FACTORS AFFECTING RESOURCE ALLOCATION IN THE LEGUME-RHIZOBIA  

SYMBIOSIS 

By 

Colleen Anne Friel 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

A DISSERTATION 

Submitted to 

Michigan State University 

in partial fulfillment of the requirements 

for the degree of  

Plant Biology—Doctor of Philosophy 

2018 

 

 

 
 
 

 

FACTORS AFFECTING RESOURCE ALLOCATION IN THE LEGUME-RHIZOBIA  

ABSTRACT 

SYMBIOSIS 

By 

Colleen Anne Friel 

 

The legume-rhizobia symbiosis is an interaction in which nitrogen fixing bacteria called 

rhizobia colonize plant roots and supply fixed nitrogen to the plant in exchange for 

photosynthetically fixed carbon. This interaction has global impacts on the nitrogen cycle and 

offers an alternative to environmentally damaging synthetic nitrogen fertilizers. Resource 

allocation between plants and rhizobia is shaped by several factors, including the abiotic 

environment and various levels of host-symbiont specificity. The first two chapters of this 

dissertation explore the effects of the abiotic environment. First, a theoretical ecological model 

was adapted to the legume-rhizobia symbiosis and parameterized with a series of detailed 

measurements of plant and nodule biomass, carbon and nitrogen content, and plant 

photosynthesis. These results were compared to model predictions, which illustrated that the 

model assumption of fair trade was invalid, that plants have more bargaining power than 

rhizobia, and that plant bargaining power is highest when soil nitrogen is lowest. In the second 

chapter, the effects of factorial manipulation of soil nitrogen and light availability on resource 

trade between legumes and rhizobia was assessed. The results revealed that plants adjusted their 

resource acquisition strategy to take up the most limiting resource, and that both nitrogen and 

light affect allocation to rhizobia, but not their symbiotic effectiveness per unit of resource 

received. Finally, the third project assessed the level of specificity between host plant and 

rhizobial symbiont by comparing the effectiveness of rhizobia isolated from the same plant 

species to that of rhizobia isolated from different plant species. The results were contrary to 

 

ecological theory predicting positive plant-soil feedbacks between legumes and rhizobia. The 

effects of environmental context dependence and plant-rhizobia specificity are vital for 

understanding the role of rhizobia in natural and agricultural ecosystems as well as the future 

development of effective rhizobial crop inoculants.  

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

ACKNOWLEDGMENTS 

 

 

I would like to thank my committee members Maren Friesen, Sheng Yang He, Yair 

Shachar-Hill, and Shinhan Shiu for their guidance and encouragement. I would also like to thank 

past and present members of the Friesen lab, including Jeff Norman, Chandra Jack, Alan 

Bowsher, Carissa Zielinski, Emily McLachlan, Michelle Afkhami, Prateek Shetty, and Eleanor 

Siler, in addition to Shachar-Hill lab members Teresa Clark and Shawna Rowe. I would also like 

to thank all the members of the Centre for Rhizobium Studies at Murdoch University in Perth, 

Western Australia, especially Talitha Rogers and Jason Terpolilli, for their hospitality and 

support. I thank my funding sources, including the MSU University Distinguished Fellowship, 

BEACON Center for Evolution in Action, the Plant Biotechnology for Heath and Sustainability 

Training Program, the Michigan State Plant Science Fellowship Award of Excellence, and 

Michigan State College of Natural Science Dissertation Continuation Fellowships..  

 

 

 

 

 

 

 

 

 

 

iv 
 

TABLE OF CONTENTS 

 

LIST OF TABLES……………………………………………...……………………………..…vii 

LIST OF FIGURES…………………………………………………………………………........ix 

KEY TO ABBREVIATIONS…………………………………………………………………….xi 

 

CHAPTER 1………………………………………………………………………………………1 
LITERATURE REVIEW…………………………………………………………………………1 
Introduction……………………………………………………………………………......1
Legumes: the plant side………………………………………………………………...…2 
Rhizobia: the bacteria side………………………………………………………………...3 
Mechanisms of the legume-rhizobia interaction…………………………………………..4
Ecological and evolutionary theory……………………………………………………….8 
The abiotic environment and the legume-rhizobia interaction…………………………..12 
The biotic environment and the legume-rhizobia interaction……………………………16 
Conclusions………………………………………………………………………………19 
REFERENCES.………………………………………………………………………………….21 

 

CHAPTER 2……………………………………………………………………………………..38 
UNFAIR TRADE UNDERGROUND REVEALED BY INTEGRATING DATA WITH NASH 
BARGAINING MODELS……………………………………………………………………….38 
Preface……………………………………………………………………………………38 
Abstract…………………………………………………………………………………..39 
Introduction………………………………………………………………………………40 
Materials and Methods…………………………………………………………………...44 
Results……………………………………………………………………………………52 
Discussion………………………………………………………………………………..61 
APPENDIX………………………………………………………………………………………66 
REFERENCES…………………………………………………………………………………..76 

CHAPTER 3……………………………………………………………………………………..81 
LEGUMES MODULATE ALLOCATION TO RHIZOBIAL NITROGEN FIXATION IN  
RESPONSE TO FACTORIAL RESOURCE MANIPULATION………………………………81 
Abstract…………………………………………………………………………………..81 
Introduction………………………………………………………………………………82 
Materials and Methods…………………………………………………………………...86 
Results……………………………………………………………………………………89 
Discussion………………………………………………………………………………..98 
APPENDIX……………………………………………………………………………………..104 
REFERENCES…..……………………………………………………………………………..106 

CHAPTER 4…………………………………………………………………………………....111 

v 
 

HOST SPECIFICITY IN THE TRIFOLIUM-RHIZOBIUM SYMBIOSIS…………………...111 
Abstract…………………………………………………………………………………111 
 
Introduction……………………………………………………………………………..112 
 
Materials and Methods………………………………………………………………….116 
 
 
Results…………………………………………………………………………………..122 
 
Discussion………………………………………………………………………………139 
APPENDIX……..………………………………………………………………………………147 
REFERENCES...……………………………………………………………………………….157 

CHAPTER 5……………………………………………………………………………………163 
CONCLUSIONS AND FUTURE DIRECTIONS.…………………………………………….163 
REFERENCES………………………………………………………………………………....170 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

vi 
 

LIST OF TABLES 

 

Table 2.1 | Model input parameters and measured values………………...…………………70 

Table 2.2 | Model predictions and corresponding values………………...…………………..71 

Table 2.3 | Plant carbon and nitrogen elemental compositions………..…………………….74 

Table 2.4 | Nodule carbon and nitrogen elemental compositions………..…………………..75 

Table 3.1 | ANOVAs summarizing the effects of light and nitrogen on nodulation………..90 

Table 3.2 | ANOVAs summarizing the effects of light and nitrogen on nodulation scaled by  
root biomass…………………………………………………………………………………….92 

Table 3.3 | ANCOVA summarizing the effects of light and nitrogen on the relationship  
between nodule biomass and shoot biomass gain…………………………………………….93 

Table 3.4 | ANOVA summarizing the effects of light, nitrogen, and rhizobia on root:shoot  
ratio……………………………………………………………………………………………...95 

Table 3.5 | ANOVAs summarizing the effects of light, nitrogen, and rhizobia on shoot and  
root biomass……………………………………………………………………………………..97 

Table 4.1 | ANOVAs summarizing the effect of isolation host on rhizobial PDI………….123 

Table 4.2 | ANOVAs summarizing the effects of plant PDI measures on plant responses to  

rhizobia……………………………………………………………………………………….125 

Table 4.3 | ANOVAs summarizing the effects of rhizobial PDI measures on plant responses  
to rhizobia……………………………………………………………………………………..126 

Table 4.4 | ANOVAs summarizing the effects of interaction type and test species on plant  
performance…………………………………………………………………………………...128 

Table 4.5 | ANOVAs summarizing the effects of interaction type and test species on  
nodulation……………………………………………………………………………………..130 

Table 4.6 | ANOVAs summarizing the effects of isolation host and test species on  
nodulation……………………………………………………………………………………..132 

Table 4 .7 | Summary of number of test species for which each isolation host was the worst  
isolation host or best isolation host…………………………………………………………..133 

vii 

 

Table 4.8 | ANCOVA summarizing the effects of nodule biomass and test species on shoot  

biomass gain…………………………………………………………………………………..138 

Table 4.9 | Percentage of inoculated individuals of each plant species that flowered…….155 

Table 4.10 | Correlation between nodule biomass and scaled shoot biomass for each test  
species…………………………………………………………………………………………..156 

Table 4.11| ANOVA summarizing the effects of inoculum and test species on plant  
performance……………………………………………………………………………………156 
 

 
 

 

 

 
 
 
 
 
 

 
 

 

 
 

 
 
 

 

 

 

 

viii 

 

LIST OF FIGURES 

 

Figure 2.1 | Pictorial representation of the legume-rhizobia nutrient exchange model……43 

Figure 2.2 | Both partners benefit more from trade at lower nitrogen availabilities………55 

Figure 2.3 | The effect of rhizobial inoculation on plant carbon and nitrogen budgets as  
functions of nitrogen availability………………………………………………………………56 

Figure 2.4 | Plant carbon allocation to nutrient uptake……………………………………...57 

Figure 2.5 | Model fit to experimentally estimated data is improved by allowing asymmetric  
bargaining power……………………………………………………………………………….60 

Figure 2.6 | Plant bargaining power as a function of nitrogen availability…………………61 

Figure 2.7 | Whole-plant photosynthesis chambers were used to measure carbon uptake  
rates……………………………………………………………………………………………..67 

Figure 2.8 | Regression analyses used to calculate model parameters……...……………….68 

Figure 2.9 | Plant growth is near linear between 4 and 6 weeks……………………………..69 

Figure 3.1 | Effects of light and nitrogen on total nodulation………………………………..90 

Figure 3.2 | Effects of light and nitrogen on investment in trade versus direct uptake and  
return on investment in trade………………………………………………………………….92 

Figure 3.3 | Effects of light, nitrogen, and rhizobia on root:shoot ratio…………………….95 

Figure 3.4 | Effects of light, nitrogen, and rhizobia on shoot and root biomass………….…97 

Figure 3.5 | Nodule efficiency in response to soil nitrogen………………………………….105 

Figure 4.1 | Specialization of plants and rhizobia as measured by Paired Difference Index  
(PDI)……………………………………………………………………………………………123 

Figure 4.2 | Effect of plant specialization on response to rhizobia…………………………125 

Figure 4.3 | Effect of rhizobial specialization on plant responses………………………….126 

Figure 4.4 | Plant performance and nodulation with conspecific versus heterospecific  
rhizobia. ………………………………………………………………………..………..…….130 

ix 
 

Figure 4.5 | Effects of isolation host on nodule number across all test species…………….132 

Figure 4.6 | Effects of isolation host on nodule biomass across all test species……………134 

Figure 4.7 | Response to rhizobia for each plant species averaged across all rhizobia  
treatments……………………………………………………………………………….……..136 

Figure 4.8 | Correlation coefficient and effect size for efficiency across all test species…..138 

Figure 4.9 | Scatterplots of mean plant responses to rhizobia versus plant PDI measures 
…………………………………………………………………………………………………..148  

Figure 4.10 | Scatterplots of maximum plant responses to rhizobia versus plant PDI  
measures………………………………………………………………………………………..149 

Figure 4.11 | Scatterplots of mean plant responses to rhizobia versus rhizobial PDI  
measures………………………………………………………………………………………..150 

Figure 4.12 | Scatterplots of maximum plant responses to rhizobia versus rhizobial PDI 
measures…………………………………………………………………………………….….151 

Figure 4.13 | Response to rhizobia versus commonness in the field………………………..152 

Figure 4.14 | Mean scaled shoot biomass and flower number for each test species  
inoculated with conspecific or heterospecific rhizobia……………………………………...153 

Figure 4.15 | Mean scaled shoot biomass for each plant species inoculated with rhizobia  
from each isolation host……………………………………………………………………….154 

Figure 4.16 | Mean flower number for each plant species inoculated with rhizobia from  
each isolation host……………………………………………………………………………..155 

 
 
 
  
 

 
 
 
 
 
 

 

x 
 

KEY TO ABBREVIATIONS 

 

ANCOVA 

Analysis of covariance 

ANOVA 

Analysis of variance 

AON 

EPS 

 

 

Autoregulation of nodulation 

exopolysaccharide 

GS-GOGAT  glutamate synthase-glutamine oxoglutarate aminotransferase 

lhk1 

 

Lotus histidine kinase 1 

LRR-RLK 

leucine-rich receptor-like protein kinase 

Nfr5 

NOP 

PBU 

PHB 

R/FR 

TCA 

 

 

 

 

 

 

Nod factor receptor 5 

nodulation outer proteins 

peribacteroid unit 

poly-β-hydroxybutyrate 

red/far red 

tricarboxylic acid 

 

 

 

 

 

 

 

xi 
 

CHAPTER 1 

 

LITERATURE REVIEW 

 

Introduction 

Nutritional mutualisms, or interactions in which participants trade nutrients for mutual 

benefit, are ancient and widespread. Mutualisms have led to innovations such as eukaryotic cells 

and the colonization of land by plants (Bronstein, 2015). Nutritional plant-microbe mutualisms 

play an important role in global nutrient cycling. For example, mycorrhizal fungi colonize plant 

roots and use their hyphal networks to supply phosphorus and other minerals in exchange for 

roughly five billion tons of photosynthetically fixed carbon globally each year (Bago et al., 

2000). Similarly, rhizobia, soil bacteria that colonize plant roots, fix more than forty million tons 

of atmospheric nitrogen and exchange it for host carbon (Udvardi & Poole, 2013). There has 

been a large body of research exploring why mutualisms have persisted over evolutionary time 

despite the large benefits mutualists stand to gain if they defect (Sachs et al., 2004; Ghoul et al., 

2014; Bronstein, 2015). However, the theoretical threat of cheaters has rarely been backed up by 

empirical evidence of cheaters prospering in contemporary mutualisms (Jones et al., 2015), and 

phylogenetic analysis demonstrates that the evolution of parasites from within mutualistic clades 

occurs rarely (Sachs et al., 2011). In wild populations, parasites of mutualisms often come from 

outside the focal interaction, as illustrated by non-fixing rhizobia that originated from non-

symbiotic lineages (Sachs et al., 2010a), and mutualistic symbioses are highly stable over 

evolutionary time (Werner et al., 2015). Together, these observations suggest the operation of 

mechanisms, potentially operating at multiple scales, that result in the evolutionary robustness of 

1 
 

mutualisms as a whole. Further complicating our understanding of mutualisms is the observation 

that interactions may be mutualistic under some contexts but parasitic under others; this context 

dependence is often linked to the availability of external resources (Johnson et al., 1997; 

Chamberlain et al., 2014). Plant-microbe nutritional mutualisms are amenable systems for 

addressing these issues because resource fluxes can be tracked in both directions and there are 

existing models at multiple biological scales in addition to relevant empirical datasets (Clark et 

al., 2017). Critical empirical work addressing these issues is lacking (Friesen & Heath, 2013), 

and larger questions about the regulation of mutualism and its evolutionary dynamics remain. 

 

Legumes: the plant side 

This dissertation will focus on the interactions between legumes and nitrogen-fixing 

bacteria known as rhizobia. Legumes belong to the family Fabaceae (Leguminosae), which 

consists of approximately 19,000 species across 750 genera, making it the third largest family of 

flowering plants (Andrews & Andrews, 2017; Sprent et al., 2017). Fabaceae had traditionally 

been divided into three subfamilies (Caesalpinioideae, Mimosoideae and Papilionoideae), but 

recent taxonomic work has suggested that it is more appropriate to divide Fabaceae into six 

subfamilies (Duparquetioideae, Cercidoideae, Detarioideae, Dialioideae, Caesalpinioideae [this 

now includes the mimosoid clade], and Papilionoideae (Azani et al., 2017)). The 

Duparquetioideae, Cercidoideae, Detarioideae and Dialioideae do not associate with rhizobia and 

thus will not be discussed further (Sprent et al., 2017). Caesalpinioideae, which now includes the 

mimosoid clade, consists of woody and herbaceous plants mainly found in the tropics. Some 

members nodulate, particularly in the mimosoid clade, but not all do (Azani et al., 2017). The 

plant species used in this work all come from Papilionoideae, a cosmopolitan subfamily that 

2 
 

includes approximately 14,000 species. The only non-legumes that are able to nodulate with 

rhizobia are in the genus Parasponia, part of the family Cannabaceae (Trinick, 1973). Legumes 

are estimated to have evolved nodulation 65 million years ago, but Parasponia seems to have 

evolved nodulation much more recently (Op den Camp et al., 2011). 

 

Rhizobia: the bacteria side 

Rhizobia is the generally accepted term for a large, phylogenetically diverse group of α- 

and β- proteobacteria that are able to form nitrogen fixing symbioses with legumes (Mus et al., 

2016). Rhizobia of the α-proteobacteria (α-rhizobia) are all members of Rhizobiales, while 

rhizobia of the β-proteobacteria (β-rhizobia) belong to Burkholderiales (Shamseldin et al., 2017). 

The same genus or species of bacteria may contain both nodulating and non-nodulating strains, 

and non-nondulating strains may vastly outnumber nodulating strains (Laguerre et al., 1993; 

VanInsberghe et al., 2015; Hollowell et al., 2016). These non-nodulating strains may reduce the 

benefits plants gain from nodulating strains because of competition on the root surface (Gano-

Cohen et al., 2016). Rhizobia tend to have large genomes (ranging from 5.4-9.2 Mb) that are 

highly plastic, using large numbers of transport and regulatory genes to be able to survive in 

highly variable soil environments (MacLean et al., 2007). The genomes of rhizobia are very 

disparate, and analysis of gene content suggests that there is not a single shared genetic program 

that allows for plant-microbe symbioses: genes that are involved in symbiosis frequently have 

close homologs in non-symbiotic relatives, or they are very narrowly distributed within the 

rhizobial phylogeny (Amadou et al., 2008; Masson-Boivin et al., 2009). All rhizobia employ the 

molybdenum-dependent nitrogenase, which requires very low oxygen levels and has a high 

3 
 

energy requirement of 16 moles of ATP per mole of nitrogen fixed, conditions that plants meet 

with high carbon supply and leghemoglobin (Masson-Boivin et al., 2009). 

