Copyright 2015 Benjamin R. Gordon

THE INFLUENCE OF LONG-TERM NITROGEN FERTILIZATION ON SYMBIOSIS AND METABOLISM IN THE

LEGUME-RHIZOBIUM MUTUALISM

BY

BENJAMIN R. GORDON

THESIS

Submitted in partial fulfillment of the requirements

for the degree of Master of Science in Plant Biology

in the Graduate College of the

University of Illinois at Urbana-Champaign, 2015

Urbana, Illinois

Master’s Committee:

Associate Professor Tom Jacobs, Chair, Director of Research

Associate Professor Rachel J. Whitaker

Associate Professor Rebecca C. Fuller

ii

ABSTRACT

Understanding how mutualisms evolve in response to a changing environment will be critical for

predicting the long-term impacts of global changes, such as increased nitrogen (N) deposition. Bacterial

mutualists in particular might evolve quickly, thanks to short generation times and the potential for

independent evolution of plasmids through recombination and/or horizontal gene transfer (HGT). In a

previous work using the legume/rhizobia mutualism, we demonstrated that long-term nitrogen

fertilization caused the evolution of less-mutualistic rhizobia. Here, I used two approaches to investigate

the influence of long-term nitrogen fertilization on symbiosis and metabolism in Rhizobium. First, I used

63 of our previously isolated rhizobium strains in comparative phylogenetic and quantitative genetic

analyses to determine the degree to which variation in partner quality is attributable to phylogenetic

relationships among strains versus recent genetic changes in response to N-fertilization. I found

evidence of distinct evolutionary relationships between chromosomal and pSym genes, and broad

similarity between pSym genes. We also find that nifD has a unique evolutionary history that explains

much of the variation in partner quality, and our results implicate MoFe subunit interaction sites in the

evolution of less-mutualistic rhizobia. Second, I determined metabolic fingerprints from our Rhizobium

isolates and integrated them with data on symbiotic quality and previously determined phylogenies,

revealing that N-fertilization significantly contributed to our observed variation in metabolic capacity.

Despite a weak general relationship between metabolism and partner quality, importantly, there appear

to be functionally significant relationships between certain C-substrates and partner quality.

These results provide insight into the evolution of less-cooperative rhizobia in response to long-

term N-fertilization, which significantly contributed to observed variation in partner quality and

metabolic capacity. Nonsynonymous substitutions in nifD appear to be major contributors to variation in

iii

partner quality. Metabolic profile generally correlates poorly with partner quality, as does individual

metabolite utilization. However, there is a relationship between functional significance of certain C-

substrates and partner quality.

iv

DEDICATION

This thesis is dedicated to the loving memory of my mother

Joey, a model of strength and compassion

who passed away from cancer on Oct. 21, 2011.

v

ACKNOWLEDGEMENTS

I would like to thank my research advisor Katy Heath for giving me fast-turnarounds on every

draft, for teaching me to focus on the question, and particularly for fostering my love of science. I thank

my committee members for all their help and suggestions, in particular Rebecca Fuller for entertaining

my microbial perspective in a megafauna world and Rachel Whitaker for lending her ear to my bacterial

bagatelles. I would also like to thank Tom Jacobs for acting as my committee chair at the eleventh hour.

This thesis would not have been possible without the help of many others. I thank Dylan Weese,

Bryn Dentinger, Katy Heath, and Jennifer Lau for previous work which formed the basis of this project. I

thank Christie Klinger for getting me started in the lab and for showing me how to stay organized. I

thank Patricia Burke for teaching me proper microbiological techniques and for always celebrating

discovery with me. I thank my undergraduate assistants, Sarah Carlisle and Taylor Weathers, for their

hard-work, patience, and dedication in the lab. In addition to the aforementioned who also helped

count, I thank Julia Ossler, Jennifer Jones, Kathryn Solórzano Lowell, Matias Fernandez, Mike Grillo,

Justin Podowski, and Leah Caplan for their help in counting seemingly endless root nodules, as nodule

counts range from zero to over 1000 per plant.

The funding for this work was made possible by the Department of Plant Biology, the School of

Integrative Biology, and the Graduate College. This work was supported by NSF DEB-1257938. Support

for this research was also provided by the NSF Long-term Ecological Research Program (DEB 1027253) at

the Kellogg Biological Station.

Finally, and most importantly, I would like to thank my family and friends, whose love and

support drove me to persevere. In particular, I thank my parents Jeff and Joey Gordon who taught me to

have a heart that is full of love, from which I draw strength, willpower, and where I sometimes find

peace.

vi

TABLE OF CONTENTS

CHAPTER 1: GENERAL INTRODUCTION………………………………………………………………………………………………..1

CHAPTER 2: DECOUPLED GENOMIC ELEMENTS SHAPE THE EVOLUTION OF PARTNER QUALITY IN R.

LEGUMINOSARUM BV. TRIFOLII…………………………………………………………………………………………………………..6

CHAPTER 3: RHIZOBIUM LEGUMINOSARUM BV. TRIFOLII STRAINS ADAPT METABOLISM IN THE

INCIPIENT DECLINE OF SYMBIOSIS.……………………………………………………………………………………………..…..….19

FIGURES AND TABLES………………………………………………………………………………………………………………………….30

REFERENCES….…………………………………………………………………………………………………………………………………….41

1

CHAPTER 1: GENERAL INTRODUCTION

Mutualisms, cooperative partnerships between species, are among the most prevalent and

economically important biological interactions, having diverse roles in the maintenance of biodiversity

and nutrient cycling (Stachowicz 2001). Despite the apparent long-term stability of many mutualistic

relationships, models predict that mutualisms should be extremely susceptible to the invasion of less-

beneficial genotypes that give less than they take from their symbiotic partners (West et al. 2007),

leading to mutualism breakdown (Sachs et al. 2004; Holland and DeAngelis 2010). Such mutualism

breakdown might be exacerbated under global change (Six 2009; Kiers et al. 2010), as mutualism serves

to tie the fate of multiple species to a common destiny, and the potential breakdown of their

interactions (from mutualism to antagonism or mutualism abandonment) could lead to an acceleration

of global change and extinction.

The global nitrogen (N) cycle is being drastically altered by fossil fuel combustion and modern

agricultural practices, with N-rich fertilizers adding millions of tons of N globally, surpassing levels of

natural N addition. Nitrogen is an essential component of many constituents of living matter such as

proteins. Unlike other nutrients that are essential to maintaining soil fertility such as potassium or

phosphate, nitrogen is not a significant product of mineral weathering. To be biologically available,

atmospheric nitrogen must be broken apart and recombined with other elements in a process called

nitrogen fixation. This flow is not spontaneous, and is naturally driven by two work intensive processes:

lightning strikes and via enzymes in N-fixing bacteria. Lightning releases heat while passing through the

atmosphere, and this heat converts atmospheric nitrogen to nitrogen oxides. Despite the frequency of

lightning strikes, on the order of 100 times per second, only a small portion of total nitrogen fixation is

achieved in this manner. The vast majority of environmental nitrogen fixation is performed by living

organisms. In the soil, nitrogen fixing bacteria such as rhizobia convert molecular nitrogen to ammonia,

2

which is then converted to ammonium. It is possession of the enzyme nitrogenase that allows rhizobia

to split atmospheric nitrogen, which is the first step in fixing nitrogen.

While N-fixing soil bacteria generate ~ 97% of non-anthropogenically fixed N in terrestrial

ecosystems (Brockwell et al., 1995), N-rich fertilizers add millions of metric tons of N to the earth,

surpassing levels of natural N addition. It is estimated that up to 50% of deposited N is lost to surface

and ground waters, the atmosphere, and uncultivated soils (Vitousek 1997; Vitousek et al. 1997; Smil

2002). Anthropogenic nitrogen (N) deposition is detrimental to natural communities with effects

spanning from decreased biodiversity (Tilman 1987; Clark and Tilman 2008) to coastal dead zones arising

from eutrophication (Diaz and Rosenberg 2008). Moreover these increases in N availability are predicted

to cause the breakdown of a resource mutualism of major ecological and economic importance – that

between legumes (plants in the family Fabaceae) and symbiotic rhizobia (Rhizobiales) (N-fixing soil

bacteria).

N is a key traded resource in the legume-rhizobium mutualistic symbiosis; theory predicts that

excess N will destabilize the legume-rhizobium mutualism and lead to the evolution of less-cooperative

rhizobia through several potential selective mechanisms (Akçay and Simms 2011). First, N-deposition

should weaken positive fitness feedbacks between N-fixation and rhizobium fitness, because plant

fitness (and its feedback on rhizobium fitness) is no longer dependent on symbiotically-fixed N (West et

al. 2002). Second, because legumes are known to limit nodulation when fixed N is available from the

abiotic environment, as previously data has shown empirically (Kiers et al., 2006), N-deposition could

result in a prolonged free-living soil phase for these non-obligate, infectiously acquired bacteria (Sprent

and Sutherland 1987). A shift in the prominent lifestyle phase could, in turn, either select for free-living

traits at the expense of partner quality, or simply allow the accumulation of mutations detrimental to

the mutualism (Sachs et al. 2011). Finally, N deposition might disrupt the stabilizing mechanisms by

which legumes are thought to maintain high partner quality in rhizobia (e.g., sanctions, partner fidelity

3

feedback, partner choice); see discussions in (Oono et al. 2009; Frederickson 2013; Heath and

Stinchcombe 2014). In essence, in an environment where N is no longer a limiting factor, legumes might

cease to select for high-quality strains because additional fixed N will have limited benefits for plants.

Bacterial mutualists are excellent models for studying the evolution and breakdown of

cooperation in changing environments, and for microbial adaptation in general. Bacteria are

experimentally tractable, have short generation times, and harbor relatively small genomes, simplifying

the detection of evolutionary responses (Elena and Lenski 2003). Ecological shifts in taxonomic

composition and/or abundance have been shown to occur within natural bacterial communities in

response to changing environments (Allison and Martiny 2008; Logares et al. 2013; Newton et al. 2013;

Youngblut et al. 2013a). Yet despite ecological evidence/investigations, we still lack fundamental

knowledge on how bacteria found in natural settings evolve in response to environmental changes.

Whether adaptation entails genome-wide sweeps (the “ecotype model”) or site-specific sweeps

(classical models of adaptive evolution) remains controversial (Polz et al. 2013). While adaptive variants

in bacteria can arise from point mutations (Stallforth et al. 2013), bacteria can also evolve rapidly via

recombination by horizontal gene transfer (HGT), a process by which foreign DNA is acquired by an

organism. By incorporating genetic segments, whole genes, or even plasmids into their genome,

bacteria need not rely on point mutation for new alleles (Osborn et al., 2000). As a result, HGT can

accelerate the process of plasmid, ergo genome, evolution (Ochman et al. 2000; Norman et al. 2009). In

addition, environmental factors such as soil type, soil moisture, and temperature have been shown to

increase rates of HGT in closely monitored culture conditions (Richaume 1989), thereby facilitating rapid

evolutionary responses to environmental stress (Jain et al. 2003; Beaber et al. 2004). Studying how

rhizobial mutualists respond to novel environments informs both global change predictions as well as

our nascent understanding of bacterial adaptation.

