHTC261 freshwater Fairport Harbor, Lake Erie, OH, DQ482549, Theriot, Stoermer & Håkansson USA DQ482550 Cyclostephanos sp. WTC161 freshwater West Lake Okoboji, IA, USA DQ482567 Cyclostephanos tholiformis E.F. Stoermer, FHTC151 freshwater Fairport Harbor, Lake Erie, OH, DQ482548 Håkansson & E.C. Theriot USA Cyclotella bodanica Grunow J98-11 freshwater Jackson Lake, WY, USA DQ482554, DQ482555 Cyclotella cf. cryptica WC03-011 freshwater Waller Creek, Austin, TX, USA DQ482487 Cyclotella cf. pseudostelligera L4351 freshwater Montezuma Well National DQ482556 Monument, Yavapai Co., AZ, USA Cyclotella distinguenda Hustedt none2 freshwater Tiplady Bog, Hell, MI, USA DQ482485, DQ482486 Cyclotella cf. menegheniana F81 freshwater Fugleso Lake, Denmark DQ482477 Cyclotella cf. menegheniana L12631 freshwater Stump Lake, ND, USA DQ482479 Cyclotella cf. menegheniana TI11 freshwater Lake Titicaca, Peru DQ482483, 141 DQ482484 Cyclotella pseudostelligera Hustedt ROR01-11 freshwater Raccourci Old River, LA, USA DQ482559, DQ482560 Cyclotella stelligera Cleve & Grunow L1360 freshwater Big Pond, Cedar Hill, TX, USA DQ482557 Stephanodiscus agassizensis H. Håkansson & H. CHTC11 freshwater Cleveland Harbor, Lake Erie, OH, DQ482543, Kling USA DQ482544 Stephanodiscus binderanus (Kützing) Krieger ESB21 freshwater Lake Erie, MI, USA DQ482545 Stephanodiscus hantzschii f. tenuis (Hustedt) WTC211 freshwater West Lake Okoboji, IA, USA DQ482568, Håkansson & Stoermer DQ482570 Stephanodiscus minutulus (Kützing) Cleve & J95-211 freshwater Jackson Lake, WY, USA DQ482552, Möller DQ482553 Stephanodiscus minutulus WTC11 freshwater West Lake Okoboji, IA, USA DQ482566 Stephanodiscus minutulus Y95-61 freshwater Yellowstone Lake, WY, USA DQ482573, DQ482574 Taxon Culture strain Habitat Locality GenBank ID Stephanodiscus minutulus Y98-11 freshwater Yellowstone Lake, WY, USA DQ482569 Stephanodiscus neoastraea H. Håkansson & B. Sneo41 freshwater Grober Müggelsee, Berlin, DQ482561, Hickel Germany DQ482562 Stephanodiscus niagarae Ehrenberg J95-161 freshwater Jackson Lake, WY, USA DQ482551 Stephanodiscus niagarae OKA-A1 freshwater Okamanpeedan Lake, MN, USA DQ482558 Stephanodiscus cf. parvus FHTC111 freshwater Fairport Harbor, Lake Erie, MI, DQ482546, USA DQ482547 Stephanodiscus reimerii Theriot WLO111 freshwater West Lake Okoboji, IA, USA DQ482563, DQ482564, DQ482565 Stephanodiscus yellowstonensis Theriot Y71 freshwater Yellowstone Lake, WY, USA DQ482571, DQ482572 Thalassiosira gessneri (Grunow) G. Fryxell & An02-081 freshwater San Joaquin River, Antioch, CA, DQ482488, Hasle USA DQ482489, DQ482490 142 Thalassiosira pseudonana ETC11 freshwater Lake Erie, MI, USA DQ482464 1Culture isolated and/or maintained in Theriot Lab at The University of Texas at Austin

2SITs were PCR-amplified and cloned directly from environmental sample

CCMP= The Provasoli–Guillard National Center for Culture of Marine Phytoplankton; L= Loras College Freshwater Diatom Culture

Collection; NEPC= Canadian Center for the Culture of Microorganisms; CCAP=and Culture Collection of Algae and Protozoa

Table 4.2. Number of free parameters (p), log-likelihood (l), and parameter estimates for random-sites and clade models. See text for model nomenclature.

Positively Model p l Parameter estimates selected sites M0 (homogenous) 1 –26724.349 ω = 0.069 not allowed M1a (neutral) 2 –26343.967 ω0 = 0.060, f0 = 0.92 none ω1 = 1.000, f1 = 0.08 M2a (selection) 4 –26343.967 ω0 = 0.060, f0 = 0.92 none ω1 = 1.000, f1 = 0.06 ω2 = 1.000, f2 = 0.02 M3 (discrete, k=2) 3 –25978.351 ω0 = 0.028, f0 = 0.73 none ω1 = 0.221, f1 = 0.27 M3 (discrete, k=3) 5 –25854.429 ω0 = 0.018, f0 = 0.58 none ω1 = 0.112, f1 = 0.32 ω2 = 0.404, f2 = 0.10 M7 (beta) 2 –25841.426 β(0.49, 3.98) not allowed ω < 0.40, f = 1.00 M8 (beta&ω) 4 –25841.426 β(0.49, 3.98) none ω0 < 0.40, f0 = 1.00 D (clade, k=2) 4 –25969.303 ω0 = 0.028, p0 = 0.73 none ωmarine = 0.251, ωfreshwater = 0.179, p1 = 0.27 D (clade, k=3) 6 –25835.950 ω0 = 0.019, p0 = 0.60 none ω1 = 0.120, p1 = 0.31 ωmarine = 0.542, ωfreshwater = 0.266, p2 = 0.09 ω = dn/ds ratio

pn = proportion of sites in ω class n

β = shape parameters for beta distribution of ω

143 Table 4.3. Likelihood ratio test statistics (2Δ l), degrees of freedom (df), and p-values for comparisons of random-sites and clade models. See text for model nomenclature.

2Δ l df P-value M0 vs. M1 760.8 1 <0.0001** M1 vs. M2 0.0 2 1.0 M0 vs. M3 (k=2) 1492.0 2 <0.0001** M0 vs. M3 (k=3) 1739.8 4 <0.0001** M3 (k=2) vs. M3 (k=3) 247.8 2 <0.0001** M3 (k=2) vs. D (k=2) 18.1 1 <0.0001** M3 (k=3) vs. D (k=3) 37.0 1 <0.0001** D (k=2) vs. D (k=3) 266.7 2 <0.0001** **significant at p<0.05

144 Table 4.4. Number of free parameters (p, including parameters for codon frequencies), log-likelihood (l), and parameter estimates for fixed-sites models that distinguished internal and external segments (partition 1) from transmembrane segments (partition 2).

Number of parameters includes branch length estimates for all 181 branches in the tree.

See text for model nomenclature.

Model p1 l rs κ ω A (homogenous) 11 –26724.349 — 1.50 0.0694 B (different rs) 12 –27297.4430.996 1.52 0.0689 C (different rs, πs) 21 –27327.9630.895 1.51 0.0683 D (different rs, κ, ω) 14 –27294.3540.972 κ1=1.46 ω1=0.0653 κ2=1.62 ω2=0.0749 E (different rs, κ, ω, πs) 23 –27324.504 0.880 κ1=1.43 ω1=0.0654 κ2=1.66 ω2=0.0734 *includes codon frequencies rs = nucleotide substitution rate

πs = codon frequencies

κ = nucleotide transition/transversion ratio

ω = dn/ds ratio

145

Table 4.5. Likelihood ratio test statistics (2Δ l), degrees of freedom (df), and p-values for comparisons of fixed-sites models. See text for model nomenclature.

2Δ ldf P-value A vs. B -1146.2 1 1 B vs. C 61.0 9 1 B vs. D 6.2 2 0.046** C vs. E 6.9 2 0.032** **significant at p<0.05

146

Chapter 5: Cell wall morphology and systematic importance of Thalassiosira

ritscheri (Hustedt) Hasle, with a description of Shionodiscus gen. nov.

Abstract. The centric diatom order Thalassiosirales includes all diatoms with a fultoportula (strutted process), a feature now recognized as a synapomorphy for the lineage. Within Thalassiosirales, Thalassiosira is perhaps the most taxonomically and morphologically diverse genus, and at least two distinct morphological groups have been recognized within it. Group "A" Thalassiosira species, which include the type species, T. nordenskioeldii, have short inward and long outward extensions of the strutted processes and a labiate process on the valve mantle. Group "B" species have exceptionally long inward and reduced outward extensions of the strutted processes, and a labiate process on the valve face. We collected and cultured Thalassiosira ritscheri, which has a combination of group A and B characters. It has a labiate process on the valve face and reduced outward extensions of the strutted processes. We show for the first time that T. ritscheri has short inward, A-type extensions of the strutted processes. A phylogenetic interpretation of these conditions suggests a close relationship between T. ritscheri and the traditionally held group B species. Species diagnosed by the autapomorphic condition of a labiate process away from the valve mantle, including many group "B" Thalassiosira species, are transferred into Shionodiscus gen. nov.

147 Introduction

The centric diatom order Thalassiosirales includes all diatoms with a fultoportula

(strutted process, SP), a feature now recognized as a synapomorphy for the lineage

(Theriot and Serieyssol, 1994). This interpretation is congruent with phylogenetic analysis of SSU rDNA data (Medlin and Kaczmarska, 2004). Thalassiosiroid diatoms are among the most abundant and diverse diatoms in the plankton of large rivers and reservoirs, freshwater lakes, brackish and estuarine waters, nearshore marine waters, and open ocean habitats (Hasle and Syvertsen, 1997; Stoermer and Julius, 2003). With nearly

400 names, Thalassiosira is one of the most taxonomically diverse genera within

Thalassiosirales (according to the computerized database of verified diatom names at the

California Academy of Sciences, E. Fourtanier and J.P. Kociolek, pers. comm.), and not surprisingly, it is among the most morphologically diverse genera as well. With few exceptions (Theriot et al., 1987), the generic level classification of Thalassiosirales is phenetic, a fact which has been supported consistently by phylogenetic analysis of SSU rDNA sequence data (most recently by Medlin and Kaczmarska, 2004). Consistent with phylogenetic results, morphological data further suggest that no single character or combination of characters can be interpreted as derived and therefore indicative of monophyly of Thalassiosira (Hasle and Syvertsen, 1997; Theriot and Serieyssol, 1994).