The largest genus in terms of number of strains isolated thus far is Bradyrhizobium, 

which nodulates the most diverse set of legume genera as well as Parasponia (Parker, 2015). 

Because Bradyrhizobium nodulates such a broad host range, and also nodulate most of the early 

diverging nodulating legumes, it has been suggested that Bradyrhizobium was the original 

legume symbiont (Parker, 2015; Sprent et al., 2017). Rhizobium and Ensifer (Sinorhizobium) are 

also common α-rhizobia that are found in soil in a wide variety of environments across many 

continents (Peix et al., 2015; Sprent et al., 2017). 

 

The mechanisms of legume-rhizobia interaction 

An important defining characteristic of different types of legumes is whether they form 

determinate or indeterminate nodules. Determinate nodules are formed by tropical legumes, such 

as Phaseolus vulgaris (common bean) and Glycine max (soybean). Determinate nodules have a 

transient meristem and grow for a certain time period (usually a few weeks) and then stop, 

resulting in round nodules (Udvardi & Poole, 2013). Inside a determinate nodule, rhizobia 

differentiate into a homogenous group of bacteroids that are similar to free living rhizobia. 

Typically several bacteroids are found in each peribacteroid unit (PBU), which is a group of 

bacteroids surrounded by the plant membrane (Denison, 2000). Bacteroids in determinate 

nodules can accumulate poly-β-hydroxybutyrate (PHB), why may allow them to hoard resources 

from their host (Lodwig & Poole, 2003). Determinate nodules convert fixed NH3 to ureides for 

export through the xylem to the rest of the plant (Baral et al., 2016).  

4 
 

In contrast, the legumes used in these studies all form indeterminate nodules. In 

indeterminate nodules, the nodule maintains an active meristem and continues to grow 

throughout its life cycle (Udvardi & Poole, 2013). This results in a nodule that is divided into 

various development zones: the meristem, the invasion zone where rhizobia move from infection 

threads to plant cells, the interzone where rhizobia are differentiated, the nitrogen fixation zone, 

and the senescence zone where the bacteroids are degraded and nitrogen fixation stops (Suzaki et 

al., 2015). Medicago truncatula and Trifolium species are members of the Inverted Repeat 

Lacking Clade (IRLC), which has lost one of two 25-kb inverted repeats in the chloroplast 

genome (Sprent et al., 2017). IRLC legumes are unique in their rhizobial specificity and the 

degree to which they control their rhizobial partners. IRLC legumes produce nodule cysteine rich 

(NCR) peptides that induce terminal bacteroid differentiation in their rhizobia while also 

increasing rhizobial membrane permeability and inducing extreme endoreduplication (Mergaert 

et al., 2006; Van de Velde et al., 2010). This means that the rhizobia that fix nitrogen in nodules 

of IRLC legumes cannot reproduce; only the undifferentiated rhizobia remaining in the infection 

threads are able to reproduce (Denison, 2000). In addition, the swollen bacteroids induced by 

NCR peptides appear to be more efficient at nitrogen fixation than those in determinate systems 

(Oono & Denison, 2010). Swollen bacteroids have evolved independently at least five times, 

suggesting that this trait is beneficial (Oono & Denison, 2010). However, this theory is difficult 

to test because a given plant species can only form one type of nodule, and few rhizobia nodulate 

plants that produce swollen and nonswollen bacteroids (Oono & Denison, 2010; Terpolilli et al., 

2012). PHB accumulation has not been detected in swollen bacteroids (Paau et al., 1980; Lodwig 

& Poole, 2003), although free living rhizobia and rhizobia in the infection thread can accumulate 

PHB (Paau et al., 1980; Tombolini & Nuti, 1989). In symbiosis, PHB is synthesized but is 

5 
 

presumably used as quickly as it is made, perhaps to fuel bacteroid differentiation (Lodwig et al., 

2005, Trainer & Charles, 2006). In addition, indeterminate nodules export fixed nitrogen as 

asparagine rather than the ureides used by determinate nodules (Poole et al., 2018).  

The initiation of nodules requires a complicated signaling interaction between legumes 

and rhizobia. Rhizobia are chemotactically attracted to plant roots (Miller et al., 2007). When the 

rhizobia get close enough to the roots, they detect flavonoids (polyhydroxy polyphenol 

secondary metabolites) released by the plant. If the plant-rhizobia combination is compatible, the 

flavonoids bind to the nodD receptor protein, which induces the production of rhizobia nod 

factors (Peck et al., 2006). Nod factors are signaling molecules with a chitooligosaccharide 

backbone consisting of several N-acetylglucosamine subunits (Long, 1996). These subunits are 

decorated with various chemical groups, and the structure of the nod factors determines its effect 

on the host plant (Roche et al., 1991; Oldroyd, 2013). If the interaction is compatible, the nod 

factor is recognized by a receptor-like kinase with extracellular LysM domains (Madsen et al., 

2003; Radutoiu et al., 2003). When nod factor is recognized by LysM receptor-like kinases, the 

receptor interacts with a number of proteins to induce calcium spiking (Ehrhardt et al., 1996; 

Kosuta et al., 2008). The calcium-activated kinase CCaMK perceives calcium oscillations first in 

epidermal cells and later in cortical cells, and activates transcription factors to change gene 

expression and lead to nodule organogenesis (Mitra & Long, 2004; Lévy et al., 2004; Guinel, 

2015). A root hair curls around a single rhizobial cell or microcolony, forming a shepherd’s hook 

shape and trapping the rhizobia, which continue to divide (Oldroyd, 2013). The infection thread 

is an invagination of the plant cell, and grows toward the nodule meristem that is developing de 

novo (Gage, 2004). Once the infection thread reaches the nodule, the rhizobia are released into 

infection droplets, which are surrounded by plant membrane tissue (Garg & Renseigne, 2007). 

6 
 

These organelle-like structures are known as symbiosomes, and rhizobia inside them 

differentiate into bacteroids and begin nitrogen fixation plant cells to begin nitrogen fixation 

(Mortier et al., 2012).  

Once rhizobia are inside the nodule, plants create an environment conducive to nitrogen 

fixation. Plants use leghemoglobin proteins to keep free oxygen levels low in nodules and 

prevent inactivation of nitrogenase (Ott et al., 2005).  Plants feed their rhizobia by transporting 

sucrose from the shoot to the root, then converting it to malate before the dicarboxylate 

transporter brings it to the bacteroids (Lodwig & Poole, 2003; Yurgel & Kahn, 2004). In the 

bacteroids, the carbon source feeds the tricarboxylic (TCA) cycle, which provides electrons and 

ATP for the energetically intensive process of nitrogen fixation (Poole & Allaway, 2000; 

Udvardi & Poole, 2013). Most rhizobia lack nifV and thus cannot produce homocitrate, a part of 

the Fe-molybdenum cofactor that is necessary for nitrogenase function (Andrews et al., 2009). 

The plant is able to synthesize homocitrate to allow the rhizobia to fix nitrogen inside a nodule 

(Hakoyama et al., 2009). Plants can also use the environment of the nodule to control rhizobia. 

Bacteroids become auxotrophs and require branched chain amino acids from the plant to fix 

nitrogen (Lodwig et al., 2003; Prell & Poole, 2006; Prell et al., 2010). The ammonia produced by 

nitrogenase is transported to the infected plant cell and assimilated through GS-GOGAT 

(Udvardi & Day, 1997). The form in which nitrogen is transported depends on the type of 

nodule: plants that form indeterminate nodules transport amides (glutamine and asparagine) 

(Prell & Poole, 2006), while plants that form determinate nodules transport nitrogen as ureides 

(Baral et al., 2016).  

As discussed above, for the nodule to be formed, plant flavonoids need to be recognized 

by nodD in the rhizobia and the rhizobial nod factors need to be recognized by plant lysM 

7 
 

receptors. Tropical legumes are typically much more promiscuous than their temperate 

counterparts, but it is not entirely clear what implications this has for signaling between tropical 

legumes and their rhizobia (Lira et al., 2015). There are many levels of specificity in the legume-

rhizobia symbiosis: some rhizobia may be able to initiate nodulation but fail to form successful 

infection threads (Simsek et al., 2007), or form nodules but fail to fix nitrogen (Yates et al., 

2005). The factors underlying these levels of specificity are not entirely clear. Some rhizobia use 

Type III secretion systems to release nodulation outer proteins (NOPs) that may suppress the 

plant immune response or modulate plant cytoskeletal rearrangement during nodule formation 

(Deakin & Broughton, 2009). Another important factor in nodulation is exopolysaccharides 

(EPS), which form part of the rhizobial cell surface. Defects in EPS production can inhibit 

nodule formation during infection thread growth and at the stage of nitrogen fixation (Finan et 

al., 1985; Leigh et al., 1985; Simsek et al., 2007).  

 

Ecological and evolutionary theory 

Since a single plant generally interacts with a number of rhizobial strains in the soil, the 

legume-rhizobia mutualism may be subject to the tragedy of the commons (Hardin, 1968). That 

is, an individual rhizobium benefits from collective nitrogen fixation (which fuels plant 

photosynthesis), but that rhizobium would benefit more from taking its resources away from 

nitrogen fixation and instead focusing them on its own reproduction (Denison, 2000; West et al., 

2002; Denison et al., 2003; Kiers & Denison, 2008). Rhizobia behaving in this way are 

commonly referred to as cheaters. The exact definition of cheating has been a controversial topic 

(Frederickson, 2013; Jones et al., 2015), but the current accepted definition is behavior that 

increases the fitness of the cheater above average fitness in that population, while decreasing the 

8 
 

fitness of the cheater’s partner below average fitness in the partner population (Jones et al., 

2015). Evolutionary theory predicts that cheating rhizobia would take over rhizobial populations 

(Denison et al., 2003), but empirical studies have found a large diversity of nitrogen fixation 

abilities in rhizobial populations (Burdon et al., 1999; Thrall et al., 2000). This leads to two 

questions: first, why do cheaters not take over rhizobial populations? And second, why do poorly 

performing rhizobia exist, if plant controls on cheaters are so effective? 

For question one, several mechanisms have been suggested for how plants control 

cheaters. The first method is partner choice, which is a pre-infection mechanism in which plants 

avoid interacting with poor-quality rhizobia (Bull & Rice, 1991; Simms & Taylor, 2002; Kiers & 

Denison, 2008). Partner choice requires plants to be able to determine a rhizobium’s nitrogen 

fixation level before nodulation, presumably through some sort of signal exchange (Bull & Rice, 

1991; Archetti et al., 2011). Partner choice would also require less investment than other 

methods of mutualism stabilization, since the plant does not have to waste energy initiating 

nodules with low quality partners (Simms & Taylor, 2002). However, effective partner choice 

would require rhizobia to offer honest signals about their nitrogen fixation level (Kiers & 

Denison, 2008). Cooperation theory suggests that low quality rhizobium would be benefited by 

sending dishonest signals about its level of cooperation (Kiers & Denison, 2008; Oono et al., 

2009; Padje et al., 2016). Given that rhizobia are also expected to be at an advantage in 

evolutionary arms races due to their shorter generation time, it seems unlikely that these signals 

would be consistently reliable (Kiers & Denison, 2008; Oono et al., 2009; Padje et al., 2016). 

However, if signals have a high mutation rate and there is linkage disequilibrium between signal 

genes and quality genes, partner choice can maintain cooperation (Jansen & van Baalen, 2006). 

Empirical testing of partner choice in legumes has produced mixed results, and the situation is 

9 
 

further complicated by differing definitions of partner choice and sanctions (Frederickson, 2013). 

Partner choice is notably difficult to test because natural strains that vary in effectiveness likely 

vary in many other ways that could affect nodulation competitiveness, and an isogenic mutant 

that has been made non-fixing would be expected to retain the wild-type signal for effectiveness. 

Partner choice (as demonstrated by preferentially nodulating with the more beneficial strain in a 

multi-strain inoculation) has been demonstrated in Medicago truncatula (Heath & Tiffin, 2009; 

Gubry-Rangin et al., 2010), Lotus strigosus (Sachs et al., 2010b), and Trifolium purpureum and 

Trifolium polymorphum (Yates et al., 2005, 2008). However, a different study in M. truncatula 

showed no evidence of partner choice between effective and ineffective mutants (Amarger, 

1981), a trend that was also detected in Pisum sativa (pea) (Westhoek et al., 2017).  

In contrast to partner choice, sanctions are a post-infection method of host control of 

rhizobia. Sanctions refer to the practice of restricting allocation to poor-performing rhizobia 

(Kiers et al., 2003; Kiers & Denison, 2008). Alternatively, this concept may be viewed in terms 

of preferential allocation, where plants reward high-performing rhizobia with more resources 

(Kiers & van der Heijden, 2006). Sanctions are predicted to require more energy than partner 

choice because the plant has to expend resource initiating nodules that may contain ineffective 

rhizobia (Kiers & Denison, 2008). However, sanctions are expected to be more reliable because 

they avoid the problem of dishonest signaling (Westhoek et al., 2017). Sanctions do not appear 

to be universal, as sanctions have been identified in pea (Westhoek et al., 2017), soybean (Kiers 

et al., 2003), and Lotus japonicus (Regus et al., 2014, 2015), but not in Medicago truncatula 

(Heath & Tiffin, 2009; Gubry-Rangin et al., 2010; Grillo et al., 2016). A major question in this 

area is the scale at which plants are able to discriminate between different rhizobial strains--on 

the whole root level, single nodule, or somewhere in between.  

10 

 

A final mechanism that may prevent the spread of noncooperative rhizobia is partner 

fidelity feedback. Partner fidelity feedback means that the benefits a symbiont will receive from 

its partner in the future depend on its investment in the present, because there are positive 

feedbacks between host and partner fitness (Bull & Rice, 1991; Sachs et al., 2004). This 

mechanism requires spatial structuring of the population or vertical transmission of symbionts 

that makes an individual or its relatives more likely to interact repeatedly over time (Yamamura, 

1996; Doebeli & Knowlton, 1998). Meta-analysis has provided empirical support for partner-

fidelity feedback by showing a positive correlation between plant performance and nodule 

number and biomass (Friesen, 2012). 

This leaves the question of why poor-performing rhizobia are able to persist. Effective 

host control mechanisms are predicted to produce rhizobial populations that consist entirely of 

cooperators (West et al., 2002; Sachs et al., 2004; Foster & Wenseleers, 2006). However, 

ineffective rhizobia are abundant in many environments (Burdon et al., 1999; Denton et al., 

2002; Sachs et al., 2010a). One possible explanation is the existence of mixed nodules. While 

most nodules consist of a rhizobial population that originated from a single cell trapped in a 

curling root hair, mixed nodules that are initiated by multiple cells of different strains are 

possible, with 2 to 74% of nodules containing two different strains under lab testing (Denison, 

2000; Gage, 2002; Westhoek et al., 2017). Depending on the scale at which sanctions operate, 

mixed nodules may allow noncooperative rhizobia to escape host control (Denison, 2000; Kiers 

& Denison, 2008; Steidinger & Bever, 2016). However, recent evidence suggests that at least 

some plants can specifically sanction noncooperative rhizobia in mixed nodules with cooperative 

rhizobia (Regus et al., 2017; Daubech et al., 2017). In addition, the mechanisms used by plants 

11 

 

to control poor performing rhizobia do not appear to be universal-different plant species use 

different mechanisms (Pahua et al., 2018).  

 

The abiotic environment and the legume-rhizobia interaction 

Legumes have two options for acquiring nitrogen: direct uptake from the soil and trade 

with rhizobia. Plants can take up either inorganic nitrogen in the form of nitrate or ammonium, or 

organic nitrogen in the form of amino acids and peptides (Mohd-Radzman et al., 2013). Nitrate 

tends to be the dominant nitrogen source in aerated soils, while ammonium is more prevalent in 

acidic and anaerobic soils (Miller & Cramer, 2005). While organic nitrogen sources may play 

important roles in boreal ecosystems, they have received relatively little attention (Näsholm et 

al., 2009) and will not be addressed further. Ammonium and nitrate tend to receive more 

attention because they are typically present in agricultural systems at much higher levels than 

any other nitrogen source (Miller & Cramer, 2005). Nitrate is the dominant form of nitrogen in 

most soils, and plants generally prefer nitrate as compared to ammonium because excess nitrate 

can be stored, while excessive ammonium levels can lead to toxicity for the plant (Glass et al., 

2002). Since nitrogen fixation is energetically expensive (Voisin et al., 2003), plants tend to 

reduce nodulation at high soil nitrogen levels and favor direct nitrogen acquisition instead 

(Voisin et al., 2002). 

High concentrations of nitrogen (above 3 mM), whether as nitrate or ammonium, tend to 

uniformly inhibit nodulation (Silsbury et al., 1986; Gan et al., 2004; Fei & Vessey, 2009; Dan & 

Brix, 2009; Mohd-Radzman et al., 2013). Lower concentrations of nitrogen generally promote 

nodulation (e.g. Weber et al., 2007), though the opposite effect has been shown in other studies 

(Dan & Brix, 2009). Not all studies assess both total nodule number and specific nodulation, or 

12 

 

nodulation per unit of root biomass, so a treatment that increases root biomass and does not 

change specific nodulation could report an increase in total nodule number simply due to 

increased plant size. In Pisum sativum, both total and specific nodulation were increased by low 

concentrations of ammonium (Gulden & Vessey, 1997), while in Glycine max, low ammonium 

concentrations increase total nodulation but decrease specific nodulation (Gulden & Vessey, 

1998; Gan et al., 2004).  