4

Rhizobial genes necessary for symbiosis with legumes (“symbiosis genes”) typically exist on a

symbiosis plasmid (pSym) separate from the main chromosome (Mazur and Koper 2012). Recent work

by Tian et al. showed that while the rate of recombination was lower than the rate of mutation in

Bradyrhizobium and Sinorhizobium, recombination (compared to point mutation) played a larger role in

shaping their overall diversity (Tian et al., 2012), and recent data suggest a high rate of HGT of rhizobial

plasmids (Epstein et al. 2012). Additionally, in Tian et al., 2012 they were specifically looking at

differences between species while it has previously been shown that differences among strains of the

same species are as great as those between species (Ormeno-Orrillo and Martínez-Romero, 2012), they

used Bradyrhizobium which lack symbiotic plasmids (Cytryn et al., 2008; Hahn and Hennecke, 1987;

Haugland and Verma, 1981), as well as using Sinorhizobium where reduced variability has previously

been observed in their plasmid profiles (López-Guerrero et al., 2012). A review of rhizobia plasmid

population structure showed that the relationship between the pSym and the chromosome could be

either clonal or panmictic, depending on the case (Souza and Eguiarte, 1997). A study of an undisturbed

population of rhizobia indicated that a tight association between the pSym and chromosome is more

relavant to natural populations, and that a panmictic structure would be more relavant to populations

that have been heavily influenced by man or under human selection (Wernegreen et al., 1997). Sachs et

al. (2010) sequenced symbiosis genes for a collection of nodulating and non-nodulating Bradyrhizobium

isolates from Lotus strigosus and found that poorly effective strains were most closely related to strains

isolated from another host genus (Lupinus), suggesting recent HGT of symbiosis genes, despite the lack

of symbiotic plasmids in Bradyrhizobium and the reduced variability observed in Sinorhizobium plasmid

profiles.

In a previous study (Weese et al., 2015), we compared strains of Rhizobium leguminosarum

(hereafter “rhizobia”) isolated from replicate long-term ecological research (LTER) field plots fertilized

with N for 22 years to those from adjacent unfertilized control plots. Consistent with theoretical

5

predictions that N-fertilization should cause the evolution of less cooperative rhizobia, we found that

clover plants (Trifolium spp.) inoculated with rhizobia from N plots (N rhizobia) yielded less plant

biomass, produced fewer leaves, and lower chlorophyll content than plants inoculated with control (C)

rhizobia (Weese et al., 2015). Analysis of a phylogeny formed from the internal transcribed spacer (ITS)

region between the 16S and 23S rRNA genes showed that clades were comprised of both C and N

strains, suggesting that the evolution of reduced cooperation stemmed from microevolutionary changes

such as point mutation and/or HGT rather than an ecological shift in community composition.

My thesis research has focused on understanding the influence of long-term nitrogen

fertilization on the symbiotic and metabolic capabilities of Rhizobium. Utilizing data from a natural

community of bacteria and long-term ecological research plots provided a perfect opportunity to

investigate the ecological (i.e. community shifts) and evolutionary processes responsible for N-driven

mutualism decline in this system. In chapter two, I investigate whether symbiosis marker genes share

similar evolutionary histories, and if they correlate with partner quality. My third chapter investigates

whether there is phylogenetic signal present in metabolism, if metabolic profile correlates with partner

quality, and whether or not strains cluster by treatment. By leveraging model-based, phylogenetic, and

quantitative genetic approaches, I am able to identify major contributors to variation in symbiotic

quality and potential relationships between functional significance of certain carbon substrates and

symbiotic quality.

6

CHAPTER 2: DECOUPLED GENOMIC ELEMENTS SHAPE THE EVOLUTION OF PARTNER QUALITY IN R. LEGUMINOSARUM BV. TRIFOLII1

Abstract

Understanding how mutualisms evolve in response to a changing environment will be critical for

predicting the long-term impacts of global changes, such as increased nitrogen (N) deposition. Bacterial

mutualists in particular might evolve quickly, thanks to short generation times and the potential for

independent evolution of plasmids through recombination and/or horizontal gene transfer (HGT). In a

previous work using the legume/rhizobia mutualism, we demonstrated that long-term nitrogen

fertilization caused the evolution of less-mutualistic rhizobia. Here, we use our 63 previously isolated

rhizobium strains in comparative phylogenetic and quantitative genetic analyses to determine the

degree to which variation in partner quality is attributable to phylogenetic relationships among strains

versus recent genetic changes in response to N-fertilization. We find evidence of distinct evolutionary

relationships between chromosomal and pSym genes, and broad similarity between pSym genes. We

also find that nifD has a unique evolutionary history that explains much of the variation in partner

quality, and our results implicate MoFe subunit interaction sites in the evolution of less-mutualistic

rhizobia. These results provide insight into the mechanisms behind the evolutionary response of rhizobia

to long-term N-fertilization, and we discuss the implications of our results on adaptation in the

mutualism.

1 Gordon, B., Klinger, C., Lau, J., Weese, D., Dettinger, B., Burke, P., Heath, K. (2015) Decoupled Genomic Elements Shape the Evolution of Partner Quality in Rhizobium leguminosarum bv. trifolii Unpublished manuscript, University of Illinois Urbana-Champaign, Urbana, IL.

7

Introduction

Nitrogen (N) is an essential nutrient for all known life (Sterner and Elser 2002), and N availability

is a major determinant of ecology in terrestrial communities (Craine 2009; Grman et al. 2010). The

global N cycle is being drastically altered by fossil fuel combustion and modern agricultural practices,

with N-rich fertilizers adding millions of tons of N globally, surpassing levels of natural N addition

(Canfield et al. 2010). It is estimated that up to 50% of anthropogenic N, which includes reduced N such

as fertilizers and oxidized N such as atmospheric pollution, is lost to water, the atmosphere, and

uncultivated soils (Vitousek 1997; Vitousek et al. 1997; Smil 2002). The ecological effects of this N

deposition are detrimental to natural communities, ranging from decreases in biodiversity (Tilman 1987;

Clark and Tilman 2008) to coastal dead zones arising from eutrophication (Diaz and Rosenberg 2008).

Furthermore, increases in N availability are predicted to cause the breakdown of a resource mutualism

of major ecological and economic importance – that between legumes (plants in the family Fabaceae)

and symbiotic rhizobia (Rhizobiales).

Mutualisms, cooperative partnerships between species, are among the most prevalent and

economically important biological interactions, having diverse roles in the maintenance of biodiversity

and nutrient cycling (Stachowicz 2001). Despite the apparent long-term stability of many mutualistic

relationships, models predict that mutualisms should be extremely susceptible to the invasion of less-

beneficial genotypes that give less than they take from their symbiotic partners (West et al. 2007),

leading to mutualism breakdown (Sachs et al. 2004; Holland and DeAngelis 2010). Such mutualism

breakdown might be exacerbated under global change (Six 2009; Kiers et al. 2010). This prediction arises

from a large body of work indicating that mutualisms are quite context-dependent, making cost-benefit

ratios for partner fitness, and ultimately mutualism stability, highly-contingent upon the surrounding

environmental conditions (Bronstein 1994; Chamberlain et al. 2014). Moreover because partner species

often have very different physiologies, phylogenic origins, and environmental responses, interaction

8

frequencies might also shift in response to a rapidly changing environment and lead to mutualism

breakdown (Sachs and Simms 2006).

Leguminous plants obtain N by pairing with soil bacteria called rhizobia, which are housed in

root nodules and therein fix atmospheric dinitrogen (N2) to ammonium (NH4+) in exchange for plant

photosynthates. Given that N is a key traded resource in this mutualistic symbiosis, theory predicts that

excess N will destabilize the legume-rhizobium mutualism and lead to the evolution of less-cooperative

rhizobia through several potential selective mechanisms (Akçay and Simms 2011). While host control

mechanisms that favor cooperative rhizobium partners have been shown to remain active in high N, at

least in the short-term (Kiers et al. 2006; Regus et al. 2014), it is acknowledged that long-term studies

are needed to understand how the mutualism evolves when exposed to increased N over many

generations. Indeed in a recent study, Weese et al. (2015) demonstrated that populations of Rhizobium

leguminosarum bv. trifolii (Rlt) were less beneficial for their clover host plants after long-term exposure

(22 years) to N fertilizer, and moreover that these N-evolved versus control strains were not

phylogenetically distinct (i.e., were not separate, monophyletic lineages). Nevertheless the evolutionary

processes leading to this phenotypic differentiation were not investigated.

Rhizobia are excellent models for studying the evolution and breakdown of cooperation in

changing environments, and for microbial adaptation in general. Bacteria are experimentally tractable,

have short generation times, and harbor relatively small genomes, simplifying the detection of

evolutionary responses (Elena and Lenski 2003). Ecological shifts in taxonomic composition and/or

abundance have been shown to occur within natural bacterial communities in response to changing

environments (Allison and Martiny 2008; Logares et al. 2013; Newton et al. 2013; Youngblut et al. 2013).

Yet despite recent investigations, we still lack fundamental knowledge on how bacterial populations in

nature evolve in response to environmental changes. While adaptive variants in bacteria can arise from

point mutations (Stallforth et al. 2013), bacteria can also evolve via recombination, or by horizontal gene

9

transfer (HGT) (Gogarten et al. 2002; Ochman & Moran 2001), ‘the non-genealogical transmission of

genetic material from one organism to another’ (Goldenfeld & Woese 2007), potentially facilitating

rapid evolutionary responses to environmental stress (Jain et al. 2003; Beaber et al. 2004). If

recombination is infrequent compared to mutation, then the entire genome should sweep to fixation

along with a newly selected locus (i.e., the “ecotype model” (Koeppel et al. 2008)); by contrast, frequent

recombination in natural bacterial populations could break up the linkage between selected and other

loci (i.e., similar to classical models of eukaryotic adaptive evolution). Existing data on recombination

within and among genomic elements in rhizobia is mixed (Orozco-Mosqueda et al. 2009), and moreover

depends on the phylogenetic scale analyzed. Some studies have found congruent phylogenies consistent

with a general lack of recombination at the species level in contrast to abundant recombination at the

population level (Wernegreen et al. 1997; Kumar et al. 2015), while others have found a high level of

consistency between the chromosome and symbiosis plasmids even at the population scale

(Wernegreen and Riley 1999; Bailly et al. 2007; Epstein et al. 2012). Studies of other microbial species

suggest that recombination is abundant (Reno et al. 2009; Cadillo-Quiroz et al. 2012; Shapiro et al. 2012;

Held et al. 2013), and that the occurrence of genome-wide sweeps is rare (Polz et al. 2013).

In rhizobia the genes necessary for nodulation (nod) and N-fixation (nif and fix) usually reside

either on a symbiosis plasmid (pSym) or as part of a genomic symbiosis island (SI), and in Rlt they have

previously been shown to localize to the pSym (López-Guerrero et al. 2012; Ormeño-Orrillo and

Martínez-Romero 2013). The ability to fix N in rhizobia is derived from nitrogenases, which constitute a

class of complex enzymes that catalyze biological nitrogen fixation by reducing dinitrogen to ammonia.

Rlt possess the commonly studied Mo-dependent nitrogenase which is primarily composed of a

heterotetrameric MoFe-core encoded by nifD and nifK that catalyzes the reduction of nitrogen to

ammonia, and a dinitrogenase reductase Fe-subunit encoded by nifH that provides electrons to

nitrogenase and ATP-derived energy for catalysis (Stacey et al. 1992; O’Carroll and Dos Santos 2011).