At least two distinct morphological groups have been recognized within

Thalassiosira on the basis of SP ultrastructure and location of the rimoportula (labiate process, LP) (Hasle, 1968; Hasle and Syvertsen, 1997). One group (group "A", Hasle,

1968; Hasle and Syvertsen, 1997) has morphological features most similar to the type species, T. nordenskioeldii Cleve. Most species in this group have short inward and long

148 outward extensions of the SP's, and an LP on the valve mantle (Hasle, 1968; Hasle and

Syvertsen, 1997). A second group (group "B", Hasle, 1968; Hasle and Syvertsen, 1997)

has exceptionally long inward and little or no outward extensions of the SP's, and an LP

on the valve face (Fryxell and Hasle, 1979a; Hasle, 1968; Hasle and Syvertsen, 1997;

Shiono and Koizumi, 2000). This group also has been referred to informally as the "T.

trifulta group" (Shiono and Koizumi, 2000).

We collected and cultured one species, Thalassiosira ritscheri (Hustedt) Hasle, which appears to have a combination of group A and B characters. Thalassiosira ritscheri

has an LP on the valve face and reduced outward extensions of the SP's, but little if

anything is known about the inward extensions of the SP's in this species (Hasle and

Heimdal, 1970; Hasle and Syvertsen, 1997; Johansen and Fryxell, 1985). Existing

photographs of T. ritscheri are either equivocal or suggestive of A-type SP morphology

(Hasle and Heimdal, 1970; Johansen and Fryxell, 1985). In any case, the position of the

LP and reduced outward extensions of the SP's suggest a possible relationship of T.

ritscheri with group B Thalassiosira species. The purpose of our study was to thoroughly

document the cell wall morphology of T. ritscheri and to interpret its features in a

phylogenetic framework, with particular reference to the group A and B classification.

Finally, we propose the transfer of T. ritscheri and many group B Thalassiosira species

into a new genus, Shionodiscus gen. nov.

Materials and Methods

During the 18th Korean Antarctic Program (KARP) expedition, data on surface

hydrography and phytoplankton (including chlorophyll a, quantitative and qualitative

data on community assemblages) were acquired at 31 stations around the South Shetland

149 Islands and northwestern Weddell Sea from 19 December 2004 to 2 January 2005.

Surface temperature and salinity were measured at each station. Water samples were

collected on the "up" casts with a Seabird rosette unit equipped with 2-liter PVC Niskin bottles. Samples for quantitative phytoplankton analyses were fixed and filtered onboard

R/V Yuhzmorgeologiya. Aliquots of 250 ml of discrete water samples were preserved

with glutaraldehyde-25% (final concentration 1% of total volume). Phytoplankton were

enumerated according to the HPMA method (Kang and Fryxell, 1991). Chlorophyll a

concentrations were measured with a Turner Design field fluorometer (TD 700).

Phytoplankton net samples were collected with vertical net tows (20 µm mesh) from 100

m to surface at each station. From this sample, one aliquot was used to isolate live cells

for culture, and the remaining sample was preserved with 25% glutaraldehyde (2% final concentration by volume). The culture of T. ritscheri used in this study was isolated from a sample taken at station LC01 (60.0023°S, 52.2554°W).

Cells were isolated under a dissecting microscope with a Pasteur pipette drawn out over a flame to a small (ca. 100 µm) diameter tip. Cells were rinsed at least twice in sterile f/2 medium (Guillard, 1975; Guillard and Ryther, 1962) before transfer into a culture tube with sterile f/2 medium. Cells were grown at 4oC on a 16:8 light-dark cycle, harvested during exponential growth, and rinsed with distilled water several times.

Rinsed cells were dried onto a coverglass and mounted onto a microscope slide with

Naphrax or filtered onto a 0.22 µm nitrocellulose filter, which then was dried and prepared for observation with a field emission scanning electron microscope (FE-SEM).

Another aliquot of rinsed cells was boiled in 30% hydrogen peroxide for one hour, and then rinsed several times before preparation for FE-SEM observation. Light microscope

150 observations were made with a Zeiss Axioskop, and FE-SEM observations were made with a Hitachi S-4500 FE-SEM.

Results

Hydrography and phytoplankton biomass

Mean temperature, salinity, and total chl-a concentration during the whole cruise period were1.71 °C, 33.97 psu, and 0.92 mg chl-a m-3, respectively. Dominant phytoplankton were nano-sized (<20 µm) phytoflagellates (e.g., Phaeocystis antarctica

Karsten [motile stage], Cryptomonas spp., Gymnodinium spp., and Pyramimonas spp.) and small diatoms (e.g., Minidiscus chilensis Rivera, Fragilariopsis pseudonana (Hasle)

Hasle, and F. cylindrus (Grunow) Krieger). Nanoplanktonic phytoplankton accounted for greater than 89% of total chl-a concentration. Mean chl-a concentration of the micro- sized phytoplankton (>20 µm) such as Thalassiosira spp. accounted for 11% of the total chl a.

Light microscope observations of Thalassiosira ritscheri

Cells are approximately 44–48 µm in diameter (Figs. 5.1 and 5.2), which is at the lower end of the size range of naturally occurring populations (Johansen and Fryxell,

1985). Areolae, approximately 15 in 10µm, occur in fascicles. Areolar size and density are uniform across the valve face (Figs. 5.1 and 5.2). A central cluster of SP's is evident, though difficult to discern in most specimens (Figs. 5.1 and 5.2). The prominent LP is radially oriented, and its position on the valve face varies from subcentral (adjacent to the central cluster of SP's; Fig. 5.2) to midway between the center and valve margin (Fig.

5.1). Marginal SP's are difficult to resolve in the light microscope. Inability to resolve prominent inward extensions in the light microscope (e.g., Figs. 51a, 60a, 67a, and 68a in

151 Johansen and Fryxell, 1985) would suggest that T. ritscheri has short inward (A-type)

extensions of the SP's.

FE-SEM observations

External foramina are occluded partially by irregular, sometimes fingerlike

projections (Figs. 5.3, 5.4). Internally, cribra appear as flat, undifferentiated pore fields

(Figs. 5.5, 5.6). As indicated previously (Hasle and Syvertsen, 1997; Johansen and

Fryxell, 1985), T. ritscheri has a central cluster of SP's, which range in number from approximately 10–30 in our culture specimens (Figs. 5.5, 5.7, 5.8). Some specimens had a few SP’s scattered on the valve face. Central area SP's are surrounded externally by a hyaline area (Fig. 5.8), and the outward extension is little more than a slightly raised rim

(Fig. 5.8). Internally, central area SP's are operculate with short, A-type extensions (Fig.

5.7). Central area SP's have 2–4 satellite pores, each type equally common in our culture specimens (Fig. 5.7). Opercula are small, rounded tabs located immediately above the satellite pores (Fig. 5.7). Satellite pores are surrounded by small, sometimes slightly raised cowlings (Fig. 5.7). FE-SEM observations confirm that T. ritscheri has a single ring of SP's on the valve mantle (Figs. 5.3–5.5, 5.9). Outward extensions of the mantle

SP's consist of slightly raised rims, similar to the valve-face SP's (Figs. 5.4, 5.10). Mantle

SP's have four satellite pores, very rarely three (Figs. 5.6, 5.11).

The LP is radially oriented and on a short stalk (Figs. 5.5, 5.7, 5.9). Externally, the LP appears as a radially oriented, slit-like aperture between the valve center and margin (Figs. 5.3, 5.8). The girdle bands are discontinuous (open), and the valve bears a prominent antiligula at the first girdle band opening (Fig. 5.12). The first girdle band opening is located on the side of valve bearing the LP (Fig. 5.13) Subsequent girdle

152 bands are right-hand spiraled. The first two girdle bands have multiple rows of simple,

round pores, and the remaining bands are imperforated pleurae (Figs. 5.12–5.14).

Discussion

Thalassiosira ritscheri is the only described species in Thalassiosirales that we

know of with this combination of characters: fasciculate striae, an LP on the valve face,

one marginal ring and one central cluster of SP's that are A-type inwardly and B-type

outwardly. The position of the LP on the valve face and the reduced outward extensions

of the SP's suggest a close relationship to group B Thalassiosira species, which are

further characterized by exceptionally long inward extensions of the SP's. Until now, the

internal ultrastructure of the SP's in T. ritscheri was undocumented, and our observations

clearly indicate that they are A-type, with short, "normal" sized inward tubes and satellite

pores covered by tabular opercula. The fossil diatoms T. bipora Shiono and T. depressa

Shiono appear to be close relatives of T. ritscheri. All three species have an LP on the

valve face, short (A-type) SP's internally, and reduced (B-type) SP extensions externally.

Thalassiosira baldaufii Bodén has a central SP cluster and an LP away from the mantle

and so might also be a close relative of T. ritscheri, with a few important differences.

Thalassiosira baldaufii has regularly spaced occluded processes (Bodén, 1993), which

are absent in T. ritscheri. Also, the LP in T. baldaufii is located at the junction of the

valve face and mantle, not on the valve face.

Although possession of a central SP cluster is common among taxa with an LP on

the valve mantle (e.g., T. rotula Meunier and T. antarctica Comber), T. ritscheri is the

only described species we are aware of with both a central SP cluster and an LP

decidedly on the valve face (Hasle and Syvertsen, 1997). Several group B taxa have

153 multiple central SP's, which are typically organized into a modified ring (T. endoseriata

Hasle and G. Fryxell and T. poro-irregulata Hasle and Heimdal) or one or more distinct

rows (T. trifulta G. Fryxell and T. poroseriata (Ramsfjell) Hasle).

At first glance, the unique combination of characters in T. ritscheri suggests

incongruence with the A–B classification of Thalassiosira species. Presence of

fasciculate striae and short inward extensions of the SP's ally T. ritscheri with group A species. An LP on the valve face and reduced outward extensions of the SP's, however, suggest a close relationship to group B species. This appearance of incongruence might simply represent the fact that the original A–B classification did not distinguish between phylogenetically ancestral (plesiomorphic) and derived (apomorphic) character states.