Nitrate and ammonium have different regulatory effects on root and nodule development, 

but when plants are exposed to both forms of nitrogen, the effects of nitrate outweigh those of 

ammonium (Bollman & Vessey, 2006). Ammonium effects on nodulation are less well studied 

than those of nitrate (Forde & Clarkson, 1999; Gan et al., 2004). Ammonium appears to inhibit 

nodulation at a relatively early stage in the signaling pathway by inhibiting root hair curling 

(Barbulova et al., 2007). In contrast, nitrate inhibits nodulation downstream of root hair curling 

(Barbulova et al., 2007). Nitrate appears to alter flavonoid metabolism, plant defense responses, 

and the redox state (van Noorden et al., 2016). Nitrate is also believed to play a role in 

autoregulation of nodulation (AON), a process plants use to control total nodule number (Mortier 

et al., 2012). The mechanisms of AON have not been fully elucidated, but AON is believed to 

occur when a signal is produced in the root in response to nodulation (Kinkema et al., 2006). 

This signal is transported to the shoot, where it induces the production of another signaling 

molecule that is transported back to the root, where it inhibits the formation of more nodules 

(Downie, 2014). Evidence suggests that CLE peptides are the root-produced signal, and that they 

are recognized by leucine-rich repeat receptor-like kinases (LRR-RLKs) in the shoot (Miyazawa 

et al., 2010; Krusell et al., 2011; Okamoto et al., 2013; Araya et al., 2016; Nishida et al., 2016). 

These receptors induce the production of cytokinin that is then transported to the roots (Sasaki et 

13 

 

al., 2014), where it downregulates nodulation through TML and NIN genes (Takahara et al., 

2013; Soyano et al., 2015). Mutants that are deficient in AON are also frequently nitrate 

insensitive, suggesting that nitrate signaling and AON may be interconnected (Schnabel et al.; 

Wopereis et al., 2000; Magori et al., 2009). However, nitrate still negatively affected nodulation 

in AON-deficient mutants (Jeudy et al., 2010; Okamoto & Kawaguchi, 2015), showing that 

nitrate affects nodulation in ways other than just the AON pathway (Nishida & Suzaki, 2018).  

While the effects of nitrogen on nodulation have been well characterized, there is less 

literature on the effects of light on the legume-rhizobia symbiosis. Nitrogen fixation by rhizobia 

is an energy-intensive process, and can require up to 28% of the plant’s total supply of 

photosynthate (Kaschuk et al., 2009). It has been suggested that rhizobia can increase the plant’s 

photosynthesis rate because they form such a strong carbon sink (Kaschuk et al., 2009). An 

interaction between Phaseolus lunatus and Rhizobium that increased plant biomass and seed 

production relative to uninoculated plants in high light (600 μmolm-2s-1) decreased plant biomass 

and seed production relative to uninoculated plants in low light conditions (300 μmolm-2s-1) 

(Ballhorn et al., 2016). Nodule biomass was unaffected by light level in this study (Ballhorn et 

al., 2016). Bradyrhizobium japonicus increased Glycine max shoot biomass and did not affect 

root biomass at high light (300 μmolm-2s-1), while it decreased root biomass and did not affect 

shoot biomass at low light (50 μmolm-2s-1) (Lau et al., 2012). In this system, decreasing light 

sharply decreased nodule biomass (Lau et al., 2012). Light effects have been more thoroughly 

studied in forest ecosystems due to the importance and prevalence of shading in those 

ecosystems. In these systems, light has been shown to change responses to mycorrhizal 

colonization (Gehring, 2003) and to qualitatively change plant-soil feedback, or the process by 

which plants alter the composition of the microbial community, which then changes the fitness 

14 

 

benefits the plant receives from that community (Smith & Reynolds, 2015). The effects of light 

on plant-mycorrhizae interactions have been more thoroughly studied. Short term (six day) 

shading significantly disrupted phosphate transportation in the Medicago truncatula-

Rhizophagus irregularis symbiosis, while long term shading (38 days) at 35% or below of full 

greenhouse daylight (~5% of natural sunlight) decreased plant growth in R. irregularis-

inoculated plants compared to control plants (Konvalinkova et al., 2015). Decreasing light levels 

from 1300 μmolm-2s-1 to 660 μmolm-2s-1 reduced the ability of Allium vineale to preferentially 

allocate carbon to more effective mycorrhizae (Zheng et al., 2015).  

The molecular pathways that translate changes in light into changes in nodulation are not 

well understood. The effects of light on a soybean variety defective in AON suggests that 

increased light and CO2 levels increase nodulation in a manner independent of the AON 

pathway (Hansen et al., 1990; Bacanamwo & Harper, 1997). Chlorophyll absorbs red light but 

does not absorb far red light. Thus, plants can use phytochrome to detect the red (R) to far red 

(FR) ratio to determine whether there are other plants nearby competing for light. Lotus 

japonicus phyB mutants show reduced nodulation, and wild-type plants respond to high R/FR 

through JA signaling (Suzuki et al., 2011). In the phyB mutant, nodulation is restricted due to 

reduced JA-Ile production and reduced transport of JA-Ile from the shoot to the root (Shigeyama 

et al., 2012). Similarly, the R/FR ratio regulates mycorrhizal colonization through JA signaling 

(Nagata et al., 2015). Light availability alters levels of flavonoids in legumes (Zavala et al., 

2015), which due to their importance in signaling with rhizobia (Zhang et al., 2009), may alter 

rhizobial colonization levels (Gundel et al., 2014). 

 

 

15 

 

The biotic environment and the legume-rhizobia interaction 

Arbuscular mycorrhizal fungi (AMF) are obligate plant symbionts that can supply 

phosphorus, nitrogen, and other resources to plants in exchange for photosynthetic carbon 

(Parniske, 2008). Dual inoculation with AMF and rhizobia is predicted to have synergistic 

effects on plant and microbial performance, since phosphate is a common limiting nutrient for 

both plant growth and rhizobial nitrogen fixation (Augusto et al., 2013). Indeed, a number of 

studies have illustrated increased nitrogen fixation and plant growth in the presence of AMF 

compared to singly-inoculated plants (Carling, 1978; Khan et al., 1995; Chalk et al., 2006; Wang 

et al., 2011). However, this synergy does not appear to be universal. A meta-analysis detected no 

synergistic effects between AMF and rhizobia, though the authors caution this may be due to 

lack of control for soil nutrient levels, which play important roles in both individual symbioses 

and their synergies (Larimer et al., 2010; Wang et al., 2011; Ossler et al., 2015). In some cases, 

it seems that dual inoculation with rhizobia and AMF may overtax the limited plant C supply, 

leading to reduced symbiont function (Bethlenfalvay et al., 1985; Brown & Bethlenfalvay, 1987) 

or shifting the plant to carbon limitation rather than nutrient limitation, particularly in lower light 

greenhouse or growth chamber studies (Bethlenfalvay et al., 1983).  

The allocation of carbon between rhizobia and AMF in tripartite symbiosis has not been 

thoroughly explored. It is believed that rhizobia co-opted mycorrhizal signaling for nodule 

establishment, and the two still share a common symbiosis signaling pathway [cite]. Despite this 

overlap, there is no evidence for genetic covariance between the establishment and regulation of 

rhizobial and mycorrhizal symbioses (Ossler & Heath, 2018). A preliminary transcriptomic 

analysis rhizobia-AMF interactions in Medicago truncatula found complex patterns of changes 

in differentially expressed genes in singly and dually-inoculated plants (Afkhami & 

16 

 

Stinchcombe, 2016). Mycorrhizal colonization displays autoregulation patterns similar to 

autoregulation of nodulation: once a threshold of colonization is reached, the plant uses systemic 

signaling to suppress further colonization (Vierheilig, 2004). In split-root systems, pre-

colonization of one half of the root system with AMF or rhizobia suppresses colonization by the 

other symbiont on the other half of the root system (Catford et al., 2003, 2006; Meixner et al., 

2005). Rhizobial inhibition of mycorrhizal colonization as well as mycorrhizal autoregulation are 

deficient in autoregulation of nodulation (AON) mutants, suggesting that autoregulation of both 

symbionts and inhibition of one symbiont by the other may use the AON pathway (Meixner et 

al., 2005; Sakamoto et al., 2013).  

 

Rhizobia are only a small part of a large, complex community of microbes in the soil. 

Plants exert strong effects on the community of microbes associated with their root surface, or 

rhizosphere (Turner et al., 2013). This community has a very different composition from and is 

much less diverse than the microbial community in bulk soil, likely due to plant selection 

(Turner et al., 2013; Tkacz et al., 2015). The effects of rhizobial symbiosis on the makeup of the 

rest of the microbiome has not been thoroughly studied. There is evidence that the nodulation 

signaling pathway plays a role in microbiome structuring. Lotus japonicus Nod factor receptor5 

(nfr5), Nodule inception (nin) and Lotus histidine kinase1 (lhk1) exhibited extremely different 

root and rhizosphere microbiomes compared to wildtype plants (Zgadzaj et al., 2016). These 

differences were maintained even under nitrogen levels that eliminated nodulation in wildtype 

plants, suggesting that these effects are due to the changes to symbiosis signaling and not just the 

loss of symbiosis (Zgadzaj et al., 2016). In addition, the H2 produced by nodules as a byproduct 

of nitrogen fixation can affect the makeup of the microbiome and may select for plant-beneficial 

bacteria (Dong et al., 2003; Maimaiti et al., 2007). 

17 

 

 

Other members of the plant microbiome may play important roles in mediating the 

legume-rhizobia symbiosis. There are a number of examples of non-rhizobial “helper strains” 

improving rhizobial colonization and plant growth benefit when co-inoculated with rhizobia 

(Sturz et al., 1997; Bai et al., 2002; Egamberdieva et al., 2010). The mechanisms by which these 

“helper strains” improve the rhizobial symbiosis are not entirely known, but may include 

cellulase (Ibáñez et al., 2009), auxin (Ibáñez et al., 2009; Mishra et al., 2009), and siderophore 

production (Maymon et al., 2015), and bacterial quorum sensing (Miao et al., 2018).  Similarly, 

soil bacteria may improve mycorrhizal colonization in a number of ways, including increasing 

the receptiveness of the host plant to colonization, improving soil conditions for fungal growth, 

and promoting spore germination and survival (Artursson et al., 2006; Pivato et al., 2009; 

Hassani et al., 2018).  

 

On a larger scale, legume-rhizobia interactions can also play a role in plant-plant 

interactions. Rhizobial symbionts may offer plants an opportunity for niche differentiation 

(Bever, 1999; Parker, 1999), which can promote coexistence and biodiversity (Chesson, 2000). 

Variation in rhizobial quality can increase the fitness advantage of competing with kin as 

opposed to nonkin for legumes (Simonsen et al., 2014).  Legumes may derive more nitrogen 

from fixation when they are competing with non-leguminous plants (Karpenstein-Machan & 

Stuelpnagel, 2000; Hodge & Fitter, 2013). Differentiating their nitrogen uptake strategy to 

nitrogen fixation and away from direct soil uptake may free up resources for the growth of non-

nitrogen fixing plants (Temperton et al., 2007; von Felten et al., 2009). Interacting with 

microbial symbionts such as rhizobia and AMF has been shown to increase plant diversity in 

model grassland communities (van der Heijden et al., 2015). However, these effects are likely 

context-dependent, as symbionts do not affect competition and coexistence in fertilized 

18 

 

ecosystems (Ren et al., 2017). In addition, in some ecosystems with diverse rhizobial 

populations, mutualism may actually lead to competitive exclusion rather than coexistence of 

multiple plant lineages (Wilkinson & Parker, 1996; Keller, 2014). Rhizobia can also affect the 

ability of invasive species to disrupt existing ecosystems: the degree of association with rhizobia 

is positively correlated with invasion success in Acacia, likely due to the benefits of high levels 

of nitrogen fixation (Rodríguez-Echeverría et al., 2009). Conversely, interactions between plants 

can also affect how legumes interact with their rhizobia. Intercropping maize with faba bean 

increases yield and nodulation because maize root exudates increase faba bean flavonoid 

biosynthesis, leading to increased nodulation and nitrogen fixation (Li et al., 2016). Allelopathic 

chemicals released by other plants can disrupt the legume-rhizobia symbiosis of another plant 

(Alsaadawi & Rice, 1982; Portales-Reyes et al., 2015). 

 

Conclusions 

The nitrogen fixing symbiosis between legumes and rhizobia is an ancient relationship 

with important implications for the functioning of agricultural and natural ecosystems. This 

resource trade has frequently been conceptualized as a biological market, with accompanying 

assumptions about how trade negotiations are conducted (Grman et al., 2012; Clark et al., 2017). 

Unfortunately, this theoretical work has rarely been parameterized with empirical data (Clark et 

al., 2017). To address this gap, chapter 1 describes empirical model parameterization that tests 

the assumption of fair trade between legume and rhizobia, one of the central assumptions of 

theoretical biological market models. Importantly, this work assesses how soil nitrogen 

availability affects trade negotiations between legume and rhizobia. In chapter 2, this assessment 

of the effects of nutrient availability is extended to explore the effects of manipulating the 

19 

 

availability of both traded resources. While the effects of nitrogen on nodulation have been 

extensively studied, the effects of carbon availability are less well understood, and there has been 

almost no exploration of their interactive effects. Thus, chapter 3 will explore how factorial 

manipulation of nitrogen and light affect the legume-rhizobia symbiosis. In chapter 4, the focus 

shifts from the effects of the external environment to how the identity of the legume and rhizobia 

affect the costs and benefits of trade. This research provides insight into how both external and 

intrinsic factors affect the balance of trade between legumes and rhizobia and will provide 

insights into the evolutionary dynamics of the symbiosis and its role in natural and agricultural 

ecosystems.  

 

 

 

 

 

 

 

 

 

 

 

 

 

 

20 

 

 

 

 

 

 

 

 

 

 

 

 

 

REFERENCES 

 

 

 

 

 

 

 

 

 

 

21 

 

REFERENCES 

 

Afkhami ME, Stinchcombe JR. 2016. Multiple mutualist effects on genomewide expression in 
the tripartite association between Medicago truncatula, nitrogen-fixing bacteria and mycorrhizal  
fungi. Molecular Ecology 25: 4946–4962. 

Alsaadawi IS, Rice EL. 1982. Allelopathic effects ofPolygonum aviculare L. II. Isolation, 
characterization, and biological activities of phytotoxins. Journal of Chemical Ecology 8: 1011– 
1023. 

Amadou C, Pascal G, Mangenot S, Glew M, Bontemps C, Capela D, Carrère S, Cruveiller 
S, Dossat C, Lajus A, et al. 2008. Genome sequence of the beta-rhizobium Cupriavidus  
taiwanensis and comparative genomics of rhizobia. Genome research 18: 1472–83. 

Amarger N. 1981. Competition for nodule formation between effective and ineffective strains of  
Rhizobium meliloti. Soil Biology and Biochemistry 13: 475–480. 

Andrews M, Andrews ME. 2017. Specificity in Legume-Rhizobia Symbioses. International  
Journal of Molecular Sciences 18: 705. 

Andrews M, Lea PJ, Raven JA, Azevedo RA. 2009. Nitrogen use efficiency. 3. Nitrogen  
fixation: genes and costs. Annals of Applied Biology 155: 1–13. 

Araya T, von Wirén N, Takahashi H. 2016. CLE peptide signaling and nitrogen interactions in  
plant root development. Plant Molecular Biology 91: 607–615. 

Archetti M, Ubeda F, Fudenberg D, Green J, Pierce NE, Yu DW. 2011. Let the right one in:  
a microeconomic approach to partner choice in mutualisms. The American naturalist 177: 75–85. 

Artursson V, Finlay RD, Jansson JK. 2006. Interactions between arbuscular mycorrhizal fungi 
and bacteria and their potential for stimulating plant growth. Environmental Microbiology 8: 1– 
10. 

Augusto L, Delerue F, Gallet-Budynek A, Achat DL. 2013. Global assessment of limitation to 
symbiotic nitrogen fixation by phosphorus availability in terrestrial ecosystems using a meta- 
analysis approach. Global Biogeochemical Cycles 27: 804–815. 

Azani N, Babineau M, Bailey CD, Banks H, Barbosa A, Pinto RB, Boatwright J, Borges L, 
Brown G, Bruneau A, et al. 2017. A new subfamily classification of the Leguminosae based on 
a taxonomically comprehensive phylogeny – The Legume Phylogeny Working Group (LPWG).  
Taxon 66: 44–77. 

Bacanamwo M, Harper JE. 1997. Response of a hypernodulating soybean mutant to increased  
photosynthate supply. Plant Science 124: 119–129. 

22 

 

Bago B, Pfeffer P, Shachar-Hill Y. 2000. Carbon Metabolism and Transport in Arbuscular  
Mycorrhizas. Plant Physiology 124: 949–958. 

Bai Y, D’Aoust F, Smith DL, Driscoll BT. 2002. Isolation of plant-growth-promoting Bacillus  
strains from soybean root nodules. Canadian Journal of Microbiology 48: 230–238. 

Ballhorn DJ, Schädler M, Elias JD, Millar JA, Kautz S, Smith S, Read D, Harley J, Sprent 
J, Carney K, et al. 2016. Friend or Foe—Light Availability Determines the Relationship 
between Mycorrhizal Fungi, Rhizobia and Lima Bean (Phaseolus lunatus L.) (R Aroca, Ed.).  
PLOS ONE 11: e0154116. 

Baral B, Teixeira da Silva JA, Izaguirre-Mayoral ML. 2016. Early signaling, synthesis,  
transport and metabolism of ureides. Journal of Plant Physiology 193: 97–109. 

Barbulova A, Rogato A, D’Apuzzo E, Omrane S, Chiurazzi M. 2007. Differential Effects of 
Combined N Sources on Early Steps of the Nod Factor–Dependent Transduction Pathway in  
Lotus japonicus. Molecular Plant-Microbe Interactions 20: 994–1003. 