10

Previous studies have found evidence that nifH has a different history of horizontal gene transfer from

the nifDK enzyme core and that genetic divergence of nifH and 16S rRNA genes does not commonly

correlate well for microbial species (Gaby and Buckley 2014). Haukka et al. (1998) concluded that the

phylogeny of nifH was generally inconsistent with 16S, but was broadly similar to that of nod genes.

Indeed, phylogenetic trees based on sequences of nod genes are generally not congruent with those

based on 16S rDNA sequences, but the nod trees show some correlation with host plant range

(Wernegreen and Riley 1999).

The extent to which sequence divergence accounts for recent phenotypic change should differ

between marker loci, with strong phylogenetic signal near causative loci; moreover, the degree to which

genome-wide (versus unlinked/site-specific) sweeps contribute to phenotypic divergence should be

reflected by the degree to which all loci in the genome display phylogenetic signal. Here we use the

collection of N-evolved and control R. leguminosarum strains previously isolated from the long-term N-

deposition experiment at the Kellogg Biological Station LTER site (Weese et al. 2015) in phylogenetically-

informed generalized linear mixed-models and comparative phylogenetic analyses to determine the

congruence between strains of Rlt based on pSym genes and the chromosomal internal transcribed

spacer (ITS) region that lies between the 16S and 23S rRNA genes, and to estimate evolutionary

relationships among N-evolved and control strains. We chose nifD and nifH as markers of nitrogen

fixation, and the nodD-A spacer as a marker of nodulation genes. If nif explains the majority of partner

quality variation, this would suggest that a reduction in N-fixation efficiency has driven changes in

partner quality. Alternatively, if nod explains the majority of partner quality variation, this would imply

that a host range shift has driven the decrease in partner quality. Finally, the degree to which these loci

are congruent is illustrative of the relative importance of recombination and HGT on their evolution.

11

Methods

Study system and sequencing: Briefly, our study focuses on a collection of Rhizobium

leguminosaurum bv. trifolii strains (from Weese et al. 2015) previously isolated from a long-term N-

deposition experiment at the Kellogg Biological Station LTER site (Huberty et al. 1998). In previous work,

we performed a single-inoculation experiment in which three species of clover were inoculated with 63

strains isolated from either N-fertilized (28 strains) or control plots (35 strains). Rhizobium strains from

N-fertilized plots were of lower mutualistic partner quality (i.e., they resulted in plants with fewer leaves

and biomass and lower chlorophyll content). See Weese et al. (2015) for full methods on rhizobium

isolations and quantitative trait phenotyping. Each strain was grown in TY (Vincent 1970) liquid media

prior to DNA extraction via the Qiagen Gentra Puregene Yeast/Bacteria kit (Qiagen, Valencia, CA). Three

loci from the pSym-located symbiosis island were amplified via PCR: nifH and nifD, which encode

dinitrogenase reductase and the α-subunit of dinitrogenase (which are the key components of the

nitrogenase enzyme complex), respectively, and the nodD-A region, which spans the intergenic region of

nodD and nodA genes. PCR was performed using the GoTaq PCR protocol (Promega) and successfully

amplified samples were sequenced at the University of Illinois at Urbana-Champaign W. M. Keck Center

for Sanger sequencing. All ambiguous base calls were manually confirmed using Sequencher 5.1 (Gene

Codes Corporation 2012). The ascension number for NifD used in the protein alignment is

YP_009081007 (Rhizobium leguminosaurum bv. trifolii) (Miller et al. 2007). All sequences were aligned

and phylogenies constructed using Seaview 4.5.3 (Gouy et al. 2010) with MUSCLE (Edgar 2004) and

default alignment options. Maximum likelihood trees were constructed in PhyML3 using the GTR model,

without invariable sites, optimized across site rate variation, NNI tree search and a BioNJ starting tree

with optimized topology; branch support was determined using the SH-like approximate likelihood-ratio

test (aLRT).

12

Phylogenetic and mixed model analyses: Unless otherwise noted, all tests of congruence and

recombination were implemented using R (R Core Team 2013). We first used the Shimodaira-Hasegawa

likelihood ratio test (Shimodaira and Hasegawa 1999), SH.test() in package phangorn (Schliep 2011),

which tests the null hypothesis of congruence by comparing the node log likelihood values from two

maximum likelihood phylogenetic reconstructions. Next, we applied a Procrustes Approach to

Cophylogeny (PACo) analysis (Balbuena et al. 2013) implemented in the ape and vegan packages

(Paradis et al. 2004; Oksanen et al. 2013) to provide a measure of topological congruence by assessing

the degree to which one phylogenetic distance matrix predicts another (i.e., is significantly better than

random).

Briefly, using packages adephylo, ape, and phylobase (Paradis et al. 2004; Jombart et al. 2010;

Bolker et al. 2013), we utilized phylo4d() to combine our phylogenies and data set to visually assess the

distribution of partner quality on the phylogenies. Finally to visually assess whether incongruence was

restricted to a particular clade or region of the tree, we formed tanglegrams using the software

Dendroscope 3 (Scornavacca et al. 2011; Huson and Scornavacca 2012). The amount of tangle, or the

degree to which lines cross, is a rough approximation of topological incongruence, and while

tanglegrams ignore support values on the trees, they allow a determination of whether incongruence

segregates by region or clade.

We used phylogenetic generalized linear mixed models (GLMMs) to estimate the proportion of

rhizobium partner quality variation attributable to phylogenetic variation (i.e., linked to the phylogenetic

marker locus) versus the recent N fertilization treatment (i.e., changes elsewhere in the genome in

response to N fertilization). This analysis, implemented using the MCMCglmm package in R (Hadfield

and Nakagawa 2010), jointly estimates the effects of a phylogenetic distance matrix and other variables

on quantitative trait variation (chlorophyll content, leaf number, root mass, shoot mass, and stolon

number), similar to the “animal model” of quantitative genetics (Wilson et al. 2010). We ran separate

13

models that included N treatment and either the ITS phylogeny or the pSym phylogenies. We followed

the recommendations specified by Hadfield (2010, 2012) and Villemereuil et al. (2012) for stipulating the

MCMCglmm priors by dividing the observed phenotypic variance evenly between the effects (the

phylogeny and N treatment), with a shape parameter of 1 (weak, to account for phylogenetic

uncertainty), and using 1,000,000 iterations with a sample interval of 50 and a burn-in of 100,000.

GLMM trace plots were free of pattern, and autocorrelation among samples calculated between

iterations were all less than 0.1, indicating that the model had performed well (Hadfield 2010).

Results

Our analyses of phylogenetic congruence among the four loci indicate recombination among

loci, but suggest that, despite the very different patterns of partner quality variation, the pSym loci are

more congruent with each other compared to the ITS. Of the three pSym loci, nifD/nifH are the least

congruent, with nifH/nodD-A most congruent in PACo, and nifD/nodD-A most congruent via the SH test

(Leigh et al. 2011). The SH test rejected congruence for all pairwise comparisons of loci in our study

(Table 1), consistent with among-locus recombination. The PACo test, however, supported congruence

among the pSym genes, but incongruence between the ITS and each of the three pSym genes (Table 1).

Finally, tanglegrams comparing the ITS to pSym loci show a high degree of entanglement and

comparisons with nifD indicate that incongruence is primarily confined to a pair of clades or regions of

the tree (Figure 1), whereas tanglegrams comparing among pSym genes appear less tangled (Figure 2).

Phylogenetic resolution at the pSym loci may be hampered by limited taxon sampling due to PCR failure

(nifD: 55/63; nifH: 54/63; nodD-A: 37/63), and increased sampling would likely improve phylogenetic

accuracy. We hypothesize that PCR failure was due to gene loss, either of the specific locus, SI, or

plasmid in its entirety.

14

We found that partitioning of partner quality variation depended considerably on the locus

analyzed. The N fertilization treatment explained the majority of rhizobium partner quality for nearly all

models that included the ITS, nifH, and nodD-A phylogenies (Figure 3). These findings would suggest that

loci elsewhere in the genome (i.e., apart from these three marker loci) have changed in response to N

fertilization. Conversely, in our analysis of nifD, phylogeny explained the vast majority of partner quality

variation (Figure 3), and two clades of generally high and low quality stand out when partner quality is

mapped to the tips of the tree (Figure 4), potentially implicating genetic changes within nifD in partner

quality variation.

Using an amino acid alignment of nifD with the reference, we determined that the reference

along with all of the strains from the high quality clade have a methionine at residue 244, whereas all of

the strains from the low quality clade possess a tyrosine. Residue 244 is a MoFe subunit interaction site,

and previous studies have shown that changes to these sites can alter an organism’s ability to fix N

(Brigle et al. 1987). Additionally, strain 116 which stands out as a low quality member of a high quality

subclade, is the only strain in the clade that possesses a mutation at residue 100, which is another MoFe

subunit interaction site.

In contrast to the other partner quality traits (leaf number, stolon number, and plant biomass),

the ITS phylogeny accounted for a large proportion (23.4%) of the variance in chlorophyll content (Figure

3); moreover, including N treatment in the model of chlorophyll content did not improve model fit

(Table 1). These results suggest that causative loci underlying variation in partner chlorophyll content

might be linked to the ITS region, most likely somewhere on the chromosome.

Discussion

Mutualisms are thought to be susceptible to anthropogenic change, although the ecological and

evolutionary processes that drive mutualism decline are not yet understood. Microbes are key players in

15

many important mutualisms, including the N-fixing legume-rhizobium symbiosis; therefore,

understanding the evolution of rhizobium partner quality in natural populations can shed light on the

evolution of bacteria in nature and the maintenance of mutualisms in general. Building on our previous

phenotypic analysis (Weese et al. 2015), which indicated that rhizobia from N-fertilized plots were less-

beneficial symbionts of clover, we compared evolutionary histories of four gene sequences (one on the

chromosome and three pSym genes) with phylogenetic mixed modelling to investigate the evolutionary

process underlying the observed reductions in partner quality. Our results indicate that the ITS region

and the pSym have different evolutionary histories and relationships with rhizobium partner quality, but

that patterns for chlorophyll content differed from other measures of partner quality. Our results differ

from expectations under the ecotype model with zero recombination, but do suggest significant

phylogenetic signal in partner quality, particularly at nifD, but also the ITS for chlorophyll content. We

discuss these main results below.

Our findings indicate both phylogenetic signal and N treatment contribute to among-strain

variation in partner quality, and thus implicate evolutionary changes in response to experimental N

treatments as well as among-lineage variation for mutualism. Given the results of our quantitative trait

modeling and various analyses of congruence/coevolution, however, it appears that the pSym and the

chromosome have different evolutionary histories. By contrast, more congruence among the three

pSym genes suggests that these genes are in partial linkage and that they are more likely to move as a

functional unit. An alternative explanation is that phylogenetic uncertainty is driving incongruence, as

coevolving genes that lack in phylogenetic signal may appear incongruent due to random noise

(Zaneveld et al. 2008). Given previous work showing that rhizobium symbiosis plasmids are self-

transmissible (Bittinger et al. 2000) and that novel symbiosis islands can confer a strong selective

advantage (Parker 2012; Horn et al. 2014), it is possible that recombination or lateral gene transfer of

the pSym across ITS backgrounds has led to the differing signals in the various loci.