Evidence from phylogenetic analysis of SSU rDNA sequences suggests that

Lauderia and Porosira are the sister clade to the remaining Thalassiosirales. These taxa

typically have a labiate process within or near the mantle and operculate SP's with short inward extensions. These features are common to most Thalassiosira, including the type species, T. nordenskioeldii. This evidence suggests that A-type features are plesiomorphic and B-type are apomorphic, including the LP location well onto the valve face. The combination of plesiomorphic and apomorphic characters in T. ritscheri is consistent with the interpretation of it being the sister species to a clade of the traditionally held group B species. That is, the valve-face LP would diagnose a clade of

T. ritscheri and all group B species. Within this large clade of Thalassiosira species that have an LP on the valve face is a sub-clade of the traditionally held group B species diagnosed by exceptionally long inward extensions of the SP's. Operculate SP's with short inward extensions place T. ritscheri at the base of the group B clade. Trifultate SP's

154 are further interpreted as apomorphic relative to operculate strutted processes, which suggests that "trifultate" species (e.g., T. oestrupii and T. trifulta) form an even less inclusive clade within group B. This interpretation is consistent with the fossil record, as elegantly demonstrated by Shiono (2001), who proposed that the trifultate SP is derived relative to the operculate SP, on the basis of stratigraphic superposition.

Although most group B species fit this hypothesis well, the character combinations present in some species suggest true conflict from a phylogenetic perspective. For example, T. confusa Makarova has reduced outward SP extensions and long, trifultate-type inward SP extensions, but it has an LP located at the edge of the valve face rather than towards the center. The two SP characters (long inward extension and trifultate structure) suggest that the single feature of the LP position near the valve face–mantle junction is a phylogenetic reversal. Thus, T. confusa is most parsimoniously placed with other group B species.

Phylogenetic interpretations of character states used for classification of

Thalassiosira species are generally consistent with stratigraphic studies of the genus

(Shiono and Koizumi, 2001). We believe this body of evidence is sufficient to warrant raising the group B Thalassiosira species to generic status, and we therefore propose a new genus, Shionodiscus gen. nov., to include Thalassiosira species with the autapomorphic conditions of an LP on the valve face, SP's with longer extensions inwards (including but not limited to species with trifultate strutted processes), and reduced or absent outward extensions of the SP's.

Diagnosis

Order: Thalassiosirales Glezer & Makarova 1986

155 Family: Thalassiosiraceae Glezer & Makarova 1986

Shionodiscus Alverson, Kang et. Theriot, gen. nov.

Thalassiosira Cleve affine, a quo rimoportula semper semota a margine, typice

in facie valvae, aliquandro ad oram faciei; fultoportulis typice extensionibus

intrisecis longis et semper extensionibus extrinsicis parvis aut absentibus.

Akin to Thalassiosira Cleve, from which it differs by the labiate process always

distant from the margin, typically on the valve face, sometimes at the edge of

the face; with strutted processes typically with long inward extensions and

always with outward extensions reduced or absent.

Etymology: Shionodiscus is named in honor of Dr. Masamichi Shiono, whose efforts

have contributed greatly to our understanding of the stratigraphic history and evolution of

this group of diatoms.

Typus generis: Shionodiscus oestrupii (Ostenfeld) Alverson, Kang et Theriot, comb.

nov.

Basionym: Coscinosira oestrupii Ostenfeld 1900. Iagttagelser over

Overfladevandets Temperatur, Saltholdighet og Plankton paa islandke og

grønlandske Skibsrouter i 1899: p. 52.

Synonym: Thalassiosira oestrupii (Ostenfeld) Hasle 1972. Taxon 21(4): p. 544.

New nomenclatural combinations for Shionodiscus

Shionodiscus bioculatus (Grunow) Alverson, Kang et Theriot, comb. nov.

Basionym: Coscinodiscus bioculatus Grunow 1884. Wien 28: p. 107, Pl. 3(C),

Fig. 30.

156 Synonym: Thalassiosira bioculata (Grunow) Ostenfeld 1903. Botany of the

Færöes Part 2: p. 564, Fig. 120.

Shionodiscus bioculatus var. exiguus (Grunow) Alverson, Kang et Theriot, comb.

nov.

Basionym: Coscinodiscus bioculatus var. exigua Grunow 1884. Wien 28: p. 108,

Pl. 4(D), Fig. 2.

Synonym: Thalassiosira bioculata var. exigua (Grunow) Hustedt 1928. Dr. L.

Rabenhorst's Kryptogamen-Flora von Deutschland, Österreich und der Schweiz

7(1): p. 332.

Shionodiscus biporus (Shiono) Alverson, Kang et Theriot, comb. nov.

Basionym: Thalassiosira bipora Shiono 2000. Diatom Research 15(1): p. 139,

Figs. 25–27.

Shionodiscus centrus (Shiono) Alverson, Kang et Theriot, comb. nov.

Basionym: Thalassiosira centra Shiono 2000. Diatom Research 15(1): p. 132,

Figs. 1, 2.

Shionodiscus confusus (Makarova) Alverson, Kang et Theriot, comb. nov.

Basionym: Thalassiosira confusa Makarova 1975. Novitates Systematicae

Plantarum Non Vascularium 12: p. 149, Pl. 1, Figs. 1, 2.

Shionodiscus depressus (Shiono) Alverson, Kang et Theriot, comb. nov.

Basionym: Thalassiosira depressa Shiono 2000. Diatom Research 15(1): p. 135,

Fig. 13.

Shionodiscus endoseriatus (Hasle & G. Fryxell) Alverson, Kang et Theriot, comb. nov.

157 Basionym: Thalassiosira endoseriata Hasle & G. Fryxell in G. Fryxell & Hasle

1977. Beiheft zur Nova Hedwigia 54: p. 78, Figs. 45–49.

Shionodiscus exceptiunculus (Shiono) Alverson, Kang et Theriot, comb. nov.

Basionym: Thalassiosira exceptiuncula Shiono 2001. Diatom Research 16(1): p.

84, Figs. 1, 2.

Shionodiscus frenguellii (Kozlova) Alverson, Kang et Theriot, comb. nov.

Basionym: Thalassiosira frenguellii Kozlova 1967. Novitates Systematicae

Plantarum Non Vascularium 1967: p. 58, Fig. 6.

Shionodiscus frenguelliopsis (Fryxell & Johansen) Alverson, Kang et Theriot, comb. nov.

Basionym: Thalassiosira frenguelliopsis Fryxell & Johansen in Johansen &

Fryxell 1985. Phycologia 24(2): p. 168, Figs. 6, 67, 68, 71, 81.

Shionodiscus gracilis (Karsten) Alverson, Kang et Theriot, comb. nov.

Basionym: Coscinodiscus gracilis Karsten 1905. Deutsche Tiefsee-Expedition

2(2): p. 78, Pl. 3, Fig. 4.

Synonym: Thalassiosira gracilis (Karsten) Hustedt 1958. Deutsche Antarktische

Expedition 1938/39 2: pp. 109, 110, Figs. 4–7.

Shionodiscus gracilis var. expectus (VanLandingham) Alverson, Kang et Theriot, comb. nov.

Basionym: Thalassiosira expecta VanLandingham 1978. Catalogue of the Fossil

and Recent Genera and Species of Diatoms and their Synonyms. Part VII.

Rhoicosphenia through Zygoceros: p. 3995.

158 Synonym: Thalassiosira gracilis var. expecta G. Fryxell & Hasle 1979.

Phycologia 18(4): p. 384.

Synonym: Thalassiosira delicatula Hustedt 1958. Deutsche Antarktische

Expedition 1938/39 2: p. 110, Figs. 8–10 (non Ostenfeld 1907).

Shionodiscus latimarginatus (Makarova) Alverson, Kang et Theriot, comb. nov.

Basionym: Thalassiosira latimarginata Makarova 1975. Novitates Systematicae

Plantarum Non Vascularium 12: p. 150, Pl. 1, Figs. 3, 4.

Shionodiscus oestrupii var. planus (Jousé) Alverson, Kang et Theriot, comb. nov.

Basionym: Thalassiosira oestrupii var. plana Jousé 1968. Novitates Systematicae

Plantarum Non Vascularium 1968: p. 15, Pl. 1, Figs. 8, 9.

Shionodiscus oestrupii var. venrickae (G. Fryxell & Hasle) Alverson, Kang et

Theriot, comb. nov.

Basionym: Thalassiosira oestrupii var. venrickae G. Fryxell & Hasle 1980.

American Journal of Botany 67(5): p. 810, Figs. 11–19.

Shionodiscus perpusillus (Kozlova) Alverson, Kang et Theriot, comb. nov.

Basionym: Thalassiosira perpusilla Kozlova 1967. Novitates Systematicae

Plantarum Non Vascularium 1967: p. 60, Figs. 12, 13.

Shionodiscus poro-irregulatus (Hasle & Heimdal) Alverson, Kang et Theriot, comb.

nov.

Basionym: Thalassiosira poro-irregulata Hasle & Heimdal 1970. Beihefte zur

Nova Hedwigia 31: p. 573, Figs. 55–64, 71, 72.

Shionodiscus poroseriatus (Ramsfjell) Alverson, Kang et Theriot, comb. nov.

159 Basionym: Coscinosira poroseriata Ramsfjell 1959. Nyatt Magasin for Botanikk

7: p. 175, Pl. 1, Fig. g and Pl. 2, Fig. a.

Synonym: Thalassiosira poroseriata (Ramsfjell) Hasle 1972. Taxon 21(4): p. 544.

Shionodiscus praeoestrupii (Dumont, Baldauf & Barron) Alverson, Kang et Theriot,

comb. nov.

Basionym: Thalassiosira praeoestrupii Dumont, Baldauf & Barron 1986.

Micropaleontology 32(4): p. 373, Pl. 1, Fig. 2.

Shionodiscus ritscheri (Hustedt) Alverson, Kang et Theriot, comb. nov.

Basionym: Coscinodiscus ritscheri Hustedt 1958. Deutsche Antarktische

Expedition 1938/39 2: pp. 117, 118, Figs. 44–46.

Synonym: Thalassiosira ritscheri (Hustedt) Hasle 1968. Nytt Magasin for

Botanikk 15(3): p. 196.

Shionodiscus tetraoestrupii (Bodén) Alverson, Kang et Theriot, comb. nov.

Basionym: Thalassiosira tetraoestrupii Bodén 1993. Terra Nova 5(1): p. 65, Pl.

1, Fig D.

Shionodiscus tetraoestrupii var. reimeri (Mahood & Barron) Alverson, Kang et

Theriot, comb. nov.