Bethlenfalvay GJ, Bayne HG, Pacovsky RS. 1983. Parasitic and mutualistic associations 
between a mycorrhizal fungus and soybean: The effect of phosphorus on host plant-endophyte  
interactions. Physiologia Plantarum 57: 543–548. 

Bethlenfalvay GJ, Brown MS, Stafford AE. 1985. Glycine-Glomus-Rhizobium Symbiosis: II. 
Antagonistic Effects between Mycorrhizal Colonization and Nodulation. Plant physiology 79:  
1054–8. 

Bever JD. 1999. Dynamics within mutualism and the maintenance of diversity: inference from a  
model of interguild frequency dependence. Ecology Letters 2: 52–61. 

Bollman MI, Vessey JK. 2006. Differential effects of nitrate and ammonium supply on nodule 
initiation, development, and distribution on roots of pea ( Pisum sativum ). Canadian Journal of  
Botany 84: 893–903. 

Bronstein JL (Ed.). 2015. Mutualism. Oxford: Oxford University Press. 

Brown MS, Bethlenfalvay GJ. 1987. Glycine-Glomus-Rhizobium Symbiosis : VI. 
Photosynthesis in Nodulated, Mycorrhizal, or N- and P-Fertilized Soybean Plants. Plant  
physiology 85: 120–123. 

Bull JJ, Rice WR. 1991. Distinguishing mechanisms for the evolution of co-operation. Journal  
of Theoretical Biology 149: 63–74. 

Burdon JJ, Gibson AH, Searle SD, Woods MJ, Brockwell J. 1999. Variation in the 
effectiveness of symbiotic associations between native rhizobia and temperate Australian Acacia:  
within-species interactions. Journal of Applied Ecology 36: 398–408. 

23 

 

Carling DE. 1978. Effects of a Vesicular-Arbuscular Mycorrhizal Fungus on Nitrate Reductase 
and Nitrogenase Activities in Nodulating and Non-Nodulating Soybeans. Phytopathology 68:  
1590. 

Catford JG, Staehelin C, Larose G, Piché Y, Vierheilig H. 2006. Systemically suppressed 
isoflavonoids and their stimulating effects on nodulation and mycorrhization in alfalfa split-root  
systems. Plant and Soil 285: 257–266. 

Catford J-G, Staehelin C, Lerat S, Piché Y, Vierheilig H. 2003. Suppression of arbuscular 
mycorrhizal colonization and nodulation in split-root systems of alfalfa after pre-inoculation and  
treatment with Nod factors. Journal of Experimental Botany 54: 1481–1487. 

Chalk PM, Souza R de F, Urquiaga S, Alves BJR, Boddey RM. 2006. The role of arbuscular  
mycorrhiza in legume symbiotic performance. Soil Biology and Biochemistry 38: 2944–2951. 

Chamberlain SA, Bronstein JL, Rudgers JA. 2014. How context dependent are species  
interactions? (R Etienne, Ed.). Ecology Letters 17: 881–890. 

Chesson P. 2000. Mechanisms of Maintenance of Species Diversity. Annual Review of Ecology  
and Systematics 31: 343–366. 

Clark TJ, Friel CA, Grman E, Shachar-Hill Y, Friesen ML. 2017. Modelling nutritional 
mutualisms: challenges and opportunities for data integration (M van Baalen, Ed.). Ecology  
Letters 20: 1203–1215. 

Dan TH, Brix H. 2009. Growth responses of the perennial legume Sesbania sesban to NH4 and  
NO3 nutrition and effects on root nodulation. Aquatic Botany 91: 238–244. 

Daubech B, Remigi P, Doin de Moura G, Marchetti M, Pouzet C, Auriac M-C, Gokhale 
CS, Masson-Boivin C, Capela D. 2017. Spatio-temporal control of mutualism in legumes helps  
spread symbiotic nitrogen fixation. eLife 6. 

Deakin WJ, Broughton WJ. 2009. Symbiotic use of pathogenic strategies: rhizobial protein  
secretion systems. Nature Reviews Microbiology 7: 312. 

Denison RF. 2000. Legume Sanctions and the Evolution of Symbiotic Cooperation by Rhizobia.  
The American Naturalist 156: 567–576. 

Denison RF, Bledsoe C, Kahn M, O’Gara F, Simms EL, Thomashow LS. 2003. Cooperation  
in the rhizosphere and the ‘free rider’ problem. Ecology 84: 838–845. 

Denton MD, Coventry DR, Murphy PJ, Howieson JG, Bellotti WD. 2002. Competition 
between inoculant and naturalised Rhizobium leguminosarum bv. trifolii for nodulation of  
annual clovers in alkaline soils. Australian Journal of Agricultural Research 53: 1019. 

Doebeli M, Knowlton N. 1998. The evolution of interspecific mutualisms. Proceedings of the 

24 

 

National Academy of Sciences of the United States of America 95: 8676–80. 

Dong Z, Wu L, Kettlewell B, Caldwell CD, Layzell DB. 2003. Hydrogen fertilization of soils -  
is this a benefit of legumes in rotation? Plant, Cell and Environment 26: 1875–1879. 

Downie JA. 2014. Legume nodulation. Current Biology 24: R184–R190. 

Egamberdieva D, Berg G, Lindström K, Räsänen LA. 2010. Co-inoculation of Pseudomonas 
spp. with Rhizobium improves growth and symbiotic performance of fodder galega (Galega  
orientalis Lam.). European Journal of Soil Biology 46: 269–272. 

Ehrhardt DW, Wais R, Long SR. 1996. Calcium Spiking in Plant Root Hairs Responding to  
Rhizobium Nodulation Signals. Cell 85: 673–681. 

Fei H, Vessey JK. 2009. Stimulation of nodulation in Medicago truncatula by low 
concentrations of ammonium: quantitative reverse transcription PCR analysis of selected genes.  
Physiologia Plantarum 135: 317–330. 

von Felten S, Hector A, Buchmann N, Niklaus PA, Schmid B, Scherer-Lorenzen M. 2009. 
Belowground nitrogen partitioning in experimental grassland plant communities of varying  
species richness. Ecology 90: 1389–1399. 

Finan TM, Hirsch AM, Leigh JA, Johansen E, Kuldau GA, Deegan S, Walker GC, Signer 
ER. 1985. Symbiotic mutants of rhizobium meliloti that uncouple plant from bacterial  
differentiation. Cell 40: 869–877. 

Forde BG, Clarkson DT. 1999. Nitrate and Ammonium Nutrition of Plants: Physiological and  
Molecular Perspectives. Advances in Botanical Research 30: 1–90. 

Foster KR, Wenseleers T. 2006. A general model for the evolution of mutualisms. Journal of  
Evolutionary Biology 19: 1283–1293. 

Frederickson ME. 2013. Rethinking Mutualism Stability: Cheaters and the Evolution of  
Sanctions. The Quarterly Review of Biology 88: 269–295. 

Friesen ML. 2012. Widespread fitness alignment in the legume-rhizobium symbiosis. The New  
phytologist 194: 1096–111. 

Friesen ML, Heath KD. 2013. One hundred years of solitude: integrating single-strain 
inoculations with community perspectives in the legume-rhizobium symbiosis. New Phytologist  
198: 7–9. 

Gage DJ. 2002. Analysis of infection thread development using Gfp- and DsRed-expressing  
Sinorhizobium meliloti. Journal of bacteriology 184: 7042–6. 

Gage DJ. 2004. Infection and Invasion of Roots by Symbiotic, Nitrogen-Fixing Rhizobia during  

25 

 

Nodulation of Temperate Legumes. Microbiology and Molecular Biology Reviews 68: 280–300. 

Gan Y, Stulen I, van Keulen H, Kuiper PJC. 2004. Low concentrations of nitrate and 
ammonium stimulate nodulation and N2 fixation while inhibiting specific nodulation (nodule 
DW g−1 root dry weight) and specific N2 fixation (N2 fixed g−1 root dry weight) in soybean.  
Plant and Soil 258: 281–292. 

Gano-Cohen KA, Stokes PJ, Blanton MA, Wendlandt CE, Hollowell AC, Regus JU, Kim 
D, Patel S, Pahua VJ, Sachs JL. 2016. Nonnodulating Bradyrhizobium spp. Modulate the 
Benefits of Legume-Rhizobium Mutualism. Applied and environmental microbiology 82: 5259– 
68. 

Garg N, Renseigne N. 2007. Symbiotic nitrogen fixation in legume nodules: process and  
signaling. A review. Sustain. Dev 27: 59–68. 

Gehring CA. 2003. Growth responses to arbuscular mycorrhizae by rain forest seedlings vary  
with light intensity and tree species. Plant Ecology 167: 127–139. 

Ghoul M, Griffin AS, West SA. 2014. TOWARD AN EVOLUTIONARY DEFINITION OF  
CHEATING. Evolution 68: 318–331. 

Glass ADM, Britto DT, Kaiser BN, Kinghorn JR, Kronzucker HJ, Kumar A, Okamoto M, 
Rawat S, Siddiqi MY, Unkles SE, et al. 2002. The regulation of nitrate and ammonium  
transport systems in plants. Journal of Experimental Botany 53: 855–864. 

Grillo MA, Stinchcombe JR, Heath KD. 2016. Nitrogen addition does not influence pre-
infection partner choice in the legume-rhizobium symbiosis. American journal of botany 103:  
1763–1770. 

Grman E, Robinson TMP, Klausmeier CA. 2012. Ecological specialization and trade affect  
the outcome of negotiations in mutualism. The American naturalist 179: 567–81. 

Gubry-Rangin C, Garcia M, Béna G. 2010. Partner choice in Medicago truncatula- 
Sinorhizobium symbiosis. Proceedings. Biological sciences / The Royal Society 277: 1947–51. 

Guinel FC. 2015. Ethylene, a Hormone at the Center-Stage of Nodulation. FRONTIERS IN  
PLANT SCIENCE 6: 1121–1121. 

Gulden RH, Vessey JK. 1997. The stimulating effect of ammonium on nodulation in Pisum 
sativum L. is not long lived once ammonium supply is discontinued. Plant and Soil 195: 195– 
205. 

Gulden RH, Vessey JK. 1998. Low concentrations of ammonium inhibit specific nodulation 
(nodule number g-1 root DW) in soybean (Glycine max [L.] Merr.). Plant and Soil 198: 127– 
136. 

26 

 

Gundel PE, Pierik R, Mommer L, Ballaré CL. 2014. Competing neighbors: light perception  
and root function. Oecologia 176: 1–10. 

Hakoyama T, Niimi K, Watanabe H, Tabata R, Matsubara J, Sato S, Nakamura Y, Tabata 
S, Jichun L, Matsumoto T, et al. 2009. Host plant genome overcomes the lack of a bacterial  
gene for symbiotic nitrogen fixation. Nature 462: 514–7. 

Hansen AP, Gresshoff PM, Pate JS, Day DA. 1990. Interactions between Irradiance Levels, 
Nodulation and Nitrogenase Activity of Soybean cv. Bragg and a Supernodulating Mutant.  
Journal of Plant Physiology 136: 172–179 

Hardin G. 1968. The tragedy of the commons. Science 162: 1243–1248. 

Hassani MA, Durán P, Hacquard S. 2018. Microbial interactions within the plant holobiont.  
Microbiome 2018 6:1 6: 58. 

Heath KD, Tiffin P. 2009. Stabilizing mechanisms in a legume-rhizobium mutualism.  
Evolution; international journal of organic evolution 63: 652–62. 

van der Heijden MG, Bruin S de, Luckerhoff L, van Logtestijn RS, Schlaeppi K. 2015. A 
widespread plant-fungal-bacterial symbiosis promotes plant biodiversity, plant nutrition and  
seedling recruitment. The ISME journal 10: 389–399. 

Hodge A, Fitter AH. 2013. Microbial mediation of plant competition and community structure  
(D Robinson, Ed.). Functional Ecology 27: 865–875. 

Hollowell AC, Regus JU, Turissini D, Gano-Cohen KA, Bantay R, Bernardo A, Moore D, 
Pham J, Sachs JL. 2016. Metapopulation dominance and genomic-island acquisition of 
Bradyrhizobium with superior catabolic capabilities. Proceedings of the Royal Society of London  
B: Biological Sciences 283. 

Ibáñez F, Angelini J, Taurian T, Tonelli ML, Fabra A. 2009. Endophytic occupation of 
peanut root nodules by opportunistic Gammaproteobacteria. Systematic and Applied  
Microbiology 32: 49–55. 

Jansen VAA, van Baalen M. 2006. Altruism through beard chromodynamics. Nature 440: 663– 
666. 

Jeudy C, Ruffel S, Freixes S, Tillard P, Santoni AL, Morel S, Journet E-P, Duc G, Gojon A, 
Lepetit M, et al. 2010. Adaptation of Medicago truncatula to nitrogen limitation is modulated  
via local and systemic nodule developmental responses. New Phytologist 185: 817–828. 

Johnson NC, Graham JH, Smith FA. 1997. Functioning of mycorrhizal associations along the  
mutualism-parasitism continuum. New Phytologist 135: 575–585. 

Jones EI, Afkhami ME, Akçay E, Bronstein JL, Bshary R, Frederickson ME, Heath KD, 

27 

 

Hoeksema JD, Ness JH, Pankey MS, et al. 2015. Cheaters must prosper: reconciling theoretical 
and empirical perspectives on cheating in mutualism (N van Dam, Ed.). Ecology Letters 18:  
1270–1284. 

Karpenstein-Machan M, Stuelpnagel R. 2000. Biomass yield and nitrogen fixation of legumes 
monocropped and intercropped with rye and rotation effects on a subsequent maize crop. Plant  
and Soil 218/2: 215–232. 

Kaschuk G, Kuyper TW, Leffelaar PA, Hungria M, Giller KE. 2009. Are the rates of 
photosynthesis stimulated by the carbon sink strength of rhizobial and arbuscular mycorrhizal  
symbioses? Soil Biology and Biochemistry 41: 1233–1244. 

Keller KR. 2014. Mutualistic rhizobia reduce plant diversity and alter community composition.  
Oecologia 176: 1101–1109. 

Khan MK, Sakamoto K, Yoshida T. 1995. Dual inoculation of peanut with Glomus sp. and 
Bradyrhizobium sp. enhanced the symbiotic nitrogen fixation as assessed by 15 N-Technique. Soil  
Science and Plant Nutrition 41: 769–779. 

Kiers ET, Denison RF. 2008. Sanctions, Cooperation, and the Stability of Plant-Rhizosphere  
Mutualisms. Annual Review of Ecology, Evolution, and Systematics 39: 215–236. 

Kiers ET, van der Heijden MGA. 2006. Mutualistic stability in the the arbuscular mycorrhizal  
symbiosis: exploring hypotheses of evolutionary cooperation. Ecology 87: 1627–1636. 

Kiers ET, Rousseau RA, West SA, Denison RF. 2003. Host sanctions and the legume– 
rhizobium mutualism. Nature 425: 78–81. 

Kinkema M, Scott PT, Gresshoff PM. 2006. Legume nodulation: successful symbiosis through  
short- and long-distance signalling. Functional Plant Biology 33: 707. 

Konvalinková T, Pachel D, Janoušková M, Gryndler M, Jansa J. 2015. Duration and 
intensity of shade differentially affects mycorrhizal growth- and phosphorus uptake responses of  
Medicago truncatula. Frontiers in Plant Science 6: 65. 

Kosuta S, Hazledine S, Sun J, Miwa H, Morris RJ, Downie JA, Oldroyd GED. 2008. 
Differential and chaotic calcium signatures in the symbiosis signaling pathway of legumes.  
Proceedings of the National Academy of Sciences 105: 9823–9828. 

Krusell L, Sato N, Fukuhara I, Koch BE V., Grossmann C, Okamoto S, Oka-Kira E, 
Otsubo Y, Aubert G, Nakagawa T, et al. 2011. The Clavata2 genes of pea and Lotus japonicus  
affect autoregulation of nodulation. The Plant Journal 65: 861–871. 

Laguerre G, Bardin M, Amarger N. 1993. Isolation from soil of symbiotic and nonsymbiotic 
Rhizobium leguminosarum by DNA hybridization. Canadian Journal of Microbiology 39: 1142– 
1149. 

28 

 

Larimer AL, Bever JD, Clay K. 2010. The interactive effects of plant microbial symbionts: a  
review and meta-analysis. Symbiosis 51: 139–148. 

Lau JA, Bowling EJ, Gentry LE, Glasser PA, Monarch EA, Olesen WM, Waxmonsky J, 
Young RT. 2012. Direct and interactive effects of light and nutrients on the legume-rhizobia  
mutualism. Acta Oecologica 39: 80–86. 

Leigh JA, Signer ER, Walker GC. 1985. Exopolysaccharide-deficient mutants of Rhizobium 
meliloti that form ineffective nodules. Proceedings of the National Academy of Sciences of the  
United States of America 82: 6231–5. 

Lévy J, Bres C, Geurts R, Chalhoub B, Kulikova O, Duc G, Journet E-P, Ané J-M, Lauber 
E, Bisseling T, et al. 2004. A putative Ca2+ and calmodulin-dependent protein kinase required  
for bacterial and fungal symbioses. Science (New York, N.Y.) 303: 1361–4. 

Li B, Li Y-Y, Wu H-M, Zhang F-F, Li C-J, Li X-X, Lambers H, Li L. 2016. Root exudates 
drive interspecific facilitation by enhancing nodulation and N2 fixation. Proceedings of the  
National Academy of Sciences of the United States of America 113: 6496–501. 

Lira MA, Nascimento LRS, Fracetto GGM. 2015. Legume-rhizobia signal exchange:  
promiscuity and environmental effects. Frontiers in Microbiology 6: 945. 

Lodwig EM, Hosie AHF, Bourdès A, Findlay K, Allaway D, Karunakaran R, Downie JA, 
Poole PS. 2003. Amino-acid cycling drives nitrogen fixation in the legume–Rhizobium  
symbiosis. Nature 422: 722–726. 