16

What do our data suggest about the evolutionary process leading to partner quality decline in

high N plots? Because we find evidence for both phylogenetic signal at our marker loci (particularly at

nifD) and unlinked evolutionary changes in response to N treatment, we can reject a strict one-to-one

relationship between phylogeny and partner quality and thus strict clonal evolution. Nevertheless the

correct interpretation of these findings regarding the evolutionary process that has occurred during the

last 22 years depends somewhat on the age of these rhizobium lineages, compared to the length of the

ecological experiment. On the one hand, if the phylogenetic signal in our dataset predates the LTER

treatments (1988), our findings are consistent with the hypothesis that a combination of lineage sorting

(changes in abundance of pre-existing lineages) and microevolutionary changes has driven the

difference in partner quality between N-evolved and control plots. On the other hand, if these lineages

have diversified since the N treatments were first established, then our findings would suggest that

particular lineages have increased in frequency along with adaptive changes in partner quality. Another

interpretation would be that 22 years is not long enough for complete lineage sorting of recent

substitutions. Given current best estimates of point mutation rates in bacteria (Wielgoss et al. 2011), we

would expect approximately four mutations over the course of our 22 year experiment, yet we find 19

informative sites, which implies that the phylogenetic signal predates the LTER treatment. Nevertheless,

all three scenarios imply a mix of clonality and recombination consistent with other recent studies of

bacteria in nature (Feil et al. 2001; Hanage et al. 2006; Narra and Ochman 2006; Pérez-Losada et al.

2006; Vos and Didelot 2009).

Our results showing that the nifD phylogeny accounts for the vast majority of variation in

partner quality suggests a causal relationship between the locus and measures of partner quality,

particularly given the lack of variation accounted for by phylogeny at nifH or nodD-A. This finding is

consistent with the known importance of nifD in the mutualism (Hennecke 1990; Spaink et al. 1998),

and previous studies of MoFe subunit interaction sites (Brigle et al. 1987), where mutations or deletions

17

lead to drastic changes in the efficiency of N-fixation. Brigle et al. (1987) found that mutations at MoFe

subunit interaction sites can lead to normal, slow, or no growth, with ancillary expression of nif-specific

components was unaffected, and normal levels of both MoFe subunits accumulated in mutants.

While we are unable to rule out the possibility of linked loci elsewhere on the plasmid, it should

be noted that there is divergent signal between nifD versus the structurally adjacent nifH/nodD-A

despite evidence of coevolution, which implies a lack of causally-linked loci near our markers. The

contribution of chromosomal loci is inconclusive based on sequencing the ITS alone. Understanding

which loci determine partner quality and underlie the difference between N-evolved and control

rhizobia will require genome-wide sequencing, which is currently ongoing. Our current evidence implies

that HGT has played a role in contributing to the reduction in partner quality in N fertilized strains.

Specifically, our findings suggest that conjugative transfer of the pSym and subsequent homologous

recombination of nifD might be a mechanism of HGT, consistent with previous findings on the

architecture and nature of rhizobial plasmids (Mercado-Blanco and Toro 1996; Herrera-Cervera et al.

1998; Ding et al. 2013).

Interestingly, the ITS phylogeny accounted for considerable variance in chlorophyll content

(>23.4%), a key measure of partner quality; this was quite different from other measures of partner

quality, where phylogeny never accounted for >3.2% of variation in partner quality. This finding

seemingly contrasts with the fact that chlorophyll content and other measures of partner quality are

correlated among strain means (chlorophyll content ~ shoot mass, r2=0.87, p < 0.0001; chlorophyll

content ~ leaf number, r2=0.80 p < 0.0001). Partner quality traits are quantitative, varying more-or-less

continuously among strains (Weese et al. 2015); therefore, these complex phenotypes are likely to be

underlain by multiple loci, some of which might contribute more to some traits than others. Our findings

might suggest that the ITS region is in partial linkage with loci that are important in determining plant

chlorophyll content, but more work would be needed to test this hypothesis.

18

Conclusion

Here, we have found that nifD explains much of the variation in partner quality, and we

implicate MoFe subunit interaction sites as the primary driver of reduction in partner quality. We find

that the phylogenies of pSym genes are not consistent with the phylogeny of the ITS region, but are

broadly similar to each other. These results suggest HGT of nifD despite the role of recombination in the

pSym. Further sequencing will be required to determine if additional changes throughout the genome

have driven reductions in partner quality, and to elucidate the role of recombination and HGT on whole-

genomes.

Acknowledgements

This work was supported by NSF DEB-1257938. Support for this research was also provided by

the NSF Long-term Ecological Research Program (DEB 1027253) at the Kellogg Biological Station and by

Michigan State University AgBioResearch. The authors thank Charles Roseman for thoughtful comments

on a draft version of the manuscript and Elida Iniguez for help with DNA preparation and sequencing.

Sequence data will be archived in Genbank and our phylogenies in TreeBASE. The authors declare no

conflict of interest.

19

CHAPTER 3: RHIZOBIUM LEGUMINOSARUM BV. TRIFOLII STRAINS ADAPT METABOLISM IN THE INCIPIENT DECLINE OF SYMBIOSIS 2

Abstract

The legume-rhizobium mutualism is one of the most ubiquitous and productive mutualisms in

nature. Nitrogen (N) is a key traded resource in the legume-rhizobium mutualistic symbiosis, and with

the global N cycle being drastically altered by fossil fuel combustion and modern agricultural practices,

theory predicts that excess N will destabilize the legume-rhizobium mutualism and lead to the evolution

of less-cooperative rhizobia through several potential selective mechanisms. A shift in lifestyle could

subsequently select for free-living traits such as increased carbon utilization at the expense of partner

quality. Previously, in work by my lab and collaborators, we demonstrated that populations of

Rhizobium were less beneficial for their clover host plants after long-term exposure (22 years) to N

fertilizer, and moreover that these N-evolved versus control strains were not phylogenetically distinct.

Here, our approach, whose aim was to determine metabolic fingerprints from R. leguminosaurum bv.

trifolii isolates and integrate them with data on symbiotic quality and previously determined

phylogenies, revealed that N-fertilization significantly contributed to our observed variation in metabolic

capacity. Despite a weak general relationship between metabolism and partner quality, importantly,

there appear to be functionally significant relationships between certain C-substrates and partner

quality.

2 Gordon, B., Heath, K. (2015) Rhizobium leguminosarum bv. trifolii Strains Adapt Metabolism in the Incipient Decline of Symbiosis Unpublished manuscript, University of Illinois Urbana-Champaign, Urbana, IL.

20

Introduction

The legume-rhizobium mutualism is widespread and productive in nature. Rhizobia live inside

specialized structures on legume roots called nodules. Once the nodule has formed, rhizobia

differentiate into bacteroids that fix atmospheric nitrogen (N) to ammonia in exchange for

photosynthates from the plant. N-fixing bacteria account for ~97% of all N fixed in natural terrestrial

ecosystems, although they now account for less than 50% globally, due to the combustion of fossil fuels

and the application of N-rich fertilizers. The ecological effects of this N deposition are detrimental to

natural communities, ranging from decreases in biodiversity (Tilman 1987; Clark and Tilman 2008)

Despite the long-term apparent stability (~100mya) of the mutualism (Soltis et al. 1995; Doyle 2011),

numerous models predict the mutualism should be susceptible to invasion by bacterial genotypes who

are less efficient than their neighbors (aka, cheaters), leading to mutualism breakdown. Such mutualism

breakdown might be exacerbated under global change, such as long-term N-fertilization.

Theory predicts that excess N will destabilize the legume-rhizobium mutualism and lead to the

evolution of less cooperative rhizobia through several potential mechanisms (Akçay and Simms 2011),

one of which is context-dependency. Because legumes are known to limit nodulation when fixed N is

available from the abiotic environment, as previous data has shown empirically (Kiers et al. 2006), N-

deposition could result in a prolonged free-living soil phase for these non-obligate, infectiously acquired

bacteria (Sprent and Sutherland 1987). A shift in the prominent lifestyle phase could, in turn, either

select for free-living traits such as increased carbon utilization at the expense of partner quality, or

simply allow the accumulation of mutations detrimental to the mutualism (Sachs et al. 2011). With a

prolonged saprobic phase, we might see a move towards increased specialization among the N strains,

to take advantage of an alternate niche, or alternatively, we might find evidence of a move towards

increased generalization to cope with the compounding stresses associated with a free-living lifestyle.

21

Metabolism of carbon and other energy sources is one of several key factors that contribute to

rhizobial fitness, such as colonization of the rhizosphere and nodule occupancy (Yost et al. 2006;

Ormeño-Orrillo et al. 2008; Ramachandran et al. 2011; Ding et al. 2012). Carbon metabolism traits have

previously been shown to be localized in plasmids in Rhizobium spp. (López-Guerrero et al. 2012; Mazur

and Koper 2012), and plasmids have been shown to confer significant metabolic flexibility to rhizobia,

which is key component of both adapting to varying soil conditions and competing for nodule occupancy

(Wielbo et al. 2010; Mazur et al. 2013). Despite significant research in the area, knowledge of how

accessory genes and the functional relationships between plasmids shape the phenotype and

metabolism of bacteria is poorly understood.

In previous work, Weese et al. (2015) demonstrated that populations of Rhizobium

leguminosarum bv. trifolii (Rlt) were less beneficial for their clover host plants after long-term exposure

(22 years) to N fertilizer, and moreover that these N-evolved versus control strains were not

phylogenetically distinct (i.e., were not separate, monophyletic lineages). Here, we investigated whether

N-evolved rhizobia differ from control rhizobia in terms of their ability to metabolize a range of

substrates (i.e., their “metabolic profile”) while in the free-living state, and more generally whether

partner quality variation among rhizobium strains correlates with metabolic profile variation. By

analyzing individual metabolic profiles at the strain level, and then comparing the results across groups,

we were able to determine whether there is a relationship between metabolism and partner quality, if

free-living or saprobic states were under increased selection among N-fertilized plots as compared with

control plots, if genome size is correlated with metabolic capacity and whether N-fertilization

significantly contributed to our observed variation in metabolic capacity.

Methods

Study system and Rlt Strains.

22

Full methods for rhizobium isolations and quantitative trait phenotyping can be found in Weese

et al. (2015). Utilizing bacteria isolated from N-fertilized and control plots at the KBS LTER site (6

replicate plots per treatment; Huberty et al. 1998) , Weese et al. (2015) used a series of common garden

greenhouse experiments and found that N-fertilized strains phenotypically differentiated from control

strains. N-evolved strains were, on average, less beneficial for host plants, as evidenced by their hosts’

lower leaf number, biomass, stolon number and chlorophyll content. Moreover, these phenotypic

differences were genetically-determined and the result of evolutionary shifts in mean trait values within

a population, rather than shifts in microbial community composition (Weese et al., 2015; Gordon et al.,

unpublished). We leverage the strain mean trait values in the quantitative analyses below.

Metabolic profiling using Biolog.