Basionym: Thalassiosira tetraoestrupii var. reimeri Mahood & Barron 1995. A

Century of Diatom Research in North America: A Tribute to the Distinguished

Careers of Charles W. Reimer and Ruth Patrick. p. 2, Fig. 23.

Shionodiscus trifultus (G. Fryxell) Alverson, Kang et Theriot, comb. nov.

Basionym: Thalassiosira trifulta G. Fryxell in Fryxell & Hasle 1979. Nova

Hedwigia Beiheft 64: p. 16, Figs. 1–24.

160 Shionodiscus variantius (Shiono) Alverson, Kang et Theriot, comb. nov.

Basionym: Thalassiosira variantia Shiono 2001. Diatom Research 16(1): p. 86,

Figs. 13, 14.

Shionodiscus voeringensis (Bodén) Alverson, Kang et Theriot, comb. nov.

Basionym: Thalassiosira voeringensis Bodén 1992. Stockholm Contributions in

Geology. 42(3): p. 201, Pl. 4, Figs. 1a–c.

It is clear from morphology and existing phylogenetic analyses of SSU rDNA sequences that Thalassiosira is not a natural evolutionary lineage. In the absence of a densely sampled and well-resolved phylogeny of Thalassiosirales, one useful strategy toward establishing a more natural classification of this group is to interpret cell wall and other characteristics in a phylogenetic framework. This type of analysis can be applied to single characters as easily as it can to entire suites of characters. All classifications are dynamic, testable hypotheses, and those based on sound phylogenetic principles are much more likely to withstand rigorous testing.

Acknowledgements

A. Alverson is especially appreciative to Dr. Greta Fryxell for several useful discussions about Thalassiosirales and for facilitating the collaboration with Dr. Sung-Ho Kang and his lab members, including J. Kang, J. Park, H. Joo, and S. Hong. S.-H. Kang was supported by KOPRI grant PE06060. A. Alverson and E. Theriot were supported by an

NSF PEET grant (DEB 0118883) and an NSF Dissertation Improvement Grant (DEB-

0407815).

161 Literature Cited

Bodén, P. 1992. Quantitative biostratigraphy of Neogene diatoms from the Norwegian

Sea, North Atlantic and North Pacific. Stockholm Contributions in Geology

42:123–215.

Bodén, P. 1993. Taxonomy and stratigraphic occurrence of Thalassiosira tetraoestrupii

sp. nov. and related species in upper Miocene and lower Pliocene sediments from

the Norwegian Sea, North Atlantic and northwest Pacific. Terra Nova 5:61–75.

Dumont, M. P., J. G. Baldauf, and J. A. Barron. 1986. Thalassiosira praeoestrupii: a new

diatom species for recognizing the Miocene/Pliocene epoch boundary in coastal

California. Micropaleontology 32:372–377.

Fryxell, G. A., and G. R. Hasle. 1977. The genus Thalassiosira: some species with a

modified ring of central strutted processes. Beiheft zur Nova Hedwigia 54:67–98.

Fryxell, G. A., and G. R. Hasle. 1979a. The genus Thalassiosira: species with internal

extensions of the strutted processes. Phycologia 18:378–393.

Fryxell, G. A., and G. R. Hasle. 1979b. The genus Thalassiosira: T. trifulta sp. nova and

other species with tricolumnar supports on strutted processes. Beihefte zur Nova

Hedwigia 64:13–40.

Fryxell, G. A., and G. R. Hasle. 1980. The marine diatom Thalassiosira oestrupii:

structure, taxonomy and distribution. American Journal of Botany 67:804–814.

Glezer, Z. I., and I. V. Makarova. 1986. New order and family of diatoms

(Bacillariophyta). Botanicheskii Zhurnal (Leningrad) 71:673–676.

162 Grunow, A. 1884. Die Diatomeen von Franz Josefs-Land. Denkschriften der kaiserlichen

Akademie der Wissenschaften. Mathematisch-naturwissenschaftliche Classe.

Wien 28:53–112.

Guillard, R. R. L. 1975. Culture of phytoplankton for feeding marine invertebrates. Pages

26–60 in Culture of Marine Invertebrate Animals (W. L. Smith, and M. H.

Chanley, eds.). Plenum Press, New York.

Guillard, R. R. L., and J. H. Ryther. 1962. Studies of marine planktonic diatoms. I.

Cyclotella nana Hustedt and Detonula confervacea (Cleve) Gran. Canadian

Journal of Microbiology 8:229–239.

Hasle, G. R. 1968. The valve processes of the centric diatom genus Thalassiosira. Nytt

Magasin for Botanikk 15:193–201.

Hasle, G. R. 1972. The inclusion of Coscinosira Gran (Bacillariophyceae) in

Thalassiosira Cleve. Taxon 21:543–544.

Hasle, G. R., and B. R. Heimdal. 1970. Some species of the centric diatom genus

Thalassiosira studied in the light and electron microscopes. Beihefte zur Nova

Hedwigia 31:543–581.

Hasle, G. R., and E. E. Syvertsen. 1997. Marine Diatoms. Pages 5–386 in Identifying

Marine Phytoplankton (C. R. Tomas, ed.) Academic Press, San Diego.

Hustedt, F. 1928. Die Kieselalgen Deutschlands, Österreichs und der Schweiz. Dr. L.

Rabenhorst's Kryptogamen-Flora von Deutschland, Österreich und der Schweiz

7:273–464.

Hustedt, F. 1958. Diatomeen aus der Antarktis und dem Südatlantik. Deutsche

Antarktische Expedition 1938/39 2:103–191.

163 Johansen, J. R., and G. A. Fryxell. 1985. The genus Thalassiosira (Bacillariophyceae)

studies on species occurring south of the Antarctic Convergence Zone.

Phycologia 24:155–180.

Jousé, A. P. 1968. Species novae Bacillariophytorum in sedimentis fundi Oceani Pacifici

et Maris Ochotensis Inventae. Novitates Systematicae Plantarum Non

Vascularium 1968:12–20.

Kang, S.-H., and G. A. Fryxell. 1991. Most abundant diatom species in water column

assemblages from five ODP Leg 119 drill sites in Prydz Bay, Antarctica:

Distributional patterns. Proceedings of the Ocean Drilling Program, Scientific

Results 119:645–666.

Karsten, G. 1905. Das Phytoplankton des Antarktischen Meeres nach dem Material der

deutschen Tiefsee-Expedition 1898–1899. Deutsche Tiefsee-Expedition 2:1–136.

Kozlova, O. G. 1967. De speciebus Bacillariophytorum novis e partibus Antarcticis

Oceanorum Indici et Pacifici. Novitates Systematicae Plantarum Non

Vascularium 1967:54–62.

Mahood, A. D., and J. A. Barron. 1995. Thalassiosira tetraoestrupii var. reimeri var.

nov., a distinctive diatom from the late Pliocene of the Southern Ocean. Pages 1–8

in A Century of Diatom Research in North America: A Tribute to the

Distinguished Careers of Charles W. Reimer and Ruth Patrick (J. P. Kociolek, and

M. J. Sullivan, eds.). Koeltz Scientific Books, Champaign.

Makarova, I. V. 1975. Species novae generis Thalassiosira Cleve e Mari Barentz.

Novitates Systematicae Plantarum Non Vascularium 12:149–152.

164 Medlin, L. K., and I. Kaczmarska. 2004. Evolution of the diatoms V: Morphological and

cytological support for the major clades and a taxonomic revision. Phycologia

43:245–270.

Ostenfeld, C. H. 1900. Plankton. Pages 45–93 in Iagttagelser over Overfladevandets

Temperatur, Saltholdighet og Plankton paa islandke og grønlandske Skibsrouter i

1899 (M. Knudsen, and C. H. Ostenfeld, eds.). C.F. Wandel, Copenhagen.

Ostenfeld, C. H. 1903. Phytoplankton from the sea around the Færöes. Pages 558–611 in

Botany of the Færöes Part 2 (E. Bojesen, ed.) Det Nordiske Forlag, Copenhagen.

Ramsfjell, E. 1959. Two new phytoplankton species from the Norwegian Sea, the diatom

Coscinosira poroseriata and the dinoflagellate Goniaulax parva. Nyatt Magasin

for Botanikk 7:175–177.

Shiono, M. 2000. Three new species in the Thalassiosira trifulta group in late neogene

sediments from the northwest Pacific Ocean. Diatom Research 15:131–148.

Shiono, M. 2001. Two new species of the Thalassiosira trifulta group in late Pliocene

sediments of the northwest Pacific Ocean. Diatom Research 16:83–92.

Shiono, M., and I. Koizumi. 2000. Taxonomy of the Thalassiosira trifulta group in late

Neogene sediments from the northwest Pacific Ocean. Diatom Research 15:355–

382.

Shiono, M., and I. Koizumi. 2001. Phylogenic evolution of the Thalassiosira trifulta

group (Bacillariophyceae) in the northwestern Pacific Ocean. Journal of the

Geological Society of Japan 107:496–514.

165 Stoermer, E. F., and M. L. Julius. 2003. Centric Diatoms. Pages 559–594 in Freshwater

Algae of North America (J. D. Wehr, and R. G. Sheath, eds.). Academic Press,

Amsterdam.

Theriot, E., and K. Serieyssol. 1994. Phylogenetic systematics as a guide to

understanding features and potential morphological characters of the centric

diatom family Thalassiosiraceae. Diatom Research 9:429–450.

Theriot, E., E. F. Stoermer, and H. Håkansson. 1987. Taxonomic interpretation of the

rimoportula of freshwater genera in the centric diatom family Thalassiosiraceae.

Diatom Research 2:251–265.

Vanlandingham, S. L. 1978. Catalogue of the Fossil and Recent Genera and Species of

Diatoms and their Synonyms. Part VII. Rhoicosphenia through Zygoceros. Verlag

Von J. Cramer, Lehre.

166 Figures 5.1–5.6. Cell wall morphology of Thalassiosira ritscheri with light microscopy

(Figs. 5.1, 5.2) and field emission microscopy (Figs. 5.3–5.6). Arrows indicate the position of the labiate process. Figs. 5.1, 5.2. Light micrographs showing the variable position of the labiate process on the valve face. Fig. 5.3. External view of valve. Fig.

5.4. External view of valve margin showing areolar structure and reduced outward openings of the marginal strutted processes. Fig. 5.5. Internal view of valve. Fig. 5.6.