Lodwig EM, Leonard M, Marroqui S, Wheeler TR, Findlay K, Downie JA, Poole PS. 2005. 
Role of Polyhydroxybutyrate and Glycogen as Carbon Storage Compounds in Pea and Bean  
Bacteroids. Molecular Plant-Microbe Interactions 18: 67–74. 

Lodwig E, Poole P. 2003. Metabolism of Rhizobium Bacteroids. Critical Reviews in Plant  
Sciences 22: 37–78. 

Long SR. 1996. Rhizobium symbiosis: nod factors in perspective. The Plant cell 8: 1885–98. 

MacLean AM, Finan TM, Sadowsky MJ. 2007. Genomes of the symbiotic nitrogen-fixing  
bacteria of legumes. Plant physiology 144: 615–22. 

Madsen EB, Madsen LH, Radutoiu S, Olbryt M, Rakwalska M, Szczyglowski K, Sato S, 
Kaneko T, Tabata S, Sandal N, et al. 2003. A receptor kinase gene of the LysM type is  
involved in legumeperception of rhizobial signals. Nature 425: 637–640. 

Magori S, Oka-Kira E, Shibata S, Umehara Y, Kouchi H, Hase Y, Tanaka A, Sato S, 
Tabata S, Kawaguchi M. 2009. TOO MUCH LOVE , a Root Regulator Associated with the 
Long-Distance Control of Nodulation in Lotus japonicus. Molecular Plant-Microbe Interactions  
22: 259–268. 

29 

 

Maimaiti J, Zhang Y, Yang J, Cen Y-P, Layzell DB, Peoples M, Dong Z. 2007. Isolation and 
characterization of hydrogen-oxidizing bacteria induced following exposure of soil to hydrogen  
gas and their impact on plant growth. Environmental Microbiology 9: 435–444. 

Masson-Boivin C, Giraud E, Perret X, Batut J. 2009. Establishing nitrogen-fixing symbiosis  
with legumes: how many rhizobium recipes? Trends in Microbiology 17: 458–466. 

Maymon M, Martínez-Hidalgo P, Tran SS, Ice T, Craemer K, Anbarchian T, Sung T, 
Hwang LH, Chou M, Fujishige NA, et al. 2015. Mining the phytomicrobiome to understand  
how bacterial coinoculations enhance plant growth. Frontiers in Plant Science 6: 784. 

Meixner C, Ludwig-Müller J, Miersch O, Gresshoff P, Staehelin C, Vierheilig H. 2005. 
Lack of mycorrhizal autoregulation and phytohormonal changes in the supernodulating soybean  
mutant nts1007. Planta 222: 709–715. 

Mergaert P, Uchiumi T, Alunni B, Evanno G, Cheron A, Catrice O, Mausset A-E, Barloy-
Hubler F, Galibert F, Kondorosi A, et al. 2006. Eukaryotic control on bacterial cell cycle and 
differentiation in the Rhizobium-legume symbiosis. Proceedings of the National Academy of  
Sciences 103: 5230–5235. 

Miao J, Zhang N, Liu H, Wang H, Zhong Z, Zhu J. 2018. Soil commensal rhizobia promote 
Rhizobium etli nodulation efficiency through CinR-mediated quorum sensing. Archives of  
Microbiology: 1–10. 

Miller AJ, Cramer MD. 2005. Root nitrogen acquisition and assimilation. In: Springer,  
Dordrecht, 1–36. 

Miller LD, Yost CK, Hynes MF, Alexandre G. 2007. The major chemotaxis gene cluster of 
Rhizobium leguminosarum bv. viciae is essential for competitive nodulation. Molecular  
Microbiology 63: 348–362. 

Mishra PK, Mishra S, Selvakumar G, Kundu S, Shankar Gupta H. 2009. Enhanced soybean 
( Glycine max L.) plant growth and nodulation by Bradyrhizobium japonicum- SB1 in presence 
of Bacillus thuringiensis- KR1. Acta Agriculturae Scandinavica, Section B - Plant Soil Science  
59: 189–196. 

Mitra RM, Long SR. 2004. Plant and bacterial symbiotic mutants define three transcriptionally 
distinct stages in the development of the Medicago truncatula/Sinorhizobium meliloti symbiosis.  
Plant physiology 134: 595–604. 

Miyazawa H, Oka-Kira E, Sato N, Takahashi H, Wu G-J, Sato S, Hayashi M, Betsuyaku S, 
Nakazono M, Tabata S, et al. 2010. The receptor-like kinase KLAVIER mediates systemic 
regulation of nodulation and non-symbiotic shoot development in Lotus japonicus. Development  
(Cambridge, England) 137: 4317–25. 

Mohd-Radzman NA, Djordjevic MA, Imin N. 2013. Nitrogen modulation of legume root 

30 

 

architecture signaling pathways involves phytohormones and small regulatory molecules.  
Frontiers in Plant Science 4: 385. 

Mortier V, Holsters M, Goormachtig S. 2012. Never too many? How legumes control nodule  
numbers. Plant, Cell & Environment 35: 245–258. 

Mus F, Crook MB, Garcia K, Costas AG, Geddes BA, Kouri ED, Paramasivan P, Ryu MH, 
Oldroyd GED, Poole PS, et al. 2016. Symbiotic nitrogen fixation and the challenges to its  
extension to nonlegumes. Applied and Environmental Microbiology 82: 3698–3710. 

Nagata M, Yamamoto N, Shigeyama T, Terasawa Y, Anai T, Sakai T, Inada S, Arima S, 
Hashiguchi M, Akashi R, et al. 2015. Red/Far Red Light Controls Arbuscular Mycorrhizal 
Colonization via Jasmonic Acid and Strigolactone Signaling. Plant and Cell Physiology 56:  
pcv135. 

Näsholm T, Kielland K, Ganeteg U. 2009. Uptake of organic nitrogen by plants. New  
Phytologist 182: 31–48. 

Nishida H, Handa Y, Tanaka S, Suzaki T, Kawaguchi M. 2016. Expression of the CLE-RS3  
gene suppresses root nodulation in Lotus japonicus. Journal of Plant Research 129: 909–919. 

Nishida H, Suzaki T. 2018. Nitrate-mediated control of root nodule symbiosis. Current Opinion  
in Plant Biology 44: 129–136. 

van Noorden G, Verbeek R, Dinh Q, Jin J, Green A, Ng J, Mathesius U. 2016. Molecular 
Signals Controlling the Inhibition of Nodulation by Nitrate in Medicago truncatula. International  
Journal of Molecular Sciences 17: 1060. 

Okamoto S, Kawaguchi M. 2015. Shoot HAR1 mediates nitrate inhibition of nodulation in  
Lotus japonicus. Plant Signaling & Behavior 10: e1000138. 

Okamoto S, Shinohara H, Mori T, Matsubayashi Y, Kawaguchi M. 2013. Root-derived CLE 
glycopeptides control nodulation by direct binding to HAR1 receptor kinase. Nature  
Communications 4: 2191. 

Oldroyd GED. 2013. Speak, friend, and enter: signalling systems that promote beneficial  
symbiotic associations in plants. Nature reviews. Microbiology 11: 252–63. 

Oono R, Denison RF. 2010. Comparing symbiotic efficiency between swollen versus  
nonswollen rhizobial bacteroids. Plant physiology 154: 1541–8. 

Oono R, Denison RF, Kiers ET. 2009. Controlling the reproductive fate of rhizobia: how  
universal are legume sanctions? The New phytologist 183: 967–79. 

Op den Camp R, Streng A, De Mita S, Cao Q, Polone E, Liu W, Ammiraju JSS, Kudrna D, 
Wing R, Untergasser A, et al. 2011. LysM-type mycorrhizal receptor recruited for rhizobium  

31 

 

symbiosis in nonlegume Parasponia. Science (New York, N.Y.) 331: 909–12. 

Ossler JN, Heath KD. 2018. Shared Genes but Not Shared Genetic Variation: Legume  
Colonization by Two Belowground Symbionts. The American Naturalist: 000–000. 

Ossler JN, Zielinski CA, Heath KD. 2015. Tripartite mutualism: facilitation or trade-offs 
between rhizobial and mycorrhizal symbionts of legume hosts. American journal of botany 102:  
1332–41. 

Ott T, van Dongen JT, Gu¨nther C, Krusell L, Desbrosses G, Vigeolas H, Bock V, 
Czechowski T, Geigenberger P, Udvardi MK. 2005. Symbiotic Leghemoglobins Are Crucial 
for Nitrogen Fixation in Legume Root Nodules but Not for General Plant Growth and  
Development. Current Biology 15: 531–535. 

Paau AS, Bloch CB, Brill WJ. 1980. Developmental fate of Rhizobium meliloti bacteroids in  
alfalfa nodules. Journal of bacteriology 143: 1480–90. 

Padje A van’t, Whiteside MD, Kiers ET. 2016. Signals and cues in the evolution of plant– 
microbe communication. Current Opinion in Plant Biology 32: 47–52. 

Pahua VJ, Stokes PJN, Hollowell AC, Regus JU, Gano-Cohen KA, Wendlandt CE, Quides 
KW, Lyu JY, Sachs JL. 2018. Fitness variation among host species and the paradox of  
ineffective rhizobia. Journal of Evolutionary Biology. 

Parker MA. 1999. Mutualism in Metapopulations of Legumes and Rhizobia. The American  
naturalist 153: S48–S60. 

Parker MA. 2015. The Spread of Bradyrhizobium Lineages Across Host Legume Clades: from  
Abarema to Zygia. Microbial Ecology 69: 630–640. 

Parniske M. 2008. Arbuscular mycorrhiza: the mother of plant root endosymbioses. Nature  
reviews. Microbiology 6: 763–75. 

Peck MC, Fisher RF, Long SR. 2006. Diverse flavonoids stimulate NodD1 binding to nod gene  
promoters in Sinorhizobium meliloti. Journal of bacteriology 188: 5417–27. 

Peix A, Ramírez-Bahena MH, Velázquez E, Bedmar EJ. 2015. Bacterial Associations with  
Legumes. Critical Reviews in Plant Sciences 34: 17–42. 

Pivato B, Offre P, Marchelli S, Barbonaglia B, Mougel C, Lemanceau P, Berta G. 2009. 
Bacterial effects on arbuscular mycorrhizal fungi and mycorrhiza development as influenced by  
the bacteria, fungi, and host plant. Mycorrhiza 19: 81–90. 

Poole P, Allaway D. 2000. Carbon and nitrogen metabolism in Rhizobium. Advances in  
microbial physiology 43: 117–63. 

32 

 

Poole P, Ramachandran V, Terpolilli J. 2018. Rhizobia: from saprophytes to endosymbionts.  
Nature Reviews Microbiology. 

Portales-Reyes C, Van Doornik T, Schultheis EH, Suwa T. 2015. A novel impact of a novel 
weapon: allelochemicals in Alliaria petiolata disrupt the legume-rhizobia mutualism. Biological  
Invasions 17: 2779–2791. 

Prell J, Bourdès A, Kumar S, Lodwig E, Hosie A, Kinghorn S, White J, Poole P. 2010. Role 
of Symbiotic Auxotrophy in the Rhizobium-Legume Symbioses (J Rutherford, Ed.). PLoS ONE  
5: e13933. 

Prell J, Poole P. 2006. Metabolic changes of rhizobia in legume nodules. Trends in  
Microbiology 14: 161–168. 

Radutoiu S, Madsen LH, Madsen EB, Felle HH, Umehara Y, Grønlund M, Sato S, 
Nakamura Y, Tabata S, Sandal N, et al. 2003. Plant recognition of symbiotic bacteria requires  
Regus JU, Gano KA, Hollowell AC, Sachs JL. 2014. Efficiency of partner choice and 
sanctions in Lotus is not altered by nitrogen fertilization. Proceedings. Biological sciences / The  
Royal Society 281: 20132587. 

Regus JU, Gano KA, Hollowell AC, Sofish V, Sachs JL. 2015. Lotus hosts delimit the 
mutualism-parasitism continuum of Bradyrhizobium. Journal of Evolutionary Biology 28: 447– 
456. 

Regus JU, Quides KW, O’Neill MR, Suzuki R, Savory EA, Chang JH, Sachs JL. 2017. Cell 
autonomous sanctions in legumes target ineffective rhizobia in nodules with mixed infections.  
American Journal of Botany 104: 1299–1312. 

Ren H, Gao T, Hu J, Yang G. 2017. The effects of arbuscular mycorrhizal fungi and root  
interaction on the competition between Trifolium repens and Lolium perenne. PeerJ 5: e4183. 

Roche P, Debellé F, Maillet F, Lerouge P, Faucher C, Truchet G, Dénarié J, Promé J-C. 
1991. Molecular basis of symbiotic host specificity in rhizobium meliloti: nodH and nodPQ  
genes encode the sulfation of lipo-oligosaccharide signals. Cell 67: 1131–1143. 

Rodríguez-Echeverría S, Crisóstomo JA, Nabais C, Freitas H. 2009. Belowground mutualists 
and the invasive ability of Acacia longifolia in coastal dunes of Portugal. Biological Invasions  
11: 651–661. 

Sachs JL, Ehinger MO, Simms EL. 2010a. Origins of cheating and loss of symbiosis in wild  
Bradyrhizobium. Journal of Evolutionary Biology 23: 1075–1089. 

Sachs JL, Mueller UG, Wilcox TP, Bull JJ. 2004. The Evolution of Cooperation. The  
Quarterly Review of Biology 79: 135–160. 

Sachs JL, Russell JE, Lii YE, Black KC, Lopez G, Patil AS. 2010b. Host control over  

33 

 

infection and proliferation of a cheater symbiont. Journal of evolutionary biology 23: 1919–27. 

Sachs JL, Skophammer RG, Regus JU. 2011. Evolutionary transitions in bacterial symbiosis.  
Sakamoto K, Ogiwara N, Kaji T. 2013. Involvement of autoregulation in the interaction 
between rhizobial nodulation and AM fungal colonization in soybean roots. Biology and Fertility  
of Soils 49: 1141–1152. 

Sasaki T, Suzaki T, Soyano T, Kojima M, Sakakibara H, Kawaguchi M. 2014. Shoot- 
derived cytokinins systemically regulate root nodulation. Nature Communications 5: 4983. 

Schnabel E, Journet E-P, De Carvalho-Niebel F, Duc G, Frugoli J. The Medicago truncatula 
SUNN gene encodes a CLV1-like leucine-rich repeat receptor kinase that regulates nodule  
number and root length. 

Shamseldin A, Abdelkhalek A, Sadowsky MJ. 2017. Recent changes to the classification of  
symbiotic, nitrogen-fixing, legume-associating bacteria: a review. Symbiosis 71: 91–109. 

Shigeyama T, Tominaga A, Arima S, Sakai T, Inada S, Jikumaru Y, Kamiya Y, Uchiumi T, 
Abe M, Hashiguchi M, et al. 2012. Additional cause for reduced JA-Ile in the root of a Lotus  
japonicus phyB mutant. Plant Signaling & Behavior 7: 746–748. 

Silsbury J, Catchpoole D, Wallace W, Silsbury J, Catchpoole D, Wallace W. 1986. Effects 
of Nitrate and Ammonium on Nitrogenase (C 2 H 2 Reduction) Activity of Swards of  
Subterranean Clover, Trifolium subterraneum L. Australian Journal of Plant Physiology 13: 257. 

Simms EL, Taylor DL. 2002. Partner choice in nitrogen-fixation mutualisms of legumes and  
rhizobia. Integrative and comparative biology 42: 369–80. 

Simonsen AK, Stinchcombe JR, Bronstein J, Boucher D, James S, Keeler K, Bronstein J, 
Sachs J, Simms E, Trivers R, et al. 2014. Standing genetic variation in host preference for 
mutualist microbial symbionts. Proceedings. Biological sciences / The Royal Society 281: 214– 
217. 

Simsek S, Ojanen-Reuhs T, Stephens SB, Reuhs BL. 2007. Strain-ecotype specificity in 
Sinorhizobium meliloti-Medicago truncatula symbiosis is correlated to succinoglycan  
oligosaccharide structure. Journal of bacteriology 189: 7733–40. 

Smith LM, Reynolds HL. 2015. Plant–soil feedbacks shift from negative to positive with  
decreasing light in forest understory species. Ecology 96: 2523–2532. 

Soyano T, Shimoda Y, Hayashi M. 2015. NODULE INCEPTION Antagonistically Regulates  
Gene Expression with Nitrate in Lotus japonicus. Plant and Cell Physiology 56: 368–376. 

Steidinger BS, Bever JD. 2016. Host discrimination in modular mutualisms: a theoretical 
framework for meta-populations of mutualists and exploiters. Proceedings. Biological sciences  
283: 20152428. 

34 

 

Sturz A V., Christie BR, Matheson BG, Nowak J. 1997. Biodiversity of endophytic bacteria 
which colonize red clover nodules, roots, stems and foliage and their influence on host growth.  
Biology and Fertility of Soils 25: 13–19. 

Suzaki T, Yoro E, Kawaguchi M. 2015. Chapter Three – Leguminous Plants: Inventors of Root 
Nodules to Accommodate Symbiotic Bacteria. International Review of Cell and Molecular  
Biology 316: 111–158. 

Suzuki A, Suriyagoda L, Shigeyama T, Tominaga A, Sasaki M, Hiratsuka Y, Yoshinaga A, 
Arima S, Agarie S, Sakai T, et al. 2011. Lotus japonicus nodulation is photomorphogenetically 
controlled by sensing the red/far red (R/FR) ratio through jasmonic acid (JA) signaling.  
Proceedings of the National Academy of Sciences 108: 16837–16842. 

Takahara M, Magori S, Soyano T, Okamoto S, Yoshida C, Yano K, Sato S, Tabata S, 
Yamaguchi K, Shigenobu S, et al. 2013. TOO MUCH LOVE, a Novel Kelch Repeat-
Containing F-box Protein, Functions in the Long-Distance Regulation of the Legume– 
Rhizobium Symbiosis. Plant and Cell Physiology 54: 433–447. 