We compared the metabolic profiles of N-evolved (28 strains) to control (35 strains) Rlt using

the GN2 MicroPlate phenotypic microarray system of Biolog (Hayward, CA). GN2 MicroPlates are 96-well

microtiter plates containing prefilled and dried selected nutrients and biochemicals along with

Tetrazolium violet, a redox dye used to colormetrically indicate the utilization of carbon sources (source

for this – the website or manual?). Thus each GN2 MicroPlate allows the simultaneous screening of

microbial utilization or oxidation of 95 different carbon sources (plus one control well containing only

dye). Previously characterized Rlt isolates were plated on agar tryptone yeast (TY) media and grown until

colonies became visible (48-72 hrs). Individual colonies were then used to inoculate 20 mL of liquid TY

media and incubated shaking at 28°C until turbid (24-48 hrs). Cells from the liquid culture were spun

down and washed twice with of 20 mL of sterile H2O (Heraeus Multifuge X1 Centrifuge Series, TX-400

Swinging Bucket Rotor, 24° C, 5 minutes, and 4700 × g).

23

One GN2 MicroPlate experiment was performed on each strain using recommended Biolog

protocols, with the following modification: To account for the slow growth of Rlt, washed/pelleted cells

derived from liquid culture were diluted with sterile H2O to an optical density (OD550) of 0.1

(approximately 108 cells/ml), and 150 μl suspension of rhizobia was inoculated into each well.

Microplates were then incubated overnight (16-24 hours) at 28° C, and absorbance at 450 and 620 nm

of each well was measured using Sunrise™ microplate reader (Tecan, Männedorf, Switzerland). To

control for any plate-level variation in inoculation density, OD values were corrected using background

color developed in the least fluorescent well (aka, control well, A1).

Sequencing.

See Klinger (2015) for complete methods. Briefly, Genomic DNA was extracted using a modified version

(http://friesen.plantbiology.msu.edu/?p=133) of the MasterPure™ DNA Purification Kit (Epicentre).

Libraries were generated using the Nextera XT DNA Sample Prep Kit (lllumina, San Diego, CA) and

quantified with a Qubit fluorometer 2.0 (Life Technologies, Carlsbad, CA) prior to normalization.

Normalized libraries were sequenced on a HiSeq2000 sequencer (Illumina, San Diego, CA) at the W.M.

Keck Center for Comparative and Functional Genomics at the University of Illinois at Urbana-Champaign.

The A5 pipeline (Tritt et al. 2012) with default parameters was used to assemble genomes. Assembled

genomes were annotated using the RAST server (Aziz et al. 2008) and genes from annotated assemblies

analyzed utilizing the ITEP toolkit (Benedict et al. 2014).

Statistical Analysis.

24

Unless otherwise noted, all calculations and analyses were performed in R (R Core Team 2013).

Using the corrected OD values from the Biolog microplate wells, we analyzed the data in two ways. We

first coded substrate utilization in the Biolog plates as binary data, with wells showing 25% greater

fluorescence than the control well (OD > (1.25 * Control well (A1))) considered positive (1), and anything

less negative (0). Second we analyzed the continuously-varying corrected OD values after normalizing so

that each plate summed to 1.

To estimate the variability of experimental effects including batch, a principal variance

component analysis (PVCA) was used. This approach leverages a principal component analysis (PCA) to

reduce data dimension and maintain variability in the data, followed by a variance components analysis

(VCA) fits a mixed linear model using factors of interest as random effects to estimate and partition the

total variability. The PVCA approach was used to determine which sources of variability (Biological, N-

fertilization, batch or plot of origin) are most prominent in our Biolog data set. Using the eigenvalues

associated with their corresponding eigenvectors as weights, associated variations of all factors are

standardized and the magnitude of each source of variability (including each batch effect) is presented

as a proportion of total variance. To determine which traits were informing the variation attributable to

N-fertilization, one‐way ANOVAs were performed to assess appreciable differences (P < 0.1) in C-

substrate utilization between N-evolved and control strains.

The relationships between the utilization of various C-substrates and the symbiotic quality of

strains was studied using principle component analyses (PCAs), followed by regression analyses.

Hierarchical clustering was then performed on all strains using pvlcust() (Suzuki and Shimodaira 2006)

with absolute corrected distances and complete clustering. Using a phylogeny constructed from whole

genome core SNP data (Klinger et al., unpublished), along with the continuous matrix and symbiotic

quality data from previous, the degree to which utilization of a given C-substrate or symbiotic quality is

25

structured on the tree was determined using a phylogenetic principle component analysis (pPCA)

(Jombart et al. 2010a,b).

Results and Discussion

N-fertilization has Driven Metabolic Variation.

Using a principle variance component analysis (PVCA), we determined that N treatment of origin

(N-fertilized or control) accounted for 13.6% (Table 2) of the overall metabolic variance in our dataset.

The vast majority of the remaining variation was accounted for by inherent among-strain variance in

metabolism, while spatial variation (plot nested within treatment) explained only 3.7% of the variation

in metabolism. We followed this multivariate analysis with univariate ANOVAs to investigate which

individual metabolites might be driving this among-treatment variance. Only six C-substrates were

appreciably influenced (alpha<0.1) by N treatment of origin: Glycogen, Tween 40, Adonitol, i-Erythritol,

Bromosuccinic Acid, and L-Pyroglutamic Acid (Table 3). Thus on the one hand, our multivariate results

suggest a strong role of N-fertilization history in shaping current rhizobium metabolism; on the other

hand, few individual substrates differ strongly between N-fertilized and control strains. These findings

suggest that small differences in metabolism across several substrates must account for our PVCA

results.

All of the aforementioned C-substrates are commonly encountered within the soil, and previous

research has indicated that their utilization plays a role in the legume-rhizobia symbiosis, as well as

being a piece of the fingerprint of disturbed environments. The differential utilization of glycogen has

been previously shown to influence and enhance symbiotic performance (Marroqui et al., 2001). The

ability to utilize adonitol and i-erythritol has been shown to lie on the same locus in S. meliloti (Geddes

and Oresnik, 2012), and to be plasmid localized (Yost et al., 2006), allowing for horizontal

26

transmissibility. Differential utilization of adonitol resulted in reduced ability to compete for nodule

occupancy (Yost et al., 2006), which is a crucial determinant of fitness among symbiotic rhizobia. Natural

homologues of Tween-40 can be found within the soil, and are known to affect biofilm formation, ergo,

bacterial aggregation. The ability to aggregate on legume roots in response to flavonoid excretion by the

plant is critical for infection, and therefore nodule occupancy, so the ability to breakdown Tween-40 and

similar compounds can be key in the establishment of the sybmiosis, and can therefore be critical in

determining rhizobial fitness.

In addition to being a key determinant of rhizbobial fitness, Tween-40 utilization has been

shown to vary in R. psuedoryzae and R. halophytocola in response to environmental disturbance (Zhang

et al., 2011; Bibi et al., 2012). The utilization of bromosuccinic acid, as well as L-Pyroglutamic acid have

both been shown to be highly correlated with microbial diversity in environments with applied

“toxicants” (Tate III and Rogers, 2002). This relationship appears to hold true for anthropogenic N-

fertilization, as we have previously noted that diversity has been greatly affected in the N-fertilized plots

(Weese et al., 2015). It may be worth noting that the utilization of these three C-substrates (Tween-40,

Bromosuccinic acid, L-Pyroglutamic Acid) could be considered primary components of the fingerprint

associated with N-fertilized environments. While partial fingerprints have been determined for certain

particular applied “toxicants,” such as heavy metals, and other environmental disturbances, the

fingerprints of many common disturbances remain as yet undetermined. Compiling a library of

fingerprints resulting from environmental disturbances would allow for quick tests to determine many

aspects of soil status, such as general community composition, health, and function.

Lack of Ecotype Redefinition.

27

While we found that a history of N-fertilization has played a significant role in shaping rhizobium

metabolic variance, treatment does not appear to have driven the formation of distinct metabolic

niches, since we do not observe distinct groups based on treatment or partner quality in the hierarchical

clustering analysis. We found three significant clusters, all of which contain both N-fertilized and

unfertilized strains of mixed partner quality (Figure 5). Moreover, when comparing the total number of

C-substrates utilized among N-fertilized and unfertilized strains, we find that the make-up and

distribution of the two groups is nearly identical (Figure 5, Table 4). These findings are consistent with

the above results suggesting that minor changes in metabolism of several substrates likely underlies the

multivariate response of metabolism to N fertilization. It is perhaps not surprising that N-evolved and

control strains are not entirely differentiated, given the 22-year timescale of N-fertilization in this

experiment and the previous observation that the phylogenies of N and control strains are intermixed

(suggesting frequent recombination between groups; Weese et al., 2015; Gordon et al., unpublished).

Genome size and metabolism are not correlated.

We found that overall genome size has no relationship to the total number C-substrates that

strains were able to utilize (r2= -0.00067; Figure 6). We considered genome size to be the complete set

of genes contained within an individual strains genome, using a binary presence/absence matrix. In

neither case did we find that there was any significant relationship with the number C-substrates utilized

with genome size. The tightly packed nature of the bacterial genome leads to a correspondence

between genome size and proportional variation in amount of genes (Gregory 2005). These variations

create a wide functional range, with smaller genomes that tend to correspond to obligatory mutualists

or pathogens (Koonin and Wolf 2008) and larger bacterial genomes that correspond to species that

28

handle diverse environments and encode complex developmental processes (Konstantinidis and Tiedje

2004). Here, we find a considerable amount of metabolic variation with proportionally little variation in

genome content. This really serves to reiterate how inadequate our current knowledge is concerning

how extrachromosomal elements and the functional relationships between these replicons shape the

phenotype and metabolism of the bacterial cell.

It is clear from the results of the pPCA that the utilization of C-substrates is not purely vertically

transmitted (Figure 7). The utilization of C-substrates runs the full gamut from highly structured on the

phylogeny, to indistinguishable from Brownian motion, all the way to highly unstructured on the

phylogeny. While clonal descent tends to leave strong phylogenetic signal, explaining Brownian motion

on the phylogeny, as well as a complete lack of phylogenetic structure is less clear cut. In either case,

horizontal gene transfer of extrachromosomal elements could create some of our observed patterns. An

alternate explanation is that enough time has passed to erase any record of phylogenetic signal, as a

result of differential linkage and selection pressures over time.

Conclusion

Here, we asked if there is phylogenetic signal present in metabolism, does metabolic profile

correlate with partner quality, and do strains cluster by treatment? By determining metabolic

fingerprints from R. leguminosaurum bv. trifolii isolates and integrating them with data on symbiotic

quality and phylogenies, we revealed that N-fertilization significantly contributed to our observed

variation in metabolic capacity. While N-fertilized and control strains do not generally differ in their

total utilization of C-substrates, metabolic profile generally correlates poorly with partner quality, as

does individual metabolite utilization. Importantly, despite the weak general relationship between C-

29

substrate utilization and partner quality, there is a relationship between functional significance of

certain C-substrates and partner quality.

Acknowledgements

This work was supported by NSF DEB-1257938. Support for this research was also provided by

the NSF Long-term Ecological Research Program (DEB 1027253) at the Kellogg Biological Station and by

Michigan State University AgBioResearch. The authors thank Patricia Burke for providing a microplate

reader, Surangi Punyasena for use of her centrifuge, Stephen Downie for use of his shaking incubator

and Justin Podowski for help inoculating microplates. The authors declare no conflict of interest.