Internal view of valve margin showing flat cribra arranged in fascicles, a marginal strutted process, and a strutted process on the valve face. Scale bar=20µm (Figs. 5.1–5.3,

5.5), Scale bar=5 µm (Figs. 5.4, 5.6).

167 Figures 5.1-5.6

168 Figures 5.7–5.12. Field emission micrographs of Thalassiosira ritscheri. Fig. 5.7.

Internal view of the central cluster of strutted processes and adjacent labiate process. Fig.

5.8. External view of the central cluster of strutted processes surrounded by hyaline area.

The arrow indicates the outward opening of the labiate process. Fig. 5.9. Internal view of valve showing single ring of marginal strutted processes. Fig. 5.10. External view of marginal strutted process openings, which consist of a slightly raised rim. The rim is not fully formed on the leftmost strutted process, exposing at least one satellite pore. Fig.

5.11. Internal view of marginal strutted processes with three or four satellite pores. Fig.

5.12. External view of valvar antiligula, first girdle-band opening, and ligula of the second girdle band. The first two girdle bands are perforated by several rows of regularly spaced, subcircular pores. Scale bar=2µm (Figs. 5.7, 5.8, 5.10–5.12), Scale bar=20 µm

(Fig. 5.9).

169 Figures 5.7-5.12

170

Figures 5.13, 5.14. Field emission micrographs of Thalassiosira ritscheri. Fig. 5.13. External view showing the relative position of the first girdle band opening to the labiate process (arrow). Scale bar=5µm. Fig. 5.14. External view showing right-hand spiral orientation of consecutive girdle bands. The first two girdle bands bear several rows of pores, whereas subsequent bands are imperforated.

171

APPENDICES. Strict consensus trees from maximum parsimony analysis of combined chloroplast, nuclear, and total combined (chloroplast+nuclear) datasets

172

Appendix 1. Strict consensus of nine most parsimonious trees from maximum parsimony analysis of the combined chloroplast dataset (See. Chapter 2 for details).

173

Appendix 2. Strict consensus of 26 most parsimonious trees from maximum parsimony analysis of the combined nuclear dataset (See. Chapter 2 for details).

174 Appendix 3. Strict consensus of two most parsimonious trees from maximum parsimony analysis of the total combined (chloroplast+nuclear) dataset (See. Chapter 2 for details).

175

Bibliography

Alverson, A. J. 2006. Phylogeny and evolutionary ecology of thalassiosiroid diatoms.

PhD thesis. The University of Texas at Austin, Plant Biology Graduate Program.

Alverson, A. J., J. J. Cannone, R. R. Gutell, and E. C. Theriot. 2006. The evolution of

elongate shape in diatoms. Journal of Phycology 42:655–668.

Alverson, A. J., and L. Kolnick. 2005. Intragenomic nucleotide polymorphism among

small subunit (18S) rDNA paralogs in the diatom genus Skeletonema

(Bacillariophyta). Journal of Phycology 41:1248–1257.

Alverson, A. J., and E. C. Theriot. 2005. Comments on recent progress toward

reconstructing the diatom phylogeny. Journal of Nanoscience and

Nanotechnology 5:57–62.

Andersen, R. A. 2004. Biology and systematics of heterokont and haptophyte algae.

American Journal of Botany 91:1508–1522.

Armbrust, E. V., J. A. Berges, C. Bowler, B. R. Green, D. Martinez, N. H. Putnam, S.

Zhou, A. E. Allen, K. E. Apt, M. Bechner, M. A. Brzezinski, B. K. Chaal, A.

Chiovitti, A. K. Davis, M. S. Demarest, J. C. Detter, T. Glavina, D. Goodstein, M.

Z. Hadi, U. Hellsten, M. Hildebrand, B. D. Jenkins, J. Jurka, V. V. Kapitonov, N.

Kroger, W. W. Y. Lau, T. W. Lane, F. W. Larimer, J. C. Lippmeier, S. Lucas, M.

Medina, A. Montsant, M. Obornik, M. S. Parker, B. Palenik, G. J. Pazour, P. M.

Richardson, T. A. Rynearson, M. A. Saito, D. C. Schwartz, K. Thamatrakoln, K.

Valentin, A. Vardi, F. P. Wilkerson, and D. S. Rokhsar. 2004. The genome of the

176 diatom Thalassiosira pseudonana: ecology, evolution, and metabolism. Science

306:79–86.

Behrenfeld, M. J., and P. G. Falkowski. 1997. Photosynthetic rates derived from satellite-

based chlorophyll concentration. Limnology & Oceanography 42:1–20.

Beszteri, B., E. Acs, and L. K. Medlin. 2005. Ribosomal DNA sequence variation among

sympatric strains of the Cyclotella meneghiniana complex (Bacillariophyceae)

reveals cryptic diversity. Protist 156:317–333.

Bhattacharyya, P., and B. E. Volcani. 1980. Sodium-dependent silicate transport in the

apochlorotic marine diatom Nitzschia alba. Proceedings of the National Academy

of Sciences of the United States of America 77:6386–6390.

Bielawski, J. P., and Z. H. Yang. 2004. A maximum likelihood method for detecting

functional divergence at individual codon sites, with application to gene family

evolution. Journal of Molecular Evolution 59:121–132.

Bienfang, P. K., P. J. Harrison, and L. M. Quarmby. 1982. Sinking Rate Response to

Depletion of Nitrate, Phosphate and Silicate in 4 Marine Diatoms. Marine

Biology 67:295–302.

Bodén, P. 1992. Quantitative biostratigraphy of Neogene diatoms from the Norwegian

Sea, North Atlantic and North Pacific. Stockholm Contributions in Geology

42:123–215.

Bodén, P. 1993. Taxonomy and stratigraphic occurrence of Thalassiosira tetraoestrupii

sp. nov. and related species in upper Miocene and lower Pliocene sediments from

the Norwegian Sea, North Atlantic and northwest Pacific. Terra Nova 5:61–75.

177 Brzezinski, M. A. 1992. Cell-cycle effects on the kinetics of silicic acid uptake and

resource competition among diatoms. Journal of Plankton Research 14:1511–

1539.

Brzezinski, M. A., R. J. Olson, and S. W. Chisholm. 1990. Silicon availability and cell-

cycle progression in marine diatoms. Marine Ecology Progress Series 67:83–96.

Buckley, T. R. 2002. Model misspecification and probabilistic tests of topology:

Evidence from empirical data sets. Systematic Biology 51:509–523.

Chen, C. P., A. Kernytsky, and B. Rost. 2002. Transmembrane helix predictions

revisited. Protein Science 11:2774–2791.

Conley, D. J. (ed) 2002. Terrestrial ecosystems and the global biogeochemical silica

cycle. Global Biogeochemical Cycles,

Conley, D. J., S. S. Kilham, and E. Theriot. 1989. Differences in silica content between

marine and freshwater diatoms. Limnology & Oceanography 34:205–213.

Conway, H., and P. Harrison. 1977. Marine diatoms grown in chemostats under silicate

or ammonium limitation IV. Transient response of Chaetoceros debilis,

Skeletonema costatum, and Thalassiosira gravida to a single addition of the

limiting nutrient. Marine Biology 43:33–43.

Conway, H., P. Harrison, and C. Davis. 1976. Marine diatoms grown in chemostats under

silicate or ammonium limitation II. Transient response of Skeletonema costatum

to a single addition of the limiting nutrient. Marine Biology 35:187–199.

Culver, M. E., and W. O. Smith. 1989. Effects of environmental variation on sinking

rates of marine phytoplankton. Journal of Phycology 25:262–270.

178 Cuthbertson, J. M., D. A. Doyle, and M. S. P. Sansom. 2005. Transmembrane helix

prediction: a comparative evaluation and analysis. Protein Engineering Design &

Selection 18:295–308.

Darwin, C. 1872. The Origin of Species, 6th edition. J.M. Dent & Sons, London.

Davis, C. O. 1976. Continuous culture of marine diatoms under silicate limitation. 2.

Effect of light intensity on growth and nutrient uptake of Skeletonema costatum.

Journal of Phycology 12:291–300.

Dawes, C. J. 1998. Abiotic factors. Pages 16–44 in Marine Botany. John Wiley & Sons,

Inc., New York.

Del Amo, Y., and M. A. Brzezinski. 1999. The chemical form of dissolved SI taken up

by marine diatoms. Journal of Phycology 35:1162–1170.

Dumont, M. P., J. G. Baldauf, and J. A. Barron. 1986. Thalassiosira praeoestrupii: a new

diatom species for recognizing the Miocene/Pliocene epoch boundary in coastal

California. Micropaleontology 32:372–377.

Edgar, L. A. 1980. Fine structure of Caloneis amphisbaena (Bacillariophyceae). Journal

of Phycology 16:62–72.

Edlund, M. B., C. M. Taylor, C. L. Schelske, and E. F. Stoermer. 2000. Thalassiosira

baltica (Grunow) Ostenfeld (Bacillariophyta), a new exotic species in the Great

Lakes. Canadian Journal of Fisheries & Aquatic Sciences 57:610–615.

Falkowski, P. G., R. T. Barber, and V. Smetacek. 1998. Biogeochemical controls and

feedbacks on ocean primary production. Science 281:200–206.

Felsenstein, J. 1985a. Confidence limits on phylogenies: an approach using the bootstrap.

Evolution 39:783–791.

179 Felsenstein, J. 1985b. Phylogenies and the comparative method. American Naturalist

125:1–15.

Field, C. B., M. J. Behrenfeld, J. T. Randerson, and P. Falkowski. 1998. Primary

production of the biosphere: integrating terrestrial and oceanic components.

Science 281:237–240.

Flynn, K. J., and V. Martin-Jezequel. 2000. Modelling Si-N-limited growth of diatoms.

Journal of Plankton Research 22:447–472.

Fritsch, F. E. 1935. The Structure and Reproduction of the AlgaeCambridge University

Press, London.

Fryxell, G. A., and G. R. Hasle. 1977. The genus Thalassiosira: some species with a

modified ring of central strutted processes. Beiheft zur Nova Hedwigia 54:67–98.

Fryxell, G. A., and G. R. Hasle. 1979a. The genus Thalassiosira: species with internal

extensions of the strutted processes. Phycologia 18:378–393.