Temperton VM, Mwangi PN, Scherer-Lorenzen M, Schmid B, Buchmann N. 2007. Positive 
interactions between nitrogen-fixing legumes and four different neighbouring species in a  
biodiversity experiment. Oecologia 151: 190–205. 

Terpolilli JJ, Hood GA, Poole PS. 2012. Chapter 5 – What Determines the Efficiency of N2- 
Fixing Rhizobium-Legume Symbioses? In: Advances in Microbial Physiology. 325–389. 

Thrall PH, Burdon JJ, Woods MJ. 2000. Variation in the effectiveness of symbiotic 
associations between native rhizobia and temperate Australian legumes: interactions within and  
between genera. Journal of Applied Ecology 37: 52–65. 

Tkacz A, Cheema J, Chandra G, Grant A, Poole PS. 2015. Stability and succession of the 
rhizosphere microbiota depends upon plant type and soil composition. The ISME Journal 9:  
2349–2359. 

Tombolini R, Nuti MP. 1989. Poly(β-hydroxyalkanoate) biosynthesis and accumulation by  
different Rhizobium species. FEMS Microbiology Letters 60: 299–304. 

Trainer MA, Charles TC. 2006. The role of PHB metabolism in the symbiosis of rhizobia with  
legumes. Applied Microbiology and Biotechnology 71: 377–386. 

Trinick MJ. 1973. Symbiosis between Rhizobium and the Non-legume, Trema aspera. Nature  
244: 459–460. 

Turner TR, James EK, Poole PS, Gilbert J, Meyer F, Jansson J, Gordon J, Pace N, Tiedje  
J, Ley R, et al. 2013. The plant microbiome. Genome Biology 14: 209. 

Udvardi MK, Day DA. 1997. METABOLITE TRANSPORT ACROSS SYMBIOTIC 

35 

 

MEMBRANES OF LEGUME NODULES. Annual Review of Plant Physiology and Plant  
Molecular Biology 48: 493–523. 

Udvardi M, Poole PS. 2013. Transport and metabolism in legume-rhizobia symbioses. Annual  
review of plant biology 64: 781–805. 

VanInsberghe D, Maas KR, Cardenas E, Strachan CR, Hallam SJ, Mohn WW. 2015. Non-
symbiotic Bradyrhizobium ecotypes dominate North American forest soils. The ISME Journal 9:  
2435–2441. 

Van de Velde W, Zehirov G, Szatmari A, Debreczeny M, Ishihara H, Kevei Z, Farkas A, 
Mikulass K, Nagy A, Tiricz H, et al. 2010. Plant peptides govern terminal differentiation of  
bacteria in symbiosis. Science (New York, N.Y.) 327: 1122–6. 

Vierheilig H. 2004. Further root colonization by arbuscular mycorrhizal fungi in already 
mycorrhizal plants is suppressed after a critical level of root colonization. Journal of Plant  
Physiology 161: 339–341. 

Voisin AS, Salon C, Jeudy C, Warembourg FR. 2003. Root and Nodule Growth in Pisum 
sativum L. in Relation to Photosynthesis: Analysis Using 13C-labelling. Annals of Botany 92:  
557–563. 

Voisin A-S, Salon C, Munier-Jolain NG, Ney B. 2002. Effect of mineral nitrogen on nitrogen 
nutrition and biomass partitioning between the shoot and roots of pea (Pisum sativum L.). Plant  
and Soil 242: 251–262. 

Wang X, Pan Q, Chen F, Yan X, Liao H. 2011. Effects of co-inoculation with arbuscular 
mycorrhizal fungi and rhizobia on soybean growth as related to root architecture and availability  
of N and P. Mycorrhiza 21: 173–181. 

Weber J, Tham FY, Galiana A, Prin Y, Ducousso M, Lee SK. 2007. Effects of nitrogen 
source on the growth and nodulation of Acacia mangium in aeroponic culture. Journal of  
Tropical Forest Science 19: 103–112. 

Werner GDA, Cornwell WK, Cornelissen JHC, Kiers ET. 2015. Evolutionary signals of 
symbiotic persistence in the legume-rhizobia mutualism. Proceedings of the National Academy  
of Sciences of the United States of America 112: 10262–9. 

West SA, Kiers ET, Simms EL, Denison RF. 2002. Sanctions and mutualism stability: why do  
rhizobia fix nitrogen? Proceedings. Biological sciences / The Royal Society 269: 685–94. 

Westhoek A, Field E, Rheling F, Mulley G, Webb I, Poole P, Turnbull L. 2017. Policing the  
legume-Rhizobium symbiosis: a critical test of partner choice. Scientific Reports 7: 1419. 

Wilkinson HH, Parker MA. 1996. Symbiotic specialization and the potential for genotypic  
coexistence in a plant-bacterial mutualism. Oecologia 108: 361–367. 

36 

 

Wopereis J, Pajuelo E, Dazzo FB, Jiang Q, Gresshoff PM, de Bruijn FJ, Stougaard J, 
Szczyglowski K. 2000. Short root mutant of Lotus japonicus with a dramatically altered  
symbiotic phenotype. The Plant Journal 23: 97–114. 

Yamamura N. 1996. Evolution of mutualistic symbiosis: A differential equation model.  
Researches on Population Ecology 38: 211–218. 

Yates RJ, Howieson JG, Real D, Reeve WG, Vivas-Marfisi A, O’Hara GW. 2005. Evidence 
of selection for effective nodulation in the Trifolium spp. saymbiosis with Rhizobium  
leguminosarum biovar trifolii. Australian Journal of Experimental Agriculture 45: 189. 

Yates RJ, Howieson JG, Reeve WG, Brau L, Speijers J, Nandasena K, Real D, Sezmis E, 
O’Hara GW. 2008. Host–strain mediated selection for an effective nitrogen-fixing symbiosis 
between Trifolium spp. and Rhizobium leguminosarum biovar trifolii. Soil Biology and  
Biochemistry 40: 822–833. 

Yurgel SN, Kahn ML. 2004. Dicarboxylate transport by rhizobia. FEMS Microbiology Reviews  
28: 489–501. 

Zavala JA, Mazza CA, Dillon FM, Chludil HD, Ballare CL. 2015. Soybean resistance to stink 
bugs ( N ezara viridula and P iezodorus guildinii ) increases with exposure to solar UV-B 
radiation and correlates with isoflavonoid content in pods under field conditions. Plant, Cell &  
Environment 38: 920–928. 

Zgadzaj R, Garrido-Oter R, Jensen DB, Koprivova A, Schulze-Lefert P, Radutoiu S. 2016. 
Root nodule symbiosis in Lotus japonicus drives the establishment of distinctive rhizosphere, 
root, and nodule bacterial communities. Proceedings of the National Academy of Sciences of the  
United States of America 113: E7996–E8005. 

Zhang J, Subramanian S, Stacey G, Yu O. 2009. Flavones and flavonols play distinct critical 
roles during nodulation of Medicago truncatula by Sinorhizobium meliloti. The Plant Journal  
57: 171–183. 

Zheng C, Ji B, Zhang J, Zhang F, Bever JD. 2015. Shading decreases plant carbon preferential  
allocation towards the most beneficial mycorrhizal mutualist. The New phytologist 205: 361–8. 

 

 

 

 

 

37 

 

CHAPTER 2 

 

UNFAIR TRADE UNDERGROUND REVEALED BY INTEGRATING DATA WITH NASH  

BARGAINING MODELS 

 

Preface 

This project began as a group proposal written for PLB801 in Fall 2013. Our original 

group included myself, Teresa Clark (TC), Chad Zirbel, Klara Scharnagl, and Pengfei Cao. We 

developed a proposal that united our interests, spanned scales from molecular biology and 

ecology, and included a modeling component. Maren Friesen (MF) and Yair Shachar-Hill (YSH) 

were interested in the ideas we proposed, and TC and I decided to move forward with a subset of 

the proposal, focusing on testing and modeling carbon-nitrogen trade in legume-rhizobia 

mutualisms.  We participated in regular meetings with MF, YSH, and Emily Grman (who 

created the model on which this project is based) to conceptualize the questions we were asking 

and the goals of the study. I led the growth system optimization for our system, and we chose to 

measure 4- to 6-week-old uninoculated and nodulated plants grown with 8, 24, 40, or 80 mg/L N. 

These conditions resulted in reproducible growth rates, large effects on growth and nodulation, 

and relatively steady soil nitrogen availability. 

Once the growth conditions were established, we conducted a number of experiments to 

collect the biomass, C:N elemental compositions, and photosynthesis data required to 

parameterize the model. TC and I were equally involved in growing the experimental plants and 

measuring the model parameters. I also led measurements (e.g., root scans to analyze root 

architecture) that were less connected to the model and thus were not included in this manuscript. 

When the data had been collected, I was responsible for statistical analyses and visualization of 

38 

 

the data, while TC carried out the modeling work. TC and I co-wrote the many drafts of the 

manuscript found here. The content of the text and figures was based on the consensus decisions 

of the larger group.  

 

Abstract 

 Mutually beneficial resource exchange is foundational to global biogeochemical cycles 

and plant and animal nutrition. However, there is inherent potential conflict in mutualisms, as 

each organism benefits more when the exchange ratio (“price”) minimizes its own costs and 

maximizes its benefits. Understanding the bargaining power that each partner has in these 

interactions is key to our ability to predict the exchange ratio and thus the functionality of the 

cell, organism, community, and ecosystem. We tested whether partners have symmetric (“fair”) 

or asymmetric (“unfair”) bargaining power in the legume-rhizobia nitrogen fixing symbiosis 

using measurements of carbon and nitrogen dynamics in a mathematical modeling framework 

derived from economic theory. A model of symmetric bargaining power was not consistent with 

our data. Instead, our data indicate that the growth benefit to the plant has greater weight in 

determining trade dynamics than the benefit to the bacteria. Quantitative estimates of the relative 

power of the plant reveal that the plant’s influence rises as soil nitrogen availability decreases 

and trade benefits to both partners increase. Our finding that legumes have more bargaining 

power than rhizobia at lower nitrogen availabilities highlights the importance of context-

dependence for the evolution of mutualism with increasing nutrient deposition. 

 

 

 

39 

 

Introduction 

Mutualistic relationships abound in nature. They are rooted in the exchange of resources 

or services between different partners whose distinct capabilities allow them to perform better 

together than either could alone. Mutualisms involving the exchange of carbon for mineral 

nutrients between plants and microbes are ancient interactions that have shaped the evolution of 

land plants and play central roles in ecosystem functioning worldwide (Bronstein, 2015). Plants 

participating in these nutritional mutualisms with microbes such as mycorrhizal fungi or 

nitrogen-fixing rhizobium prokaryotes must optimize their allocation of photosynthate between 

taking up nutrients directly and trading for them with mutualists (Bloom et al., 1985). Indeed, 

plants exhibit considerable plasticity in partitioning carbon among shoots, roots, and mutualistic 

partners in response to environmental cues (Harris et al., 1985; Wang et al., 2011). This optimal 

allocation is determined by soil nutrient availability and the cost:benefit ratio of acquiring the 

nutrient through trade. However, it is in each partner’s best interest to influence the carbon-for-

nutrient exchange ratio (“price”) to maximize the benefit to itself (Akçay & Roughgarden, 2007; 

Grman et al., 2012), conditions that should lead to a power struggle over the price. Considerable 

effort has been devoted to applying economic principles to analyzing nutrient exchange, stability, 

and other aspects of mutualisms (Weyl et al., 2010; Werner et al., 2014; Clark et al., 2017), but 

there is a major gap: we do not understand how the exchange ratio and the quantity traded 

between plant and microbe are determined.  

This question has been explored using mathematical models in which partners have 

disparate abilities to acquire resources and divergent resource requirements (Akçay & 

Roughgarden, 2007; Grman et al., 2012; Franklin et al., 2014). In these models, mutualistic 

partners negotiate based on the principles of the Nash bargaining solution, an axiomatically 

40 

 

derived result describing the expected distribution of benefits after bargaining between self-

interested partners that are able to regulate their participation in trade in response to the benefits 

they receive from trade (Nash, 1950; Binmore et al., 1986; Akçay & Roughgarden, 2007).  

A central assumption underlying the Nash bargaining solution is symmetry in bargaining 

power between the partners, where the bargaining power of a partner is defined as the weight 

given to the benefit received by that partner in the determination of trade dynamics. The best 

indicator of symbiotic benefit is reproductive fitness in the field, but biomass is conventionally 

used as a proxy for fitness (Younginger at el 2017). Consequently, the symmetric Nash product 

is the product of the partners’ growth gains from trade: 

 

 

    (gPtrade – gPnotrade) (gRtrade – gRnoTrade) 

 

 

  (1) 

where gPtrade and gRtrade are the plant and rhizobial growth rates with trade, and gPnoTrade and 

gRnoTrade are the respective growth rates without trade. An extension of this framework allows for 

unequal power between partners through the asymmetric Nash product: 

  

  (gPtrade – gPnotrade)β (gRtrade – gRnoTrade)1-β 

 

 

(2) 

  

which arises when bargaining power differs between partners (Binmore et al., 1986). β is a 

scaling exponent that assigns different weights to the gains from trade by the plant and microbe. 

Increases in β correlate with increases in plant bargaining power relative to the microbial 

symbiont, and bargaining is symmetrical when β = 0.5 (Binmore et al., 1986).  

 

Previous modeling analyses of nutrient exchange symbioses have been limited by the 

41 

 

absence of quantitative experimental studies in which all the major relevant parameters were 

measured in a single study across a range of environmental conditions and used explicitly within 

a mathematical framework (Clark et al., 2017). To address this knowledge gap, we measured 

biomass distributions, nitrogen uptake and exchange rates, photosynthetic carbon assimilation 

fluxes, and carbon and nitrogen compositions in the model Medicago truncatula-Ensifer medicae 

(legume-rhizobia) symbiosis under conditions ranging from low to high nitrogen availability. We 

used the measurements to quantitatively test whether trade in a mutualism follows the 

predictions of the Nash bargaining solution under different conditions of resource availability. 

To do this, we refined and parameterized a mechanistic model of resource trade between a plant 

and microbe (Figure 2.1; Grman et al., 2012). This model assumes that the growth of each 

partner is limited by its ability to obtain carbon and/or a mineral nutrient and determines the 

exchange ratio to be the one that maximizes the product of partner benefits from trade, consistent 

with the Nash bargaining solution (Nash, 1950; Akçay & Roughgarden, 2007; Grman et al., 

2012). The exchange ratio is then used to predict per capita partner growth rates and allocation to 

growth versus trade. We tested the assumption of symmetric bargaining by comparing the 

accuracy of predictions made by the model with experimental measurements. The experimental 

results were inconsistent with predictions based on the symmetric Nash bargaining solution, so 

we explored how predictions based on asymmetries in relative bargaining power aligned with 

experimental measurements and determined the value of β for which model fit was greatest.  

 

 

 

 

42 

 

 

Figure 2.1 | Pictorial representation of the legume-rhizobia nutrient exchange model. Plant 
and rhizobial growth are limited by the ability to obtain carbon (C) or nitrogen (N). In general, 
the growth rates are the rates of biomass carbon increase that complement via yield parameters 
the net amount of carbon or nitrogen obtained directly and/or from trade. Carbon and/or nitrogen 
can be lost during trade or respiration. More information about model construction is provided in 
the Methods. 
 

 

 

 

 

 

43 

 

Materials and Methods 

Experimental methods 

SC10 Cone-Tainer pots (Steuwe and Sons Inc., Corvallis, OR, USA) were plugged with 

⅜” diameter cotton wicks leading to opaque 50 mL reservoirs and filled with medium grain 

vermiculite. Pots were wetted with 25 mL deionized water, covered, and autoclaved for 45 

minutes. After 24 hours, pots were wetted with a further 25 mL of deionized water and 

autoclaved again for 45 minutes. 24 hours later, the pots were wetted with 25 mL of Fahraeus 

nutrient solution (Fahraeus, 1957) supplemented with 8, 24, 40, or 80 mg L-1 N in the form of 

NH4NO3.  

 

Medicago truncatula A17 seeds (Young et al., 2011) were scarified with 600 grit 

sandpaper, sterilized in commercial bleach (8.25% NaHClPO3) for 3 minutes, and rinsed at least 

6 times with sterile deionized water. Following 3 hours of incubation in sterile deionized water at 

room temperature, the seeds were re-sterilized in 0.825% NaHClPO3 for 30 seconds and rinsed at 

least 6 times with sterile deionized water. Seeds were then incubated in sterile deionized water 

for 48 hours at 4°C. The water was replaced approximately every twelve hours during this 

incubation. The seedlings were then transferred to sterile petri dishes, sealed with Parafilm, and 

germinated at room temperature for 48 hours. Seedlings with 1 cm or longer radicles were 

aseptically transplanted into prepared pots. After planting, the plants were fully randomized and 

grown at 22°C with a 16 hour day/8 hour night cycle at approximately 250 µmol m-2 s-1. 

Seedlings were misted daily with sterile deionized water for the first week. Throughout the 

growth period, plants were continuously supplied via their reservoirs with Fahraeus nutrient 

solution.  

44 

 

 

Ensifer medicae WSM419 (Reeve et al., 2010) was grown for 48 hours in tryptone-yeast 

broth at 30°C with rotary shaking at 200 RPM. The OD600 of the culture was measured to 

estimate cell density. Half of the week-old plants were inoculated with 1 mL of 106 CFU mL-1 

inoculum in ½ x phosphate buffered saline (+ rhizobia treatment), while the other half were 

mock-inoculated with sterile buffer (- rhizobia treatment). At least 7 plants per nitrogen level and 

inoculation status were harvested after 4 or 6 weeks of growth for biomass measurements (in 

total, 64 or 101 plants were harvested after 4 or 6 weeks, respectively, half of which were 

nodulated). Roots and shoots were separated, and roots were carefully washed in deionized water 

to remove vermiculite. Washed roots were checked for nodulation, and nodules were removed 

and counted. All tissue was dried at 60°C for at least one week. Of these plants, 3 plants per 

nitrogen level, inoculation status, and age were analyzed for carbon and nitrogen elemental 

compositions of roots, shoots, and nodules (48 plants in total). Dried plant tissues were ground 

using a NutriBullet, LLC household blender followed by a Retsch MM301 Mixer Mill. 2-5 mg 

dried tissue was weighed, packaged in tin capsules, and analyzed by the Robertson lab at 

Michigan State University’s Kellogg Biological Station using a Costech ECS4010 analyzer. 