30

FIGURES AND TABLES

Table 1: SH-test/PACo results – Pairwise comparisons of congruence among loci; values closer to 0 represent improved fit. Upper right: SH-test – ln L diff; Lower left: PACo – m^2 = lack-of-fit. Sig. code: *** < 0.001, ** < 0.01, * < 0.05, " " > 0.5

Locus ITS nifD nifH nodD

ITS 7127 *** 4237 *** 4372 ***

nifD 0.256 3939 *** 995 ***

nifH 0.272 0.036 *** 1479 ***

nodD 0.202 0.021 ** 0.007 ***

31

Figure 1: Tanglegram, ITS/nifD – A tanglegram highlighting patterns of phylogenetic variation between the ITS and nifD phylogenies. Strains in blue represent poor partners and strains in red represent superior partners visually identified from trait-mapping. Bootstrap support values >0.70 have been projected onto the trees.

32

Figure 2: Tanglegrams illustrating reduced tangle between pSym phylogenies.

33

Figure 3: Mosaic plot, GLMM attributable variance – The percent of variance attributable to phylogenetic relationships versus N fertilization treatment. Black: variation attributable to phylogeny; Grey: variation attributable to N fertilization; White: residual variance.

34

Figure 4: Bullseye plot, nifD/Partner Quality – (A.) Fan phylogeny of nifD with measure of partner quality mapped to the tips. Traits, from inside out: shoot mass, leaf number, stolon number, and chlorophyll content. Darker colors represent values that are further away from the mean, with blue indicating below the mean (-) and red indicating above the mean (+). Strains from two clades are highlighted: the clade in blue is predominantly below the mean, while the clade in red is predominantly above the mean. (B.) Amino acid alignment of residues 241-250 of nifD, with a MoFe subunit interaction site (residue 244) outlined in red; the same strains are highlighted as above.

35

Table 2: PVCA – The proportion of variance from the first principle component, which accounts for 97.04586 percent of the total variation in the data, attributable to either the plot of origin, nitrogen fertilization treatment, and residual genetic variation.

Effects Weighted Average Proportion Variance

Treatment/Plot 0.037

Treatment 0.136

Residual 0.827

36

Table 3: ANOVA tables for carbon (C) substrates where utilization differed between treatments. C-substrates with a p-value<0.1 are shown; * indicate C-substrates that varied significantly (p-value<0.05) between treatments. Caution should be exercised when interpreting these results, as all values fail to meet significance after FDR adjustment, so they may be considered marginally significant at best.

C-substrate Sum Sq Df Mean Sq F-value P-value

Glycogen*

Treatment 3.00243e-05 1 3.00243e-05 5.31142 0.026199

Plot 2.90383e-05 10 2.90383e-06 0.51370 0.870738

Error 2.37417e-05 42 5.65278e-06

Tween 40

Treatment 8.41298e-06 1 8.41298e-06 3.06867 0.08711

Plot 8.63783e-06 10 8.63783e-07 0.31507 0.97288

Error 1.15146e-04 42 2.74157e-06

Adonitol*

Treatment 0.00005155 1 5.51550e-05 6.36836 0.01549

Plot 0.00010833 10 1.08334e-05 1.25086 0.28867

Error 0.00036375 42 8.66078e-06

i-Erythritol

Treatment 2.52464e-05 1 2.52464e-05 2.85912 0.09827

Plot 1.62471e-04 10 1.62471e-05 1.83996 0.08280

Error 5.19437e-04 42 8.83014e-06

Bromosuccinic Acid

Treatment 0.000132605 1 1.32605e-04 2.95500 0.09298

Plot 0.000529883 10 5.29883e-05 1.18080 0.33057

Error 0.001884746 42 4.48749e-05

L-Pyroglutamic Acid

Treatment 8.01214e-05 1 8.01214e-05 3.25712 0.07829

Plot 4.19281e-04 10 4.19281e-05 1.70447 0.11165

Error 1.03315e-03 42 2.45989e-05

37

Figure 5: Hierarchical clustering of principle components 1 and 2, which account for approximately 50% of metabolic variance; N-fertilized strains are in blue and unfertilized strains in red, with green dots highlighting poor mutualists and orange dots highlighting quality mutualists. Three significant clusters are formed: cluster 1=yellow, cluster 2=green and cluster 3=purple.

38

Table 4: Metabolic Fingerprints in total numbers of substrates utilized, by Treatment-Group and broken down by substrate type.

Total Substrates Sugars Acids Amino Acids Others

N-fertilized 70 31.88 16 18.24 3.88

Unfertilized 67.28571429 30.77142857 15.65714286 17.6 3.257142857

Percent Difference 1.040339703 1.036025998 1.02189781 1.036363636 1.19122807

39

Figure 6: Scatterplot of total gene content in R. leguminosarum bv. trifolii against the number of carbon substrates they could successfully utilize. Overlain in green is the simple linear regression line.

0

2000

4000

6000

8000

10000

12000

14000

0 10 20 30 40 50 60 70 80 90 100

Tota

l Ge

ne

Co

nte

nt

C-substrates utilized

40

Figure 7: Phylogeny derived from whole-genome sequence data (Klinger et al., unpublished) with carbon-substrate utilization mapped to the tips. Each column (1-96) represents a well (A1-H12) from the Biolog GN2 microplate (Hayward, CA)The size of circle represents the degree away from the mean, with white below the mean and black above the mean.

41

REFERENCES Akçay, E., and E. L. Simms. 2011. “Negotiation, sanctions, and context dependency in the legume-

Rhizobium mutualism.” Am. Nat. 178(1):1–14. doi: 10.1086/659997.

Allison, S. D., and J. B. H. Martiny. 2008. “Resistance, resilience, and redundancy in microbial

communities.” Proc. Natl. Acad. Sci. U. S. A. 105(Suppl 1):11512-11519. doi:

10.1073/pnas.0801925105. Epub 2008 Aug 11.

Aziz, R. K., D. Bartels, A. A. Best, M. DeJongh, T. Disz, R. A. Edwards, K. Formsma, et al. 2008. “The RAST

Server: Rapid Annotations Using Subsystems Technology.” BMC Genomics 9:75. doi:

10.1186/1471-2164-9-75.

Bailly, X., I. Olivieri, B. Brunel, J.-C. Cleyet-Marel, and G. Béna. 2007. “Horizontal gene transfer and

homologous recombination drive the evolution of the nitrogen-fixing symbionts of Medicago

species.” J. Bacteriol. 189(14):5223–5236. doi: 10.1128/JB.00105-07. Epub 2007 May 11.

Balbuena, J. A., R. Míguez-Lozano, and I. Blasco-Costa. 2013. “PACo: a novel procrustes application to

cophylogenetic analysis.” PLoS One 8(4):e61048. doi: 10.1371/journal.pone.0061048. Print

2013.

Beaber, J. W., B. Hochhut, and M. K. Waldor. 2004. “SOS response promotes horizontal dissemination of

antibiotic resistance genes.” Nature 427(6969):72–74. doi: 10.1038/nature02241. Epub 2003

Dec 21.

Benedict, M. N., J. R. Henriksen, W. W. Metcalf, R. J. Whitaker, and N. D. Price. 2014. “ITEP: An

Integrated Toolkit for Exploration of Microbial Pan-Genomes.” BMC Genomics 15(8):1– 11.

doi:10.1186/1471-2164-15-8.

Bittinger, M. A., J. A. Gross, J. Widom, J. Clardy, and J. Handelsman. 2000. “Rhizobium etli CE3 carries vir

gene homologs on a self-transmissible plasmid.” Mol. Plant Microbe Interact. 13(9):1019–1021.

doi: 10.1094/MPMI.2000.13.9.1019.

Bolker, B., M. Butler, P. Cowan, D. de Vienne, D. Eddelbuettel, M. Holder, T. Jombart, S. Kembel, F.

Michonneau, B. O’meara, E. Paradis, J. Regetz, and D. Zwickl. 2013. “phylobase: Base package

for phylogenetic structures and comparative data.” R package version 0.6.8. http://CRAN.R-

project.org/package=phylobase.

Brigle, K. E., R. A. Setterquist, D. R. Dean, J. S. Cantwell, M. C. Weiss, and W. E. Newton. 1987. “Site-

directed mutagenesis of the nitrogenase MoFe protein of Azotobacter vinelandii.” Proc. Natl.

Acad. Sci. U. S. A. 84(20):7066–7069.

Brockwell, J., P. J. Bottomley, and J. E. Thies. 1995. “Manipulation of rhizobia microflora for improving

legume productivity and soil fertility: A critical assessment.” Plant Soil 174(1):143–180.

42

Bronstein, J. L. 1994. “Conditional outcomes in mutualistic interactions.” Trends Ecol. Evol. 9(6):214–

217. doi: 10.1016/0169-5347(94)90246-1.

Cadillo-Quiroz, H., X. Didelot, N. L. Held, A. Herrera, A. Darling, M. L. Reno, D. J. Krause, and R. J.

Whitaker. 2012. “Patterns of gene flow define species of thermophilic Archaea.” PLoS Biol.

10(2): e1001265. doi: 10.1371/journal.pbio.1001265. Epub 2012 Feb 21.

Canfield, D. E., A. N. Glazer, and P. G. Falkowski. 2010. “The evolution and future of Earth’s nitrogen

cycle.” Science 330(6001):192–196. doi: 10.1126/science.1186120.

Chamberlain, S., D. P. Vázquez, L. Carvalheiro, E. Elle, and J. C. Vamosi. 2014. “Phylogenetic tree shape

and the structure of mutualistic networks.” J. Ecol. 102(5):1234–1243. doi: 10.1111/1365-

2745.12293.

Clark, C., and D. Tilman. 2008. “Loss of plant species after chronic low-level nitrogen deposition to

prairie grasslands.” Nature 451(7179):712–715. doi: 10.1038/nature06503.

Craine, J. M. 2009. “Resource Strategies of Wild Plants.” Princeton University Press, Princeton, NJ. ISBN:

9780691139128.

Diaz, R. J., and R. Rosenberg. 2008. “Spreading dead zones and consequences for marine ecosystems.”

Science 321(5891):926–929. doi: 10.1126/science.1156401.

Ding, H., C. B. Yip, B. A. Geddes, I. J. Oresnik, and M. F. Hynes. 2012. “Glycerol utilization by Rhizobium

leguminosarum requires an ABC transporter and affects competition for nodulation.”

Microbiology 158(Pt 5):1369–1378. doi: 10.1099/mic.0.057281-0. Epub 2012 Feb 16.

Ding, H., C. B. Yip, and M. F. Hynes. 2013. “Genetic characterization of a novel rhizobial plasmid

conjugation system in Rhizobium leguminosarum bv. viciae strain VF39SM.” J. Bacteriol.

195(2):328–339. doi: 10.1128/JB.01234-12. Epub 2012 Nov 9.

Doyle, J. J. 2011. “Phylogenetic perspectives on the origins of nodulation.” Mol. Plant Microbe Interact.

24(11):1289–1295. doi: 10.1094/MPMI-05-11-0114.

Edgar, R. C. 2004. “MUSCLE: Multiple sequence alignment with high accuracy and high throughput.”

Nucleic Acids Res. 32(5):1792–1797. doi: 10.1093/nar/gkh340.

Elena, S. F., and R. E. Lenski. 2003. “Evolution experiments with microorganisms: the dynamics and

genetic bases of adaptation.” Nat. Rev. Genet. 4(6):457–469. doi: 10.1038/nrg1088.

Epstein, B., A. Branca, J. Mudge, A. K. Bharti, R. Briskine, A. D. Farmer, M. Sugawara, N. D. Young, M. J.