Fryxell, G. A., and G. R. Hasle. 1979b. The genus Thalassiosira: T. trifulta sp. nova and

other species with tricolumnar supports on strutted processes. Beihefte zur Nova

Hedwigia 64:13–40.

Fryxell, G. A., and G. R. Hasle. 1980. The marine diatom Thalassiosira oestrupii:

structure, taxonomy and distribution. American Journal of Botany 67:804–814.

Genkal, S. I., and H. Håkansson. 1990. Problem of distinguishing the newly described

diatom genus Pseudostephanodiscus. Diatom Research 5:15–24.

Glezer, Z. I., and I. V. Makarova. 1986. New order and family of diatoms

(Bacillariophyta). Botanicheskii Zhurnal (Leningrad) 71:673–676.

180 Goldman, N., J. P. Anderson, and A. G. Rodrigo. 2000. Likelihood-based tests of

topologies in phylogenetics. Systematic Biology 49:652–670.

Goldman, N., and Z. H. Yang. 1994. Codon-based model of nucleotide substitution for

protein-coding DNA sequences. Molecular Biology and Evolution 11:725–736.

Goloboff, P. A., J. S. Farris, and K. C. Nixon. 2004. TNT. Cladistics 20:84.

Gordon, R., F. Sterrenberg, and K. Sandhage (eds) 2005. A Special Issue on Diatom

Nanotechnology. Journal of Nanoscience and Nanotechnology, 178 pages.

Grachev, M., T. Sherbakova, Y. Masyukova, and Y. Likhoshway. 2005. A potential zinc-

binding motif in silicic acid transport proteins of diatoms. Diatom Research

20:409–411.

Grachev, M. A., N. N. Denikina, S. I. Belikov, E. V. Likhoshvai, M. V. Usol'tseva, I. V.

Tikhonova, R. V. Adel'shin, S. A. Kler, and T. A. Scherbakova. 2002. Elements

of the active center of silicon transporters in diatoms. Molecular Biology 36:534–

536.

Grunow, A. 1884. Die Diatomeen von Franz Josefs-Land. Denkschriften der kaiserlichen

Akademie der Wissenschaften. Mathematisch-naturwissenschaftliche Classe.

Wien 28:53–112.

Guillard, R. R. L. 1975. Culture of phytoplankton for feeding marine invertebrates. Pages

26–60 in Culture of Marine Invertebrate Animals (W. L. Smith, and M. H.

Chanley, eds.). Plenum Press, New York.

Guillard, R. R. L., P. Kilham, and T. A. Jackson. 1973. Kinetics of silicon-limited growth

in the marine diatom Thalassiosira pseudonana Hasle and Heimdal (Cyclotella

nana Hustedt). Journal of Phycology 9:233–237.

181 Guillard, R. R. L., and J. H. Ryther. 1962. Studies of marine planktonic diatoms. I.

Cyclotella nana Hustedt and Detonula confervacea (Cleve) Gran. Canadian

Journal of Microbiology 8:229–239.

Håkansson, H. 2002. A compilation and evaluation of species in the general

Stephanodiscus, Cyclostephanos and Cyclotella with a new genus in the family

Stephanodiscaceae. Diatom Research 17:1–139.

Håkansson, H., and G. Khursevich. 1997. Tertiarius gen. nov., a new genus in the

Bacillariophyceae, the transfer of some cyclotelloid species and a comparison to

closely related genera. Diatom Research 12:19–33.

Harper, H. E., and A. H. Knoll. 1974. Silica, diatoms, and Cenozoic radiolarian

evolution. Geology 3:175–177.

Harrison, P. J., H. L. Conway, R. W. Holmes, and C. O. Davis. 1977. Marine diatoms

grown in chemostats under silicate or ammonium limitation. 3. Cellular chemical

composition and morphology of Chaetoceros debilis, Skeletonema costatum, and

Thalassiosira gravida. Marine Biology 43:19–31.

Harvey, P. H., and M. Pagel. 1991. The Comparative Method in Evolutionary

BiologyOxford University Press, Oxford.

Hasle, G. R. 1968. The valve processes of the centric diatom genus Thalassiosira. Nytt

Magasin for Botanikk 15:193–201.

Hasle, G. R. 1972. The inclusion of Coscinosira Gran (Bacillariophyceae) in

Thalassiosira Cleve. Taxon 21:543–544.

Hasle, G. R. 1973. Thalassiosiraceae, a new diatom family. Norwegian Journal of Botany

20:67–69.

182 Hasle, G. R. 1978a. Some freshwater and brackish water species of the diatom genus

Thalassiosira Cleve. Phycologia 17:263–292.

Hasle, G. R. 1978b. Some Thalassiosira species with one central process. Norwegian

Journal of Botany 25:77–110.

Hasle, G. R., and G. A. Fryxell. 1977. The genus Thalassiosira: some species with a

linear areola array. Nova Hedwigia Beiheft 54:14–60.

Hasle, G. R., and B. R. Heimdal. 1970. Some species of the centric diatom genus

Thalassiosira studied in the light and electron microscopes. Beihefte zur Nova

Hedwigia 31:543–581.

Hasle, G. R., and C. B. Lange. 1989. Freshwater and brackish water Thalassiosira

(Bacillariophyceae) taxa with tangentially undulated valves. Phycologia 28:120–

135.

Hasle, G. R., and E. E. Syvertsen. 1985. Thalassiosiropsis, a new diatom genus from the

fossil records. Micropaleontology (New York) 31:82–91.

Hasle, G. R., and E. E. Syvertsen. 1997. Marine Diatoms. Pages 5–386 in Identifying

Marine Phytoplankton (C. R. Tomas, ed.) Academic Press, San Diego.

Herth, W. 1978. Special chitin-fibril-synthesizing apparatus in centric diatom Cyclotella.

Naturwissenschaften 65:260–261.

Herth, W. 1979. Site of beta-chitin fibril formation in centric diatoms. 2. Chitin-forming

cytoplasmic structures. Journal of Ultrastructure Research 68:16–27.

Herth, W., and W. Barthlott. 1979. Site of beta-chitin fibril formation in centric diatoms.

1. Pores and fibril formation. Journal of Ultrastructure Research 68:6–15.

183 Hildebrand, M. 2000. Silicic acid transport and its control during cell wall silicification in

diatoms. Pages 171–188 in Biomineralization: From Biology to Biotechnology

and Medical Applications (E. Bäurlein, ed.) Wiley-VCH, Weinheim.

Hildebrand, M., K. Dahlin, and B. E. Volcani. 1998. Characterization of a silicon

transporter gene family in Cylindrotheca fusiformis: sequences, expression

analysis, and identification of homologs in other diatoms. Molecular and General

Genetics 260:480–486.

Hildebrand, M., D. R. Higgins, K. Busser, and B. E. Volcani. 1993. Silicon-responsive

cDNA clones isolated from the marine diatom Cylindrotheca fusiformis. Gene

132:213–218.

Hildebrand, M., B. E. Volcani, W. Gassmann, and J. I. Schroeder. 1997. A gene family of

silicon transporters. Nature 385:688–689.

Hildebrand, M., and R. Wetherbee. 2003. Components and control of silicification in

diatoms. Progress in Molecular and Subcellular Biology 33:11–57.

Hillis, D. M., B. K. Mable, and C. Moritz. 1996. Applications of molecular systematics:

the state of the field and a look into the futurein Molecular Systematics (D. M.

Hillis, C. Moritz, and B. K. Mable, eds.). Sinauer Associates, Sunderland,

Massachusetts.

Hirokawa, T., S. Boon-Chieng, and S. Mitaku. 1998. SOSUI: classification and

secondary structure prediction system for membrane proteins. Bioinformatics

14:378–379.

Huelsenbeck, J. P., D. M. Hillis, and R. Jones. 1996. Parametric bootstrapping in

molecular phylogenetics: applications and performance. Pages 19–45 in

184 Molecular Zoology: Advances, Strategies and Protocols (J. D. Ferraris, and S. R.

Palumbi, eds.). Wiley-Liss, New York.

Hustedt, F. 1928. Die Kieselalgen Deutschlands, Österreichs und der Schweiz. Dr. L.

Rabenhorst's Kryptogamen-Flora von Deutschland, Österreich und der Schweiz

7:273–464.

Hustedt, F. 1958. Diatomeen aus der Antarktis und dem Südatlantik. Deutsche

Antarktische Expedition 1938/39 2:103–191.

Iler, R. K. 1979. The Chemistry of Silica: Solubility, Polymerization, Colloid and Surface

Properties, and BiochemistryJohn Wiley & Sons, New York.

Johansen, J. R., and G. A. Fryxell. 1985. The genus Thalassiosira (Bacillariophyceae)

studies on species occurring south of the Antarctic Convergence Zone.

Phycologia 24:155–180.

Jousé, A. P. 1968. Species novae Bacillariophytorum in sedimentis fundi Oceani Pacifici

et Maris Ochotensis Inventae. Novitates Systematicae Plantarum Non

Vascularium 1968:12–20.

Julius, M. L. 2000. Phylogeny of the Cyclostephanoid Diatoms: An Investigation of their

Morphology and Stratigraphy. PhD thesis. University of Michigan, Natural

Resources and Environment, 209 pages.

Julius, M. L., and Y. Tanimura. 2001. Cladistic analysis of plicated Thalassiosira

(Bacillariophyceae). Phycologia 40:111–122.

Kaczmarska, I., M. Beaton, A. C. Benoit, and L. K. Medlin. 2006. Molecular phylogeny

of selected members of the order Thalassiosirales (Bacillariophyta) and evolution

of the fultoportula. Journal of Phycology 42:121–138.

185 Kang, S.-H., and G. A. Fryxell. 1991. Most abundant diatom species in water column

assemblages from five ODP Leg 119 drill sites in Prydz Bay, Antarctica:

Distributional patterns. Proceedings of the Ocean Drilling Program, Scientific

Results 119:645–666.

Karsten, G. 1905. Das Phytoplankton des Antarktischen Meeres nach dem Material der

deutschen Tiefsee-Expedition 1898–1899. Deutsche Tiefsee-Expedition 2:1–136.

Kass, R. E., and A. E. Raftery. 1995. Bayes Factors. Journal of the American Statistical

Association 90:773–795.

Katz, M. E., Z. V. Finkel, D. Grzebyk, A. H. Knoll, and P. G. Falkowski. 2004.