To measure direct nitrogen uptake with 15N enrichment, 24 plants (3 per nitrogen level 

and inoculation status) were grown as described above until 4.5 weeks, whereupon the pots and 

reservoirs were flushed with 500 mL N-free Fahraeus solution to remove soluble unlabeled 

nitrogen. The plants were then watered with 25 mL of Fahraeus solution containing the 

appropriate concentration of 15NH4

15NO3, supplied with this solution via their reservoirs for 1 

week and harvested. Plants were harvested, weighed, dried, ground, and packaged into tin 

capsules as described above. The nitrogen content and 15N abundance were then analyzed by the 

Stable Isotope Lab at Utah State University using a Europa Scientific SL-2020 system. 

45 

 

Photosynthesis rates were measured on 5-week-old nodulated and uninoculated plants (4-

6 plants per nitrogen level and inoculation status; 41 plants total) using whole-plant (6”x6”x12”) 

photosynthesis chambers connected to LI-COR LI-6400 apparatuses, illuminated with LED 

lights, and provided a constant airflow of 1000 µmol s-1 with 400 µmol CO2 per mol of air 

(Figure 2.7a,b). CO2 from below ground was excluded using modeling clay. Assay conditions 

(e.g., light, humidity, temperature) matched the growth conditions. Steady-state photosynthetic 

rates were measured for at least 90 minutes after a pre-equilibration period in the chambers with 

the light on for at least 1 hour (Figure 2.7c). The final steady state CO2 assimilation rate (µmol 

CO2 sec-1), was converted to a daily rate (mg C day-1), assuming a constant rate throughout the 

16h light period. After measuring photosynthesis rates, the plants were harvested, dried, and 

weighed as described above. 

To test the effects of soil nitrogen and rhizobia on root and shoot biomass, we used a 

linear model ANOVA with Type II sum of squares (aov and car packages, R 3.3.1) with soil 

nitrogen, rhizobia, and the nitrogen by rhizobia interaction as fixed effects. Since we detected 

significant main effects of nitrogen and rhizobia, we conducted post-hoc testing with the Tukey 

test at a significance level of 0.05 (lsmeans package, R 3.3.1) to determine whether group means 

were significantly different. 

 

Model construction  

The legume-rhizobia model (Figure 2.1) was derived from the model of Grman et al. 

(2012) that assumes plants adjust carbon allocation to roots or shoots on a faster timescale than 

the carbon-for-nutrient exchange ratio is negotiated between the partners. Refinements to model 

structure equations were made using Wolfram Mathematica 11.3. Model predictions include 

46 

 

plant and rhizobial specific growth rates (gP and gR, respectively) as functions of biomass 

carbon gain, the proportion of plant carbon allocated to roots (aNP) or shoots, and the carbon-for-

nitrogen exchange ratio (T). The model assumes organismal growth is limited by the ability to 

obtain nitrogen and/or carbon for biomass production. Consequently, for each partner, the 

predicted growth rate is the minimum growth predicted when nitrogen (gPNlim, gRNlim) or carbon 

(gPClim, gRClim) is limiting: 

 

 

  

gP = min(gPClim, gPNlim) 

gR = min(gRClim, gRNlim). 

When nitrogen is limiting, the organismal growth rates are represented as 

  

gPNlim = (fnp + X/(p T))*ynp 

gRNlim = (fnr – X/(r T))*ynr, 

, 

 where X is the rate of carbon traded from the plant to the rhizobia and, for the plant and rhizobia, 

respectively, p and r are the organismal carbon contents, fnp and fnr are the rates of nitrogen 

uptake per organismal carbon content, and ynp and ynr are the carbon biomass yields per unit 

nitrogen. Briefly in the vernacular, the growth rates are the rates of biomass carbon increase that 

complement (via the yield parameters) the amount of nitrogen obtained directly and/or from 

trade. Comparably, when carbon is limiting, the organismal growth rates are represented as 

  

gPClim = fcp - X/p 

47 

 

gRClim = X/r, 

, 

 where fcp is the rate of photosynthetic carbon uptake per organismal carbon content. Yield 

parameters are not necessary here because the growth rates are in units of biomass carbon gain.  

Respiratory costs associated with nitrogen fixation were added to the model by reducing 

the rhizobial growth rate by the rate of nitrogen fixation (fnf) multiplied by the biochemical 

stoichiometric trade constraints of 2.57 g C g-1 N (Phillips, 1980). Consequently, the rhizobial 

growth rate equations become 

  

gRNlim = (fnr – X/(r T))*ynr – 2.57*fnr 

gRClim = X/r – 2.57*fnr. 

. 

Plant respiratory costs are not as well-defined (Wardlaw, 1990) and thus were incorporated via 

the plant carbon uptake (fcp) measurement as described above. 

  

In addition to serving as model predictions, the growth rate equations provide the 

foundation for how the model predicts the rate of carbon trade, root-to-shoot allocation, and the 

carbon-for-nitrogen exchange ratio. Because rhizobia in nodules are unable to take up external 

carbon, we assumed the modeled trade is rhizobia-limited. Consequently, the rhizobial partner 

trades away all surplus nitrogen (i.e., nitrogen unnecessary for growth) in exchange for carbon 

from the plant. The traded nitrogen and that received by root uptake are used for plant growth 

and thus determine the plant surplus carbon (via the yield parameter ynp) that is traded to the 

rhizobia (i.e., X is the plant surplus carbon). As described in Grman et al. (2012), the model 

predicts that the optimal root-to-shoot allocation is when, after trade, the total carbon uptake by 

48 

 

the shoots complements (via the yield parameter) the total nitrogen uptake by the roots. Partner 

negotiations for the carbon-for-nitrogen exchange ratio are predicted to be consistent with 

economic modeling methods. When bargaining is symmetric, as in the Grman et al. (2012) 

model, the negotiated ratio is assumed to result in the Nash bargaining solution, in which the 

Nash product (Equation 1) is maximized. If bargaining is asymmetric, then negotiations should 

result in maximizing the asymmetric Nash product (Equation 2). 

 

Computational methods 

The model derivation (Grman et al. 2012) does not rely on the assumption of extended 

steady state because the solutions are instantaneous for any given set of input values for the plant 

(Medicago truncatula) and symbiont (Ensifer medicae). However, we first confirmed that over 

the period in which measurements were made (4-6 weeks), the plant per capita growth rate was 

near linear (Figure 2.7), thus its growth and nutrient uptake rates should change slowly compared 

to the negotiations. Measurements were either taken at 5 weeks of age or averaged between 

measurements at 4 and 6 weeks to represent 5-week-old systems. Goodness of fits for model 

predictions and parameters were assessed using 90% confidence intervals from the results of 

modeling 50 pseudo datasets per growth condition and age (Methods S2). The pseudo datasets 

were generated by Monte Carlo sampling of the experimental measurements; i.e., for each 

measurement, we generated random pseudo data points with normal distribution around the 

measured average with the measured standard deviation. To be consistent with biological reality, 

we assumed that biomasses and nutrient uptake rates could not be negative, so any randomly 

generated negative values were rounded up to zero. Each pseudo set contained one point for 

every measurement and were used as model inputs to generate one set of model predictions. This 

49 

 

generated 50 sets of model predictions per nitrogen level, which were assessed using confidence 

intervals. This statistical method allowed us to evaluate how variations in measurements affected 

model predictions.  

Because the model tracked how carbon was obtained and allocated, most model inputs 

(Table 2.1) and predictions (Table 2.2) were expressed per unit of biomass carbon content. 

Carbon and nitrogen contents were calculated using biomass and elemental composition 

measurements of carbon and nitrogen, respectively (Table 2.3-2.4). Both contents were used to 

calculate the carbon per nitrogen organismal yield parameters, and allocation to root biomass 

was calculated as the root carbon content divided by whole plant carbon content. Per capita 

growth rates were calculated as the carbon content gained between 4 and 6 weeks of age, divided 

by 5-week carbon content. For the plant, photosynthesis and soil nitrogen uptake rates were 

expressed per shoot carbon and root carbon, respectively, while rhizobial nitrogen fixation was 

per nodule carbon. Please note: when the abundance of nodule carbon is low (e.g., at 80 mg L-1 

N), this requires dividing by a small number which amplifies the associated uncertainties. 

Nitrogen elemental composition and 15N enrichment measurements were used to 

differentiate between the rates of soil nitrogen uptake and nitrogen trade in nodulated plants. As 

described above, nodulated and uninoculated plants were labeled by flushing the soil with 

nitrogen-free media and then watering for 1 week with 15NH4

15NO3. Labeling revealed that 

96.7% of the variation in nitrogen uptake per total biomass was due to the abundance of nitrogen 

in the nutrient solution (p < 0.001; Figure 2.8a) and at 24, 40, and 80 mg L-1 N, there was no 

significant difference in 15N uptake per total plant biomass between nodulated and uninoculated 

plants. Consequently, at these nitrogen levels, we concluded that the rate of soil nitrogen uptake 

in nodulated plants is equivalent to the rate of biomass nitrogen uptake in uninoculated plants, 

50 

 

and the remaining biomass nitrogen content in nodulated plants can be attributed to trade. When 

analyzing the model with pseudo datasets, this was implemented by generating pseudo data for 

the uninoculated plants as well as nodulated. At 8 mg L-1 N, the average 15N uptake rate was 2.4-

fold greater in uninoculated plants than in nodulated plants. This average difference was 

incorporated into the model analysis by dividing the corresponding (per pseudo replicate) 

uninoculated nitrogen uptake rate by 2.4.  

A minimum of 2.57 g C is biochemically required to produce the ATP and reductant 

necessary for rhizobia to fix 1 g N (Phillips, 1980). Using this biochemical minimum and the 

amount of nitrogen traded to the plant, we obtained a minimum for carbon to be traded for and 

consumed by rhizobial respiration. In addition to respiratory costs, the rhizobia used carbon for 

biomass production. We experimentally derived the quantity of biomass carbon using biomass 

and carbon composition measurements. Consequently, the predicted carbon-for-nitrogen 

exchange ratio was experimentally estimated as the amount of carbon traded to the rhizobia for 

biomass and respiration, divided by the amount of nitrogen traded to the plant. These estimates 

should be regarded as conservative (low) because they do not include any additional respiration 

costs for growth or maintenance; however, we found that modest increases (e.g., 3.6 mg C mg-1 

N, Ryle et al 1984) beyond the minimum had little effect on model predictions. 

Measurements from uninoculated plants were used to test the effect of including plant 

respiration in the model because the presence of nodules prevented accurate measurement of 

below-ground (i.e., root) respiration in nodulated plants. It has been shown that when relative 

plant growth is constant (as in the near linear legume-rhizobia system), the rate of plant 

respiration per unit of biomass is also constant (Lambers et al., 1983). Furthermore, the rate of 

whole-plant respiration has been found to be a linear function of plant biomass and rate of gross 

51 

 

photosynthesis. Therefore, we assumed that the rate of respiration per unit plant biomass was the 

same for plants with and without rhizobia. We found that in uninoculated plants, there was a 

significant correlation (R2 = 0.346, p < 0.001) between biomass carbon content and the 

proportion of photosynthetic carbon used for biomass (Figure 2.8b). We used this relationship to 

estimate the amount of carbon used for plant respiration in nodulated plants. Plant respiratory 

costs were included in the model by reducing the measured photosynthesis rate by the estimated 

respiration rate. 

Overall model fits were quantified using the sum of squared differences between 

predicted and measured allocation to root biomass, plant growth, and rhizobial growth. The 

carbon-for-nitrogen exchange ratio predictions were not used in this assessment because the ratio 

was experimentally estimated (as described above) but not directly measured. The best fit value 

of β was identified as the one yielding the smallest normalized sum of squares. The sum of 

squares for each β in a pseudo dataset was normalized by dividing by the average sum of squares 

for that pseudo dataset. This allowed fit comparisons to be between the β values without bias 

from differences between pseudo datasets. 

 

Results 

Changes in soil nitrogen alter carbon and nitrogen uptake and the benefits of the legume-

rhizobia mutualism 

We determined the values of model input parameters (Table 2.1) for the Medicago 

truncatula-Ensifer medicae (legume-rhizobia) symbiosis by measuring biomasses, nutrient 

uptake rates, and elemental compositions, and of the resulting model predictions, including 

growth rates, root-to-shoot allocation, and the carbon-for-nitrogen exchange ratio (Table 2.2). 

52 

 

The collective nodules of a single plant were used as a proxy for the rhizobial partner because, 

although nodules contain plant cells as well as rhizobia, the whole nodule represents the plant’s 

investment toward trade. In addition, nodule biomass correlates well with rhizobial abundance 

(Ratcliff et al 2011), thus larger nodules indicate greater plant investment as well as greater 

rhizobial benefits from trade. 

Plants were grown in the presence or absence of rhizobia and fertilized with 8, 24, 40, or 

80 mg L-1 N (Figure 2.2a; Valladares et al., 2002). Total plant biomass was significantly 

increased by soil nitrogen (p < 0.001). Inoculation with rhizobia significantly increased total 

plant biomass at 8, 24, and 40 mg L-1 N, but not at 80 mg L-1 N (Figure 2.2b), showing that 

rhizobia enhanced plant growth at low and intermediate nitrogen levels. Nodule biomass 

significantly decreased with increasing soil nitrogen (p < 0.001, Fig 2.2c). Together, these 

findings confirm that mutualistic relationships are more beneficial for both partners at lower 

nitrogen availabilities and that nitrogen is a key limiting resource in these experiments (Regus,  

et al., 2017). 

Plants interacting with rhizobia have two options for acquiring nitrogen: taking it up 

directly from the soil or trading for it with rhizobia. These two nitrogen acquisition routes are 

rarely differentiated experimentally, but this differentiation is essential to quantifying symbiotic 

nutrient exchange. To do so, we measured direct nitrogen uptake using 15N enriched nutrient 

solution and compared it to the total increase in plant nitrogen content. Our findings revealed that 

fixed nitrogen from trade adds to but does not replace nitrogen obtained by direct uptake. On a 

per-plant basis, plants with rhizobia obtained at least as much nitrogen by direct uptake as those 

without rhizobia at all nitrogen levels (Figure 2.3a). Furthermore, in comparison to uninoculated 

plants, the rate of direct nitrogen uptake per root carbon was slightly lower in nodulated plants at 

53 

 

the lowest soil nitrogen, modestly greater at 24 and 40 mg L-1 N, and the same at 80 mg L-1 N 

(Figure 2.4b). 

In addition to increasing plant growth rates and supplementing nitrogen acquisition, the 

presence of rhizobia significantly altered the pattern of carbon allocation within the plant (Voisin 

et al., 2002; Goh et al., 2016). Increasing soil nitrogen significantly decreased root:shoot ratio in 

uncolonized plants (Figure 2.4a; Rufty et al., 1984; Paponov et al., 2000), while the root:shoot 

ratio of nodulated plants did not vary with soil nitrogen (Figure 2.4a). This shows that plants 

adjust their relative allocation of biomass between roots and shoots depending on their ability to 

take up nitrogen directly from the soil or trade for it with rhizobia. However, root:shoot ratios are 

an incomplete measure of carbon allocation because they do not account for allocation to 

mutualists, respiration, exudates, and other carbon sinks (Wardlaw, 1990). To determine how 

organ allocation relates to the total carbon budget, we measured total plant carbon content and 

whole-plant photosynthetic carbon uptake. As expected from their larger shoot biomasses, we 

measured higher rates of carbon uptake per plant in nodulated plants grown at 8, 24, and 40 mg 

L-1 N than in uninoculated plants (Figure 2.3b, Figure 2.4c). Plant respiration ranged from 11-

65% of total photosynthetic carbon, and carbon allocation to nodules ranged from 2-10% (Figure 

2.3b). The magnitude of respiratory costs highlights the importance of accounting for respiration 

in models of plant carbon budgets (Table 2.2); the strong dependence of respiration rates on 

nitrogen levels points to the value of measuring them directly.  

 

54 

 

Figure 2.2 | Both partners benefit more from trade at lower nitrogen availabilities. (a) 
Images of 5-week-old nodulated and uninoculated M. truncatula at 8, 24, 40, and 80 mg L-1 N. 
(b,c) Average 5-week root (b, dark green), shoot (b, light green), and nodule (c) biomasses. Error 
bars indicate standard error (n = 7-15), and bars with the same letter within the same panel do not 
differ significantly after post hoc testing with the Tukey test (p < 0.05). (b) Capital letters refer to 
shoot biomass and lowercase letters refer to root biomass.  

 

55 

 

Figure 2.3 | The effect of rhizobial inoculation on plant carbon and nitrogen budgets as 
functions of nitrogen availability. Whole-plant nitrogen (a) and carbon (b) uptake and 
allocation rates by nodulated (+) and uninoculated (-) M. truncatula at 8, 24, 40, and 80 mg L-1 
N. (a) Direct nitrogen uptake by roots (dark green) and nitrogen received from trade with 
rhizobia (medium green). (b) Photosynthetic carbon allocated to plant biomass (dark green), 
plant respiration (medium green), and nodule growth or respiration (light green). Values and 
error bars represent the average and 90% confidence intervals, respectively, of 50 pseudo 
datasets generated by Monte Carlo sampling. 