Sadowsky, and P. Tiffin. 2012. “Population genomics of the facultatively mutualistic bacteria

43

Sinorhizobium meliloti and S. medicae.” PLoS Genet. 8(8):e1002868. doi:

10.1371/journal.pgen.1002868. Epub 2012 Aug 2.

Feil, E. J., E. C. Holmes, D. E. Bessen, M. S. Chan, N. P. Day, M. C. Enright, R. Goldstein, D. W. Hood, A.

Kalia, C. E. Moore, J. Zhou, and B. G. Spratt. 2001. “Recombination within natural populations of

pathogenic bacteria: short-term empirical estimates and long-term phylogenetic

consequences.” Proc. Natl. Acad. Sci. U. S. A. 98(1):182–187.

Gaby, J. C., and D. H. Buckley. 2014. “A comprehensive aligned nifH gene database: a multipurpose tool

for studies of nitrogen-fixing bacteria.” Database 2014:bau001. doi: 10.1093/database/bau001.

Gouy, M., S. Guindon, and O. Gascuel. 2010. “SeaView version 4: A multiplatform graphical user

interface for sequence alignment and phylogenetic tree building.” Mol. Biol. Evol. 27(2):221–

224. doi: 10.1093/molbev/msp259. Epub 2009 Oct 23.

Gregory, T. R. 2005. “Synergy between sequence and size in large-scale genomics.” Nat. Rev. Genet.

6(9):699–708. doi: 10.1038/nrg1674.

Grman, E., J. A. Lau, D. R. Schoolmaster, and K. L. Gross. 2010. “Mechanisms contributing to stability in

ecosystem function depend on the environmental context.” Ecol. Lett. 13(11):1400–1410. doi:

10.1111/j.1461-0248.2010.01533.x. Epub 2010 Sep 16.

Hadfield, J. 2010. “MCMC methods for multi-response generalized linear mixed models: the MCMCglmm

R package.” J. Stat. Softw. 33(2):1-22.

Hadfield, J. D., and S. Nakagawa. 2010. “General quantitative genetic methods for comparative biology:

phylogenies, taxonomies and multi-trait models for continuous and categorical characters.” J.

Evol. Biol. 23(3):494–508. doi: 10.1111/j.1420-9101.2009.01915.x. Epub 2010 Jan 7.

Hanage, W. P., C. Fraser, and B. G. Spratt. 2006. “The impact of homologous recombination on the

generation of diversity in bacteria.” J. Theor. Biol. 239(2):210–219. doi:

10.1016/j.jtbi.2005.08.035. Epub 2005 Oct 19.

Held, N. L., A. Herrera, and R. J. Whitaker. 2013. “Reassortment of CRISPR repeat-spacer loci in

Sulfolobus islandicus.” Environ. Microbiol. 15(11):3065–3076. doi: 10.1111/1462-2920.12146.

Epub 2013 May 22.

Hennecke, H. 1990. “Nitrogen fixation genes involved in the Bradyrhizobium japonicum-soybean

symbiosis.” FEBS Lett. 268(2):422–426. doi: 10.1016/0014-5793(90)81297-2.

Herrera-Cervera, J. A., J. M. Sanjuan-Pinilla, J. Olivares, and J. Sanjuan. 1998. “Cloning and identification

of conjugative transfer origins in the Rhizobium meliloti genome.” J. Bacteriol. 180(17):4583–

4590.

44

Holland, J. N., and D. L. DeAngelis. 2010. “A consumer-resource approach to the density-dependent

population dynamics of mutualism.” Ecology 91(5):1286–1295. doi: 10.1890/09-1163.1.

Horn, K., I. M. Parker, W. Malek, S. Rodríguez-Echeverría, and M. A. Parker. 2014. “Disparate origins of

Bradyrhizobium symbionts for invasive populations of Cytisus scoparius (Leguminosae) in North

America.” FEMS Microbiol. Ecol. 89(1):89–98. doi: 10.1111/1574-6941.12335. Epub 2014 Apr 28.

Huberty, L., K. Gross, and C. Miller. 1998. “Effects of nitrogen addition on successional dynamics and

species diversity in Michigan old‐fields.” J. Ecol. 86:794–803. doi: 10.1046/j.1365-2745

Huson, D. H., and C. Scornavacca. 2012. “Dendroscope 3: an interactive tool for rooted phylogenetic

trees and networks.” Syst. Biol. 61(6):1061–1067. doi: 10.1093/sysbio/sys062. Epub 2012 Jul 10.

Jain, R., M. C. Rivera, J. E. Moore, and J. A. Lake. 2003. “Horizontal gene transfer accelerates genome

innovation and evolution.” Mol. Biol. Evol. 20(10):1598–1602. doi: 10.1093/molbev/msg154.

Epub 2003 May 30.

Jombart, T., F. Balloux, and S. Dray. 2010a. “Adephylo: new tools for investigating the phylogenetic

signal in biological traits.” Bioinformatics 26(15):1907–1909. doi:

10.1093/bioinformatics/btq292. Epub 2010 Jun 4.

Jombart, T., S. Pavoine, S. Devillard, and D. Pontier. 2010b. “Putting phylogeny into the analysis of

biological traits: a methodological approach.” J. Theor. Biol. 264(3):693–701. doi:

10.1016/j.jtbi.2010.03.038. Epub 2010 Mar 31.

Kiers, E. T., R. A. Rousseau, and R. F. Denison. 2006. “Measured sanctions: Legume hosts detect

quantitative variation in rhizobium cooperation and punish accordingly.” Evol. Ecol. Res. 8:1077–

1086.

Kiers, T. E., T. M. Palmer, A. R. Ives, J. F. Bruno, and J. L. Bronstein. 2010. “Mutualisms in a changing

world: an evolutionary perspective.” Ecol. Lett. 13(12):1459–1474. doi: 10.1111/j.1461-

0248.2010.01538.x. Epub 2010 Oct 19.

Klinger, C. 2015. “Ecological genomics of mutualism decline in nitrogen-fixing bacteria” (Master's thesis).

Retrieved from https://www.ideals.illinois.edu/bitstream/handle/2142/78425/KLINGER-THESIS-

2015.pdf?sequence=1

Koeppel, A., E. B. Perry, J. Sikorski, D. Krizanc, A. Warner, D. M. Ward, A. P. Rooney, E. Brambilla, N.

Connor, R. M. Ratcliff, E. Nevo, and F. M. Cohan. 2008. “Identifying the fundamental units of

bacterial diversity: a paradigm shift to incorporate ecology into bacterial systematics.” Proc.

Natl. Acad. Sci. U. S. A. 105(7):2504–2509. doi: 10.1073/pnas.0712205105. Epub 2008 Feb 12.

45

Konstantinidis, K. T., and J. M. Tiedje. 2004. “Trends between gene content and genome size in

prokaryotic species with larger genomes.” Proc. Natl. Acad. Sci. U. S. A. 101(9):3160–3165. doi:

10.1073/pnas.0308653100. Epub 2004 Feb 18.

Koonin, E. V., and Y. I. Wolf. 2008. “Genomics of bacteria and archaea: The emerging dynamic view of

the prokaryotic world.” Nucleic Acids Res. 36(21):6688–6719. doi: 10.1093/nar/gkn668. Epub

2008 Oct 23.

Kumar, N., G. Lad, E. Giuntini, M. E. Kaye, P. Udomwong, N. J. Shamsani, J. P. W. Young, and X. Bailly.

2015. “Bacterial genospecies that are not ecologically coherent: population genomics of

Rhizobium leguminosarum.” Open Biol. 5(1):140133. doi: 10.1098/rsob.140133.

Logares, R., E. S. Lindström, S. Langenheder, J. B. Logue, H. Paterson, J. Laybourn-Parry, K. Rengefors, L.

Tranvik, and S. Bertilsson. 2013. “Biogeography of bacterial communities exposed to progressive

long-term environmental change.” ISME J. 7(5):937–948. doi: 10.1038/ismej.2012.168. Epub

2012 Dec 20.

López-Guerrero, M. G., E. Ormeño-Orrillo, J. L. Acosta, A. Mendoza-Vargas, M. A. Rogel, M. A. Ramírez,

M. Rosenblueth, J. Martínez-Romero, and E. Martínez-Romero. 2012. “Rhizobial

extrachromosomal replicon variability, stability and expression in natural niches.” Plasmid

68(3):149–158. doi: 10.1016/j.plasmid.2012.07.002. Epub 2012 Jul 16.

Mazur, A., and P. Koper. 2012. “Rhizobial plasmids — replication, structure and biological role.” Cent.

Eur. J. Biol. 7(4):571–586. doi: 10.2478/s11535-012-0058-8.

Mazur, A., G. Stasiak, J. Wielbo, P. Koper, A. Kubik-Komar, and A. Skorupska. 2013. “Phenotype profiling

of Rhizobium leguminosarum bv. trifolii clover nodule isolates reveal their both versatile and

specialized metabolic capabilities.” Arch. Microbiol. 195(4):255–267. doi: 10.1007/s00203-013-

0874-x. Epub 2013 Feb 16.

Mercado-Blanco, J., and N. Toro. 1996. “Plasmids in Rhizobia: The Role of Nonsymbiotic Plasmids.” Mol.

Plant Microbe Interact. 9(7):535–545. doi: 10.1094/MPMI-9-0535.

Miller, S. H., R. M. Elliot, J. T. Sullivan, and C. W. Ronson. 2007. “Host-specific regulation of symbiotic

nitrogen fixation in Rhizobium leguminosarum biovar trifolii.” Microbiology 153(Pt 9):3184–

3195. doi: 10.1099/mic.0.2007/006924-0.

Narra, H. P., and H. Ochman. 2006. “Of What Use Is Sex to Bacteria?” Curr. Biol. 16(17):R705–R710. doi:

10.1016/j.cub.2006.08.024.

Newton, R. J., S. M. Huse, H. G. Morrison, C. S. Peake, M. L. Sogin, and S. L. McLellan. 2013. “Shifts in the

microbial community composition of Gulf Coast beaches following beach oiling.” PLoS One

8(9):e74265. doi: 10.1371/journal.pone.0074265. eCollection 2013.

46

O’Carroll, I. P., and P. C. Dos Santos. 2011. “Genomic Analysis of Nitrogen Fixation.” Methods Mol. Biol.

766:49-65. doi: 10.1007/978-1-61779-194-9_4.

Oksanen, J., F. Blanchet, R. Kindt, P. Legendre, P. R. Minchin, R. B. O’Hara, G. L. Simpson, P. Solymos, M.

H. H. Stevens, and H. Wagner. 2014. “vegan: Community Ecology Package.” R package version

2.2-0. http://CRAN.R-project.org/package=vegan.

Ormeño-Orrillo, E., M. Rosenblueth, E. Luyten, J. Vanderleyden, and E. Martínez-Romero. 2008.

“Mutations in lipopolysaccharide biosynthetic genes impair maize rhizosphere and root

colonization of Rhizobium tropici CIAT899.” Environ. Microbiol. 10(5):1271–1284. doi:

10.1111/j.1462-2920.2007.01541.x. Epub 2008 Feb 27.

Ormeño-Orrillo, E., and E. Martínez-Romero. 2013. “Phenotypic tests in Rhizobium species description:

An opinion and (a sympatric speciation) hypothesis.” Syst. Appl. Microbiol. 36(3):145–147. doi:

10.1016/j.syapm.2012.11.009. Epub 2013 Feb 13.