Evolutionary trajectories and biogeochemical impacts of marine eukaryotic

phytoplankton. Annual Review of Ecology Evolution and Systematics 35:523–

556.

Kilham, S. S., D. A. Kreeger, C. E. Goulden, and S. Lynn. 1998. COMBO: A defined

freshwater culture medium for algae and zooplankton. Hydrobiologia 377:147–

159.

Kozlova, O. G. 1967. De speciebus Bacillariophytorum novis e partibus Antarcticis

Oceanorum Indici et Pacifici. Novitates Systematicae Plantarum Non

Vascularium 1967:54–62.

Krebs, W. N. 1990. The biochronology of freshwater planktonic diatom communities in

western North America. Pages 485–499 in Proceedings of the 11th International

Diatom Symposium (J. P. Kociolek, ed.) California Academy of Sciences, San

Francisco.

186 Krebs, W. N., and J. P. Bradbury. 1982. Neogene Lacustrine Diatom Biostratigraphy of

Western Snake River Basin, Idaho and Oregon. AAPG Bulletin 66:1693–1693.

Krebs, W. N., J. P. Bradbury, and E. Theriot. 1987. Neogene and Quaternary lacustrine

diatom biochronology western USA. Palaios 2:505–513.

Lane, T. W., M. A. Saito, G. N. George, I. J. Pickering, R. C. Prince, and F. M. M.

Morel. 2005. A cadmium enzyme from a marine diatom. Nature 435:42–42.

Lebour, M. V. 1930. The planktonic diatoms of northern seas. Royal Society Publication

no. 116:1–244.

Loginova, L. P. 1990. Classification of the diatom genus Cyclotella. Pages 37–46 in

Proceedings of the 10th International Symposium on Recent and Fossil Diatoms

(H. Simola, ed.) Otto Koeltz, Koenigstein.

Lowe, R. L. 1975. Comparative ultrastructure of the valves of some Cyclotella species

(Bacillariophyceae). Journal of Phycology 11:415–424.

Maddison, W. P., and D. R. Maddison. 2003. MacClade. Sinauer Associates, Inc.

Maddison, W. P., and D. R. Maddison. 2006. Mesquite: a modular system for

evolutionary analysis, version 1.1.

Mahood, A. D., and J. A. Barron. 1995. Thalassiosira tetraoestrupii var. reimeri var.

nov., a distinctive diatom from the late Pliocene of the Southern Ocean. Pages 1–8

in A Century of Diatom Research in North America: A Tribute to the

Distinguished Careers of Charles W. Reimer and Ruth Patrick (J. P. Kociolek, and

M. J. Sullivan, eds.). Koeltz Scientific Books, Champaign.

Makarova, I. V. 1975. Species novae generis Thalassiosira Cleve e Mari Barentz.

Novitates Systematicae Plantarum Non Vascularium 12:149–152.

187 Mann, D. G. 1999. Crossing the Rubicon: the effectiveness of the marine/freshwater

interface as a barrier to the migration of diatom germplasm. Pages 1–21 in

Proceedings of the 14th International Diatom Symposium (S. Mayama, M. Idei,

and I. Koizumi, eds.). Koeltz Scientific Books, Koenigstein.

Martin-Jezequel, V., M. Hildebrand, and M. A. Brzezinski. 2000. Silicon metabolism in

diatoms: Implications for growth. Journal of Phycology 36:821–840.

Medlin, L. K., and I. Kaczmarska. 2004. Evolution of the diatoms V: Morphological and

cytological support for the major clades and a taxonomic revision. Phycologia

43:245–270.

Medlin, L. K., W. H. C. F. Kooistra, R. Gersonde, and U. Wellbrock. 1996. Evolution of

the diatoms (Bacillariophyta): III. Molecular evidence for the origin of the

Thalassiosirales. Nova Hedwigia Beiheft 112:221–234.

Moore, T. C. 1969. Radiolaria: change is skeletal weight and resistance to solution.

Bulletin of the Geological Society of America 80:2103–2108.

Muse, S. V., and B. S. Gaut. 1994. A likelihood approach for comparing synonymous and

nonsynonymous nucleotide substitution rates, with application to the chloroplast

genome. Molecular Biology and Evolution 11:715–724.

Nelson, D. M., P. Treguer, M. A. Brzezinski, A. Leynaert, and B. Queguiner. 1995.

Production and dissolution of biogenic silica in the ocean: revised global

estimates, comparison with regional data and relationship to biogenic

sedimentation. Global Biogeochemical Cycles 9:359–372.

188 Nielsen, R., and Z. H. Yang. 1998. Likelihood models for detecting positively selected

amino acid sites and applications to the HIV-1 envelope gene. Genetics 148:929–

936.

Nylander, J. A. A. 2004. MrModeltest v2. Program distributed by the author.

Evolutionary Biology Centre, Uppsala University.

Nylander, J. A. A., F. Ronquist, J. P. Huelsenbeck, and J. L. Nieves-Aldrey. 2004.

Bayesian phylogenetic analysis of combined data. Systematic Biology 53:47–67.

Ohno, S. 1970. Evolution by gene duplicationSpringer-Verlag, Berlin, New York.

Olsen, S., and E. Paasche. 1986. Variable kinetics of silicon limited growth in

Thalassiosira pseudonana (Bacillariophyceae) in response to changed chemical

composition of the growth medium. British Phycological Journal 21:183–190.

Ostenfeld, C. H. 1900. Plankton. Pages 45–93 in Iagttagelser over Overfladevandets

Temperatur, Saltholdighet og Plankton paa islandke og grønlandske Skibsrouter i

1899 (M. Knudsen, and C. H. Ostenfeld, eds.). C.F. Wandel, Copenhagen.

Ostenfeld, C. H. 1903. Phytoplankton from the sea around the Færöes. Pages 558–611 in

Botany of the Færöes Part 2 (E. Bojesen, ed.) Det Nordiske Forlag, Copenhagen.

Paasche, E. 1973a. The influence of cell size on growth rate, silica content, and some

other properties of four marine diatom species. Norwegian Journal of Botany

20:197–204.

Paasche, E. 1973b. Silicon and the ecology of marine plankton diatoms. I. Thalassiosira

pseudonana (Cyclotella nana) grown in a chemostat with silicate as limiting

nutrient. Marine Biology (Berlin) 19:117–126.

189 Paasche, E. 1973c. Silicon and the ecology of marine plankton diatoms. II. Silicate-

uptake kinetics in five diatom species. Marine Biology (Berlin) 19:262–269.

Paasche, E. 1975. Growth of the plankton diatom Thalassiosira nordenskioldii Cleve at

low silicate concentrations. Journal of Experimental Marine Biology and Ecology

18:173–183.

Paasche, E. 1980a. Silicon. Pages 259–284 in The Physiological Ecology of

Phytoplankton (I. Morris, ed.) University of California Press, Berkeley.

Paasche, E. 1980b. Silicon content of 5 marine plankton diatom species measured with a

rapid filter method. Limnology & Oceanography 25:474–480.

Pagel, M. 1994. Detecting correlated evolution on phylogenies: a general method for the

comparative analysis of discrete characters. Proceedings of the Royal Society of

London Series B-Biological Sciences 255:37–45.

Pagel, M. 1998. Discrete, version 4.0. Program distributed by the author. School of

Animal and Microbial Sciences, University of Reading.

Pagel, M. 1999. The maximum likelihood approach to reconstructing ancestral character

states of discrete characters on phylogenies. Systematic Biology 48:612–622.

Persson, B., and P. Argos. 1997. Prediction of membrane protein topology utilizing

multiple sequence alignments. Journal of Protein Chemistry 16:453–457.

Pither, J., and L. W. Aarssen. 2005a. Environmental specialists: their prevalence and their

influence on community-similarity analyses. Ecology Letters 8:261–271.

Pither, J., and L. W. Aarssen. 2005b. The evolutionary species pool hypothesis and

patterns of freshwater diatom diversity along a pH gradient. Journal of

Biogeography 32:503–513.

190 Pither, J., and L. W. Aarssen. 2006. How prevalent are pH-specialists diatoms? A reply to

Telford et al. (2006). Ecology Letters 9:E6–E12.

Ramsfjell, E. 1959. Two new phytoplankton species from the Norwegian Sea, the diatom

Coscinosira poroseriata and the dinoflagellate Goniaulax parva. Nyatt Magasin

for Botanikk 7:175–177.

Reeder, T. W. 2003. A phylogeny of the Australian Sphenomorphus group (Scincidae :

Squamata) and the phylogenetic placement of the crocodile skinks (Tribolonotus):

Bayesian approaches to assessing congruence and obtaining confidence in

maximum likelihood inferred relationships. Molecular Phylogenetics and

Evolution 27:384–397.

Ronquist, F., and J. P. Huelsenbeck. 2003. MrBayes 3: Bayesian phylogenetic inference

under mixed models. Bioinformatics 19:1572–1574.

Rost, B., P. Fariselli, and R. Casadio. 1996. Topology prediction for helical

transmembrane proteins at 86% accuracy. Protein Science 5:1704–1718.

Round, F. E., R. M. Crawford, and D. G. Mann. 1990. The Diatoms: Biology &

Morphology of the GeneraCambridge University Press, Cambridge.

Round, F. E., and P. A. Sims. 1980. The distribution of diatom genera in marine and

freshwater environments and some evolutionary considerations. Pages 301–320 in

Proceedings of the Sixth Symposium on Recent and Fossil Diatoms (R. Ross, ed.)

Otto Koeltz Science Publishers, Hirschberg.

Ruck, E. C., and J. P. Kociolek. 2004. Preliminary Phylogeny of the Family Surirellaceae

(Bacillariophyta)J. Cramer, Berlin.

191 Rueter, J. G., S. W. Chisholm, and F. M. M. Morel. 1981. Effects of copper toxicity on

silicic-acid uptake and growth in Thalassiosira pseudonana. Journal of Phycology

17:270–278.

Rueter, J. G., and F. M. M. Morel. 1981. The Interaction between zinc deficiency and

copper toxicity as it affects the silicic-acid uptake mechanisms in Thalassiosira

pseudonana. Limnology and Oceanography 26:67–73.

Schluter, D., T. Price, A. O. Mooers, and D. Ludwig. 1997. Likelihood of ancestor states

in adaptive radiation. Evolution 51:1699–1711.