Orozco-Mosqueda, M. D. C., J. Altamirano-Hernandez, R. Farias-Rodriguez, E. Valencia-Cantero, and G.

Santoyo. 2009. “Homologous recombination and dynamics of rhizobial genomes.” Res.

Microbiol. 160(10):733–741. doi: 10.1016/j.resmic.2009.09.011. Epub 2009 Oct 8.

Paradis, E., J. Claude, and K. Strimmer. 2004. APE: Analyses of Phylogenetics and Evolution in R language.

Bioinformatics 20(2):289–290. doi: 10.1093/bioinformatics/btg412.

Parker, M. A. 2012. “Legumes select symbiosis island sequence variants in Bradyrhizobium.” Mol. Ecol.

21(7):1769–1778. doi: 10.1111/j.1365-294X.2012.05497.x. Epub 2012 Feb 27.

Pérez-Losada, M., E. B. Browne, A. Madsen, T. Wirth, R. P. Viscidi, and K. A. Crandall. 2006. “Population

genetics of microbial pathogens estimated from multilocus sequence typing (MLST) data.”

Infect. Genet. Evol. 6(2):97–112. doi: 10.1016/j.meegid.2005.02.003. Epub 2005 Mar 24.

Polz, M., E. Alm, and W. Hanage. 2013. “Horizontal gene transfer and the evolution of bacterial and

archaeal population structure.” Trends Genet. 29(3):170–175. doi: 10.1016/j.tig.2012.12.006.

Epub 2013 Jan 15.

R Core Team. 2013. “R: A Language and Environment for Statistical Computing.” R Foundation for

Statistical Computing, Vienna, Austria. http://www.R-project.org/.

Ramachandran, V. K., A. K. East, R. Karunakaran, J. A. Downie, and P. S. Poole. 2011. “Adaptation of

Rhizobium leguminosarum to pea, alfalfa and sugar beet rhizospheres investigated by

comparative transcriptomics.” Genome Biol. 12(10):R106. doi: 10.1186/gb-2011-12-10-r106.

Regus, J. U., K. A. Gano, A. C. Hollowell, and J. L. Sachs. 2014. “Efficiency of partner choice and sanctions

in Lotus is not altered by nitrogen fertilization.” Proc. R. Soc. B 281(1781):20132587. doi:

10.1098/rspb.2013.2587. Epub 2014 Feb 26.

47

Reno, M. L., N. L. Held, C. J. Fields, P. V. Burke, and R. J. Whitaker. 2009. “Biogeography of the Sulfolobus

islandicus pan-genome.” Proc. Natl. Acad. Sci. U. S. A. 106(21):8605–8610. doi:

10.1073/pnas.0808945106. Epub 2009 May 12.

Sachs, J. L., U. G. Mueller, T. P. Wilcox, and J. J. Bull. 2004. “The evolution of cooperation.” Q. Rev. Biol.

79(2):135–160. doi: 10.1086/383541

Sachs, J. L., and E. L. Simms. 2006. “Pathways to mutualism breakdown.” Trends Ecol. Evol. 21(10):585–

92. doi: 10.1016/j.tree.2006.06.018.

Sachs, J. L., A. C. Hollowell, and J. E. Russell. 2011. “Evolutionary Instability of Symbiotic Function in

Bradyrhizobium japonicum.” PLoS One 6(11):e26370. doi: 10.1371/journal.pone.0026370. Epub

2011 Nov 2.

Schliep, K. P. 2011. “phangorn: phylogenetic analysis in R.” Bioinformatics 27(4):592–593. doi:

10.1093/bioinformatics/btq706. Epub 2010 Dec 17.

Scornavacca, C., F. Zickmann, and D. H. Huson. 2011. Tanglegrams for rooted phylogenetic trees and

networks. Bioinformatics 27:i248–56.

Shapiro, B. J., J. Friedman, O. X. Cordero, S. P. Preheim, S. C. Timberlake, G. Szabo, M. F. Polz, and E. J.

Alm. 2012. “Population Genomics of Early Events in the Ecological Differentiation of Bacteria.”

Science 336(6077):48–51. doi: 10.1126/science.1218198.

Shimodaira, H., and M. Hasegawa. 1999. “Multiple comparisons of log-likelihoods with applications to

phylogenetic inference.” Mol. Biol. Evol. 16(8):1114–1116.

Six, D. L. 2009. “Climate change and mutualism.” Nat. Rev. Microbiol. 7:686–686. doi:

10.1038/nrmicro2232.

Smil, V. 2002. “Nitrogen and food production: proteins for human diets.” Ambio 31(2):126–131.

Soltis, D. E., P. S. Soltis, D. R. Morgan, S. M. Swensen, B. C. Mullin, J. M. Dowd, and P. G. Martin. 1995.

“Chloroplast gene sequence data suggest a single origin of the predisposition for symbiotic

nitrogen fixation in angiosperms.” Proc. Natl. Acad. Sci. U. S. A. 92(7):2647–2651. doi:

10.1073/pnas.92.7.2647.

Spaink, H. P., A. Konodorosi, and P. Hooykaas. 1998. “The Rhizobiaceae: Molecular Biology of Model

Plant-Associated Bacteria.” 1st ed. Kluwer Academic Publishers, Dordrecht, Netherlands. doi:

10.1007/978-94-011-5060-6.

48

Sprent, J. I., J. M. Sutherland, S. M. de Faria, M. J. Dilworth, H. D. L. Corby, J. H. Becking, L. A. Materon,

and J. W. Drozd. 1987. “Some Aspects of the Biology of Nitrogen-Fixing Organisms.” Philos.

Trans. R. Soc. Lond. B. Biol. Sci. 317(1184):111–129.

Stacey, G., R. Burris, and H. Evans. 1992. “Biological Nitrogen Fixation.” Chapman and Hall, New York.

ISBN: 978-0-412-02421-4.

Stachowicz, J. J. 2001. “Mutualism, Facilitation, and the Structure of Ecological Communities.”

Bioscience 51(3):235–246. doi: 10.1641/0006-3568(2001)051[0235:MFATSO]2.0.CO;2.

Stallforth, P., D. A. Brock, A. M. Cantley, X. Tian, D. C. Queller, J. E. Strassmann, and J. Clardy. 2013. “A

bacterial symbiont is converted from an inedible producer of beneficial molecules into food by a

single mutation in the gacA gene.” Proc. Natl. Acad. Sci. U. S. A. 110(36):14528–14533. doi:

10.1073/pnas.1308199110.

Sterner, R. W., and J. J. Elser. 2002. “Ecological stoichiometry: The biology of elements from molecules

to the biosphere.” Princeton University Press, Princeton, NJ. ISBN: 9780691074917.

Suzuki, R., and H. Shimodaira. 2006. “Pvclust: an R package for assessing the uncertainty in hierarchical

clustering.” Bioinformatics 22(12):1540–1542. doi: 10.1093/bioinformatics/btl117. Epub 2006

Apr 4.

Tilman, D. 1987. “Secondary succession and the pattern of plant dominance along experimental

nitrogen gradients.” Ecol. Monogr. 57(3):189–214. doi: 10.2307/2937080.

Tritt, A., J. A. Eisen, M. T. Facciotti, and A. E. Darling. 2012. “An Integrated Pipeline for de Novo Assembly

of Microbial Genomes.” PloS One 7(9):e42304. doi: 10.1371/journal.pone.0042304.

Vincent, J. M. 1970. “A Manual for the Practical Study of the Root-Nodule Bacteria.” IBP Handbuch No.

15 des International Biology Program, London. XI u. 164 S., 10 Abb., 17 Tab., 7 Taf. Oxford-

Edinburgh.

Vitousek, P. M., H. A. Mooney, J. Lubchenco, and J. M. Melillo. 1997. “Human Domination of Earth's

Ecosystems.” Science 277(5325):494–499. doi: 10.1126/science.277.5325.494.

Vitousek, P. M., J. D. Aber, R. W. Howarth, G. E. Likens, P. a. Matson, D. W. Schindler, W. H. Schlesinger,

and D. G. Tilman. 1997. “Human Alteration of the Global Nitrogen Cycle: Sources and

Consequences.” Ecol. Appl. 7(3):737–750. doi: 10.1890/1051-

0761(1997)007[0737:HAOTGN]2.0.CO;2.

Vos, M., and X. Didelot. 2009. “A comparison of homologous recombination rates in bacteria and

archaea.” ISME J. 3(2):199–208. doi: 10.1038/ismej.2008.93. Epub 2008 Oct 2.

49

Weese, D. J., K. D. Heath, B. T. M. Dentinger, and J. A. Lau. 2015. “Long-term nitrogen addition causes

the evolution of less-cooperative mutualists.” Evolution, 69(3):631–642. doi:

10.1111/evo.12594.

Wernegreen, J. J., E. E. Harding, and M. A. Riley. 1997. “Rhizobium gone native: unexpected plasmid

stability of indigenous Rhizobium leguminosarum.” Proc. Natl. Acad. Sci. U. S. A. 94(10):5483–

5488.

Wernegreen, J. J., and M. A. Riley. 1999. “Comparison of the evolutionary dynamics of symbiotic and

housekeeping loci: a case for the genetic coherence of rhizobial lineages.” Mol. Biol. Evol.

16(1):98–113.

West, S. A., A. S. Griffin, and A. Gardner. 2007. “Evolutionary explanations for cooperation.” Curr. Biol.

17(16):R661–R672. doi: 10.1016/j.cub.2007.06.004.

Wielbo, J., M. Marek-Kozaczuk, A. Mazur, A. Kubik-Komar, and A. Skorupska. 2010. “Genetic and

metabolic divergence within a Rhizobium leguminosarum bv. trifolii population recovered from

clover nodules.” Appl. Environ. Microbiol. 76(14):4593–4600. doi: 10.1128/AEM.00667-10. Epub

2010 May 14.

Wielgoss, S., J. E. Barrick, O. Tenaillon, S. Cruveiller, B. Chane-Woon-Ming, C. Médigue, R. E. Lenski, D.

Schneider, B. J. Andrews. 2011. “Mutation rate inferred from synonymous substitutions in a

long-term evolution experiment with Escherichia coli.” G3 1(3):183–186. doi:

10.1534/g3.111.000406.

Wilson, A. J., D. Réale, M. N. Clements, M. M. Morrissey, E. Postma, C. A. Walling, L. E. B. Kruuk, and D.

H. Nussey. 2010. “An ecologist’s guide to the animal model.” J. Anim. Ecol. 79(1):13–26. doi:

10.1111/j.1365-2656.2009.01639.x.

Yost, C. K., A. M. Rath, T. C. Noel, and M. F. Hynes. 2006. “Characterization of genes involved in

erythritol catabolism in Rhizobium leguminosarum bv. viciae.” Microbiology 152(Pt 7):2061–

2074. doi: 10.1099/mic.0.28938-0.

Youngblut, N. D., M. Dell’aringa, and R. J. Whitaker. 2013. “Differentiation between sediment and

hypolimnion methanogen communities in humic lakes.” Environ. Microbiol. 16(5):1411–1423.

doi: 10.1111/1462-2920.12330. Epub 2013 Dec 10.

Zaneveld, J. R., D. R. Nemergut, and R. Knight. 2008. “Are all horizontal gene transfers created equal?

Prospects for mechanism-based studies of HGT patterns.” Microbiology 154(Pt 1):1–15. doi:

10.1099/mic.0.2007/011833-0.