Schoniger, M., and A. Von Haeseler. 1994. A Stochastic model for the evolution of

autocorrelated DNA sequences. Molecular Phylogenetics and Evolution 3:240–

247.

Serieyssol, K. K. 1980. Cyclotella species of late Miocene age from St. Bauzile, France.

Pages 27–42 in Proceedings of the Sixth Symposium on Recent and Fossil

Diatoms (R. Ross, ed.) Otto Koeltz Science Publishers, Koenigstein.

Shapiro, B., A. Rambaut, and A. J. Drummond. 2006. Choosing appropriate substitution

models for the phylogenetic analysis of protein-coding sequences. Mol Biol Evol

23:7–9.

Sherbakova, T. A., Y. A. Masyukova, T. A. Safonova, D. P. Petrova, A. L. Vereshagin,

T. V. Minaeva, R. V. Adelshin, T. I. Triboy, I. V. Stonik, N. A. Aizdaitcher, M.

V. Kozlov, Y. V. Likhoshway, and M. A. Grachev. 2005. Conserved motif

CMLD in silicic acid transport proteins of diatoms. Molecular Biology 39:269–

280.

192 Shiono, M. 2000. Three new species in the Thalassiosira trifulta group in late neogene

sediments from the northwest Pacific Ocean. Diatom Research 15:131–148.

Shiono, M. 2001. Two new species of the Thalassiosira trifulta group in late Pliocene

sediments of the northwest Pacific Ocean. Diatom Research 16:83–92.

Shiono, M., and I. Koizumi. 2000. Taxonomy of the Thalassiosira trifulta group in late

Neogene sediments from the northwest Pacific Ocean. Diatom Research 15:355–

382.

Shiono, M., and I. Koizumi. 2001. Phylogenic evolution of the Thalassiosira trifulta

group (Bacillariophyceae) in the northwestern Pacific Ocean. Journal of the

Geological Society of Japan 107:496–514.

Siever, R. 1991. Silica in the oceans: biological-geochemical interplay. Pages 287–295 in

Scientists on Gaia (S. H. Schneider, and P. J. Boston, eds.). MIT Press,

Cambridge.

Spamer, E. E., and E. C. Theriot. 1997. "Stephanodiscus minutulus", "S. minutus", and

similar epithets in taxonomic, ecological, and evolutionary studies of modern and

fossil diatoms (Bacillariophyceae: Thalassiosiraceae): A century and a half of

uncertain taxonomy and nomenclatural hearsay. Proceedings of the Academy of

Natural Sciences of Philadelphia 148:231–272.

Stoermer, E. F., and M. L. Julius. 2003. Centric Diatoms. Pages 559–594 in Freshwater

Algae of North America (J. D. Wehr, and R. G. Sheath, eds.). Academic Press,

Amsterdam.

Stoermer, E. F., H. S. Pankratz, and R. W. Drum. 1964. The fine structure of Mastogloia

grevillei Wm. Smith. Protoplasma 70:1–12.

193 Strelnikova, N. I., and T. V. Lastivka. 1999. The problem of the origin of marine and

freshwater diatoms. Pages 113–123 in Proceedings of the 14th International

Diatom Symposium (S. Mayama, M. Idei, and I. Koizumi, eds.). Koeltz Scientific

Books, Koenigstein.

Strzepek, R. F., and P. J. Harrison. 2004. Photosynthetic architecture differs in coastal

and oceanic diatoms. Nature 431:689–692.

Sullivan, C. W. 1976. Diatom mineralization of silicic acid. I. Si(OH)4 transport

characteristics in Navicula pelliculosa. Journal of Phycology 12:390–396.

Sullivan, C. W. 1977. Diatom mineralization of silicic acid. II. Regulation of Si(OH)4

transport rates during the cell cycle of Navicula pelliculosa. Journal of Phycology

13:86–91.

Swofford, D. L. 2001. PAUP*: Phylogenetic analysis using parsimony (*and other

methods), version 4.0b8. Academic Press.

Tanimura, Y. 1996. Fossil marine plicated Thalassiosira: Taxonomy and an idea on

phylogeny. Diatom Research 11:165–202.

Telford, R. J., V. Vandvik, and H. J. B. Birks. 2006. How many freshwater diatoms are

pH specialists? A response to Pither & Aarssen (2005). Ecology Letters 9:E1–E5.

Thamatrakoln, K., A. J. Alverson, and M. Hildebrand. 2006. Comparative sequence

analysis of diatom silicon transporters: towards a mechanistic model of silicon

transport. Journal of Phycology (in press).

Theriot, E. 1987. Principal component analysis and taxonomic interpretation of

environmentally related variation in silicification in Stephanodiscus

(Bacillariophyceae). British Phycological Journal 22:359–374.

194 Theriot, E., and K. Serieyssol. 1994. Phylogenetic systematics as a guide to

understanding features and potential morphological characters of the centric

diatom family Thalassiosiraceae. Diatom Research 9:429–450.

Theriot, E., E. F. Stoermer, and H. Håkansson. 1987. Taxonomic interpretation of the

rimoportula of freshwater genera in the centric diatom family Thalassiosiraceae.

Diatom Research 2:251–265.

Theriot, E. C. 1992. Clusters, species concepts, and morphological evolution of diatoms.

Systematic Biology 41:141–157.

Theriot, E. C., S. C. Fritz, C. Whitlock, and D. J. Conley. 2006. Late Quaternary rapid

morphological evolution of an endemic diatom in Yellowstone Lake, Wyoming.

Paleobiology 32:38–54.

Tilman, D. 1977. Resource competition between planktonic algae: experimental and

theoretical approach. Ecology 58:338–348.

Tilman, D. 1981. Tests of resource competition theory using 4 species of Lake Michigan

algae. Ecology 62:802–815.

Tilman, D., and S. S. Kilham. 1976. Phosphate and silicate growth and uptake kinetics of

the diatoms Asterionella formosa and Cyclotella meneghiniana in batch and

semicontinuous culture. Journal of Phycology 12:375–383.

Treguer, P., D. M. Nelson, A. J. Vanbennekom, D. J. Demaster, A. Leynaert, and B.

Queguiner. 1995. The silica balance in the world ocean: a reestimate. Science

268:375–379.

195 Tuchman, M. L., E. C. Theriot, and E. F. Stoermer. 1984. Effects of low level salinity

concentrations on the growth of Cyclotella meneghiniana Kütz. (Bacillariophyta).

Archiv Fuer Protistenkunde 128:319–326.

Tusnady, G. E., and I. Simon. 2001. The HMMTOP transmembrane topology prediction

server. Bioinformatics 17:849–850.

Vanlandingham, S. L. 1978. Catalogue of the Fossil and Recent Genera and Species of

Diatoms and their Synonyms. Part VII. Rhoicosphenia through ZygocerosVerlag

Von J. Cramer, Lehre.

Walsby, A. E., and A. Xypolyta. 1977. Form resistance of chitan fibers attached to cells

of Thalassiosira fluviatilis Hustedt. British Phycological Journal 12:215–223.

Weber, C. I. 1970. A new freshwater centric diatom Microsiphona potamos gen. et sp.

nov. Journal of Phycology 6:149–153.

Werner, D. 1977. Introduction with a note on taxonomy. Pages 1–17 in The Biology of

Diatoms (D. Werner, ed.) University of California Press, Berkeley.

Yang, Z., and W. J. Swanson. 2002. Codon substitution models to detect adaptive

evolution that account for heterogeneous selective pressures among site classes.

Mol Biol Evol 19:49–57.

Yang, Z. H. 1997. PAML: a program package for phylogenetic analysis by maximum

likelihood. Computer Applications in the Biosciences 13:555–556.

Yang, Z. H. 1998. Likelihood ratio tests for detecting positive selection and application to

primate lysozyme evolution. Molecular Biology and Evolution 15:568–573.

Yang, Z. H. 2002. Inference of selection from multiple species alignments. Current

Opinion in Genetics & Development 12:688–694.

196 Yang, Z. H., and R. Nielsen. 2002. Codon-substitution models for detecting molecular

adaptation at individual sites along specific lineages. Molecular Biology and

Evolution 19:908–917.

Yang, Z. H., R. Nielsen, N. Goldman, and A. M. K. Pedersen. 2000a. Codon-substitution

models for heterogeneous selection pressure at amino acid sites. Genetics

155:431–449.

Yang, Z. H., W. J. Swanson, and V. D. Vacquier. 2000b. Maximum-likelihood analysis

of molecular adaptation in abalone sperm lysin reveals variable selective

pressures among lineages and sites. Molecular Biology and Evolution 17:1446–

1455.

Zhang, J. Z., R. Nielsen, and Z. H. Yang. 2005. Evaluation of an improved branch-site

likelihood method for detecting positive selection at the molecular level.

Molecular Biology and Evolution 22:2472–2479.

Zwickl, D. J. 2006. GARLI, version 0.94. Program distributed by the author. Section of

Integrative Biology, The University of Texas at Austin.

197 VITA

Andrew James Alverson was born April 23, 1975, in Howell, Michigan, and is the son of

Rebecca E. and W. Edward Alverson. Notably, Andrew shares a birthday with such legendary figures as Shirley Temple, William Shakespeare, and Robert K. ("Bob")

Jansen. Andrew, who saw his first Neil Young concert at age 17, developed an interest in diatoms during an introductory level Phycology course taught by Dr. Mark Luttenton at

Grand Valley State University. He applied this interest to an undergraduate research project with Dr. Gregory Courtney, studying the gut contents of a group of aquatic flies that feeds on diatoms. Upon graduation in 1997, Andrew worked as Head of Maintenance at Metropolitan Title Company in Holland, Michigan. Although Andrew enjoyed the opportunities for fishing and leisure afforded during this time, he felt that he must do something besides fish and watch football. Dr. Courtney accepted a position at Iowa State

University and invited Andrew to continue his insect study as a M.S. student. Shortly before departing for Ames in January 1998, Lisa Collins showed the first signs of romantic interest in Andrew, and as he drove to Ames, he knew they would marry. Lisa and Andrew married in July 1999, and Andrew graduated from ISU with a M.S. degree in

Entomology the following May. Andrew taught Biology and Chemistry at Fowlerville

High School for one year before moving to Austin in August 2001 to attend The

University of Texas, where he diligently studied diatoms and lamented the heat.

Permanent Address: 7990 Colleen Drive, Fowlerville, Michigan, 48836

This dissertation was typed by the author.

198