 

56 

 

 
Figure 2.4 | Plant carbon allocation to nutrient uptake. Root-to-shoot ratio (a), direct nitrogen 
uptake per root carbon (b), and photosynthesis per shoot carbon (c) of 5-week-old nodulated (+) 
and uninoculated (-) plants at 8, 24, 40, and 80 mg L-1 N. a, Values and error bars represent the 
average and standard error, respectively, of 7-15 biological replicates. Bars with the same letter 
do not differ significantly after post hoc testing with the Tukey test (p > 0.05). b, c, Values and 
error bars represent the average and 90% confidence intervals, respectively, of 50 pseudo 
datasets generated by Monte Carlo sampling. 
 

 

 

57 

 

The benefits and exchange ratio of legume-rhizobia trade are better explained by asymmetric 

bargaining  

We used the measurements described above to predict the carbon-for-nitrogen exchange 

ratio consistent with the Nash bargaining solution in the legume-rhizobium model. Using the 

biochemical stoichiometric trade constraints for nitrogen fixation of at least 2.57 mg C mg-1 N 

(Phillips, 1980) and measured volumes of nitrogen traded and nodule biomass carbon contents 

(Figure 2.2 and 2.3), we estimate that the nodules received 2.72-3.25 mg C mg-1 N for growth 

and respiration (Figure 2.5a). The model-predicted exchange ratios approached these estimates at 

the highest nitrogen level, but at the lowest, the predicted exchange ratio was 2.5-fold higher 

than experimentally estimated (Figure 2.5a).   These estimates are consistent with measurements 

on the soybean-rhizobia mutualism, where nodules received 3.6 mg C mg-1 N (Ryle et al., 1984). 

In addition to comparing experimentally estimated and model-predicted exchange ratios, we also 

examined how the exchange ratio influences partner benefits from trade, and thus how well the 

model predicts partner growth rates and the proportion of plant carbon allocated to roots or 

shoots. Using the predicted exchange ratio, the symmetric bargaining model predicted plant 

growth and carbon allocation to roots within 50% of measured (Figure 2.5b,c), but predicted 

nodule growth to be 3 to 7-fold higher than measured rates (Figure 2.5d). These discrepancies 

could not be resolved by modifications in respiratory cost interpretations, growth rate definitions 

(specific versus organismal), assuming that only 50% of the nodule biomass represents the 

rhizobial partner (Table AI-2), or assumptions concerning the timescale of adjusting root-to-

shoot allocation (Grman et al., 2012), and thus suggest that this legume-rhizobia mutualism fails 

to meet fundamental conditions of the symmetric Nash bargaining solution (Equation 1; Nash, 

1950). 

58 

 

Consequently, we investigated asymmetric bargaining using asymmetric Nash products 

in our model (Equation 2) and found that it could better explain the observed growth rates, 

nutrient allocations and trade across the range of growth conditions. We determined how values 

of β between 0 and 1 affected model fit for the carbon-for-nitrogen exchange ratio and other 

model predictions. We found that across the soil nitrogen levels, the best fit value of β was 0.70 

(Figure 2.6). Using this estimate, model predictions of nodule growth at 8, 24, and 40 mg L-1 N 

were improved up to 2-fold, but there was little improvement in plant growth, root:shoot 

allocation, or 80 mg L-1 N predictions (Figure 2.5). Next, we let β vary among the nitrogen levels 

and found that the best fit value of β increased from 0.57 to 0.86 as soil nitrogen decreased 

(Figure 2.6, Table 2.2). The 90% confidence intervals of β for 8 and 80 mg L-1 N were 

inconsistent with the value of 0.70 obtained assuming a constant value of β across nitrogen 

levels. Importantly, β for 8, 24, and 40 mg L-1 N indicate asymmetric trade (i.e., β ≠ 0.50), and 

model fit was dramatically improved compared to symmetric predictions (Figure 2.5). For 

example, at 8 mg L-1 N, nodule growth was predicted within 3% of measured values (Figure 

2.5d), and overall agreement between predicted and measured growth rate and root:shoot 

allocation predictions was increased 7-fold (Table 2.2). For all soil nitrogen levels, plant growth 

and root allocation predictions were improved, and the exchange ratio was predicted within 25% 

of experimental estimates (Figure 2.5b,c). Together, these findings indicate that plant bargaining 

power rises as its investment in trade and the benefits to both partners increase. 

59 

 

 

Figure 2.5 | Model fit to experimentally estimated data is improved by allowing asymmetric 
bargaining power. Empirically estimated (measured) and model predicted carbon-for-nitrogen 
exchange ratios (a), percent allocations toward roots (b), plant growth rates (c), and 
nodulerhizobial  growth rates (d) of 5-week-old nodulated M. truncatula at 8, 24, 40, and 80 mg 
L-1 N. The symmetric (sym) bargaining predictions correspond to those predicted using 
symmetric Nash products, while the asymmetric (asym) bargaining predictions correspond to 
using asymmetric Nash products with either β fitted across the soil nitrogen levels (average β) or 
β fitted to individual soil nitrogen levels (variable β). The dotted line in (a) represents the 
biochemical minimum value of the exchange ratio. Values and error bars represent the average 
and 90% confidence intervals, respectively, of 50 pseudo datasets generated by Monte Carlo 
sampling. 
 

60 

 

 

Figure 2.6 | Plant bargaining power as a function of nitrogen availability. Best fit average β 
(dotted line; shaded area indicates the 90% confidence interval) and the best fit variable β at 8, 
24, 40, and 80 mg L-1 N (bars; error bars indicate 90% confidence intervals of 50 pseudo 
datasets). Best fit β were those associated with the smallest sum of squared differences between 
predicted (using asymmetric Nash products, Equation 2) and measured allocation to root 
biomass, plant growth, and nodule  growth in 5-week-old nodulated M. truncatula. The solid line 
represents symmetric bargaining.  
 

 

Discussion 

Potential mechanisms underlying asymmetric bargaining 

Our measurements of nutrient uptake, per capita growth, and partner composition in a 

legume-rhizobia system were more consistent with the plant having greater bargaining power 

than its symbiotic partner, than with a model of fair trade. Asymmetries in bargaining power can 

arise by three mechanisms that may be operating in the legume-rhizobia mutualism. First, one 

partner can have a lower effective “discount rate” (the rate at which benefits lose value to that 

partner if negotiations are prolonged), thus conferring greater bargaining power to the partner 

61 

 

able to endure longer negotiations (Kawamori, 2014). In the legume-rhizobium mutualism, the 

plant’s longer lifespan and potentially larger nutrient reserves could serve to provide a lower 

discount rate. Second, group bargaining dynamics can affect bargaining power. In “pure 

bargaining” situations, groups of individuals negotiating as a single partner have less apparent 

bargaining power than independent partners because group benefits would be divided among 

group members after negotiation (Chae & Heidhues, 2001). In this case, if the bacteroids that 

comprise a nodule (Udvardi & Poole, 2013) negotiate as a group (e.g., at the nodule level), then 

the rhizobial partner would have less apparent bargaining power because the carbon received by 

the nodule would be divided among the bacteroids. A third mechanism that can increase an 

individual’s bargaining power is the ability to simultaneously negotiate with multiple trade 

partners (Chakraborty et al., 2009; Chakraborty, 2011). In this case, the presence of multiple 

nodules on a plant and/or multiple bacteroids within a nodule could lead to bargaining power 

asymmetry. This mechanism could be particularly important if bacteroids within nodules bargain 

independently with the plant, which is consistent with the plant being able to interact differently 

with different bacteroids within a nodule (Daubech et al., 2017; Regus, JU et al., 2017). Each of 

these mechanisms can explain why the plant has more bargaining power than the rhizobia, but 

alone, they do not explain why plant bargaining power appears to be higher when nitrogen is 

scarce (e.g., 8 mg L-1 N). At lower soil nitrogen levels, the plant would be expected to have 

fewer nitrogen reserves and nitrogen received from trade comprises a larger proportion of the 

plant’s nitrogen budget (Figure 2.3a), which could increase its discount rate and lower its 

bargaining power. However, we found that as soil nitrogen decreases, total nodule biomass 

(Figure 2.2) increases. This apparent increase in the number of bacteroids (i.e., trade partners) 

could lead to increased plant bargaining power even as the plant’s reliance on the rhizobia 

62 

 

increases. Studies in other systems would be important for assessing the prevalence of 

asymmetric bargaining, and manipulating microbial numbers would allow the influence of 

partner number to be further investigated. 

 

Relating trade conflict to the concept of cheating 

Our finding that the plant host has a high degree of control over the carbon-for-nitrogen 

exchange ratio when soil nitrogen is limiting has important implications for conceptualizing 

conflict and cheating within mutualisms. Cheaters have been the focus of much empirical and 

theoretical work as they have the potential to lead to mutualism collapse, yet there has been 

debate in the literature regarding how to identify cheaters (Frederickson, 2013). A recent 

synthesis defines cheating as increasing one’s own relative fitness while decreasing that of the 

partner (Jones et al., 2015). However, cheating is not synonymous with conflict, and in fact 

Jones et al. (2015) show that under both fitness conflict and fitness alignment cheating genotypes 

may be present within the population. Within the economic model that we use in our study, there 

is a fundamental conflict between plant and symbiont over the exchange ratio (Schwartz & 

Hoeksema, 1998; Grman et al., 2012). Our finding of asymmetric bargaining can be interpreted 

as evidence of this conflict because it demonstrates that partners may not benefit equally from 

trade, even if trade is mutually beneficial. However, given that the exchange ratio seems to be 

determined almost entirely by the plant, this control could in effect force the fitness interests of 

the symbiont to align with the host. Consequently, we predict that with multiple genotypes of 

varying fixation abilities there would be a strong signal of fitness alignment—even in the face of 

underlying conflict. 

 

63 

 

Limitations and future work 

The nutrient-exchange model employed connects nutrient uptake rates to partner growth 

by predicting the allocation strategies that maximize growth, given the organismal stoichiometric 

compositions. More sophisticated models that include detailed chemical reactions can predict the 

metabolic processes involved (Resendis et al 2011, Zhao et al 2012), but they also rely on 

observed stoichiometries and maximizing growth rates. Consequently, one limitation of our 

approach is that nutrient stoichiometries and rates are treated as constants, which is not 

necessarily consistent with biological systems (Näsholm et al 2009, Wolf et al. 2017). We 

addressed this challenge by independently measuring model parameters and predictions at all 

four nitrogen levels. Another limitation is that stoichiometric models do not account for other 

benefits of mutualisms, such as rhizobia acting as biocontrol agents to protect the plant from 

fungal pathogens (Das et al 2017) or association with rhizobia leading to the induction of defense 

signaling pathways (Dean et al 2014). We sought to minimize the influence of these other 

benefits by protecting the plants from pathogens and keeping them well-watered and under stable 

conditions.    

To our knowledge, the symmetry of bargaining power has not yet been directly tested in 

other plant-microbe mutualisms, but researchers have proposed several mechanisms that 

influence the power dynamics, such as symbionts conferring greater competitive power to more 

cooperative plants (Bücking et al 2016), partner strategies for preventing imbalances in benefits 

received from trade (Kiers et al 2011), and outside negotiator options leading to joint control of 

the mutualism (Ackay and Simms 2011). Further work in other plant-microbe symbioses is 

needed before it can be determined if asymmetric bargaining power is a broad feature of 

nutritional symbioses and how it relates to these mechanisms. 

64 

 

Conclusion 

Our finding of experimental support for a model of asymmetric (“unfair”) bargaining 

power between legumes and rhizobia versus a model of “fair” trade highlights the power of 

integrating quantitative models with data in the study of mutualisms (Clark et al., 2017), which 

we believe and will be broadly applicable to other systems. This work improves our 

understanding of the drivers of quantitative variation in symbiotic nitrogen fixation, a process 

that makes a major contribution to the global nitrogen cycle (Fowler et al., 2013) and is critical 

for agricultural sustainability (Herridge et al., 2008). Our finding of a reduction in plant 

bargaining power at higher nitrogen levels combined with the frequently high rates of fertilizer 

application to legume crops underscores the potential evolutionary danger of relaxed host control 

under anthropogenic inputs (Kiers et al., 2007; Weese et al., 2015). The ability to estimate 

bargaining power using this approach will allow bargaining strength to be compared with 

measurements of natural selection. We also believe these estimates contribute to investigating 

the genetic, metabolic, and physiological mechanisms underlying the regulation of resource 

exchange and could facilitate future efforts to breed crops that can maintain their own beneficial 

microbiomes (Busby et al., 2017). 

 

 

 

 

 

 

 

65 

 

 

 

 

 

 

 

 

 

 

 

 

 

APPENDIX  

 

 

 

 

 

 

 

 

 

 

66 

 

SUPPLEMENTAL INFORMATION TO CHAPTER 2 

Figure 2.7 | Whole-plant photosynthesis chambers were used to measure carbon uptake 
rates. (a) Chambers were connected to LI-COR LI-6400 apparatuses, illuminated with LED 
lights, and used in a growth chamber to ensure assay conditions matched growth conditions. (b) 
Chambers included fans to promote air circulation. (c) Steady-state photosynthesis was measured 
by collecting readings every 5 min for at least 90 min in the light. Respiration was estimated with 
dark measurements. Data shown is from a nodulated plant grown at 40 mg L-1 N, but is 
representative of all measured plants.  

 

67 

 

 

Figure 2.8 | Regression analyses used to calculate model parameters. (a) Direct nitrogen 
uptake (squares) and trade (triangles) rates per total plant biomass were experimentally estimated 
for uninoculated (blue) and/or nodulated (red) plants at 8, 24, 40, and 80 mg L-1 N. Linear or 
exponential regression was used to analyze the correlation between direct uptake or trade, 
respectively, and nitrogen availability. Values and error bars represent average and 90% 
confidence intervals, respectively, of 50 pseudo datasets generated by Monte Carlo sampling. (b) 
The percent of photosynthetic carbon uptake used for biomass is directly correlated with total 
carbon content in uninoculated plants. This correlation was analyzed with linear regression and 
used to estimate respiration rates in nodulated plants. Values represent 50 pseudo datasets per 
nitrogen treatment. (a,b) The linear equations were used in model analysis as described in the 
Methods. 

 

68 

 

Figure 2.9 | Plant growth is near linear between 4 and 6 weeks. Nodulated (triangles, dashed 
lines) and uninoculated (circles, solid lines) plant dry weights were measured at 8 (a), 24 (b), 40 
(c), and 80 (d) mg L-1 N. Growth rates were analyzed using linear regression. All rates were 
found to have p < 0.001, and equations and R2 values as indicated. Error bars represent standard 
error (n = 7-15). Plant ages with asterisks indicate significant differences between inoculated and 
uninoculated plants at that age; plant ages without asterisks do not significantly differ after post 
hoc testing with the Tukey test (p < 0.05).  

 

 

 

 

 

 

69 

 

Table 2.1 | Model input parameters and measured values. Parameters were measured and 
used in model construction as described in the Methods. Values correspond to the average ± 90% 
confidence interval by Monte Carlo sampling with 50 pseudo datasets.  

Parameter 

Biological 
interpretation 

plant size 

8 mg/L 

24 mg/L 

40 mg/L 

80 mg/L 

N 

N 

N 

N 

172.87 ± 

231.54 ± 

285.49 ± 

389.92 ± 

6.15 
3.91 ± 
0.21 

8.22 
3.45 ± 
0.19 

9.97 
2.58 ± 
0.12 

13.61 
0.42 ± 
0.05 

11.94 ± 

11.48 ± 

11.17 ± 

10.39 ± 

0.20 

0.12 

0.23 

0.05 

6.06 ± 
0.04 

6.17 ± 
0.05 

6.01 ± 
0.03 

6.19 ± 
0.06 

0.12 ± 
0.00 

0.01 ± 
0.00 
0.29 ± 
0.03 

0.15 ± 
0.01 

0.01 ± 
0.00 
0.27 ± 
0.04 

0.13 ± 
0.01 

0.02 ± 
0.00 
0.36 ± 
0.07 

0.11 ± 
0.00 

0.03 ± 
0.00 
3.05 ± 
3.56 

p 

r 

ynp 

ynr 

fcp' 

fnp' 

fnr' 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Units 

mg C 

rhizobial partner size  mg C 

plant carbon biomass 
yield per unit N 
nodule carbon 
biomass yield per unit 
N 

photosynthetic carbon 
uptake rate 

soil nitrogen uptake 
rate 

nitrogen fixation rate 

mg C/mg 
N 

mg C/mg 
N 

mg 
C/shoot C 
/day 
mg N/root 
C /day 
mg N/nod 
C/day 

70 

 

T 

aNP 

gP 

gR 

2.98 ± 0.07 

0.29 ± 0.02 

0.10 ± 0.00 

Table 2.2 | Model predictions and corresponding values. Experimentally estimated or model 
predicted carbon-for-nitrogen exchange ratio (T; mg C mg-1 N), proportion of plant carbon 
allocated to roots (aNP; g root C g-1 plant C), plant per capita growth rate (gP; g C day-1 g-1 plant 
C), rhizobial per capita growth rate (gR; g C day-1 g-1 rhizobial partner C), and proportion of 
bargaining power possessed by the plant (β). Model predictions include those predicted with or 
without organismal respiration, use the plant’s collective nodules as a proxy for the rhizobial 
partner or assume only 50% of the nodule biomass represents the rhizobial partner, and assume 
symmetric power (i.e., β = 0.50), average asymmetric power across the observed soil nitrogen 
range (i.e., β = 0.70), or asymmetric power which varies with soil nitrogen. Values correspond to 
the average ± 90% confidence interval by Monte Carlo sampling with 50 pseudo datasets (see 
Methods for further details).  
8 mg/L N 
experimentally 
estimated 
symmetric, without 
respiration 
symmetric, with 
respiration 
symmetric, with 
respiration, assume 
50% rhizobia 
average β, without 
respiration 
average β, with 
respiration 
variable β, without 
respiration 
variable β, with 
respiration 
variable β, with 
respiration, assume 
50% rhizobia 

0.09 ± 0.00  0.92 ± 0.01 

0.08 ± 0.01 

0.41 ± 0.07 

0.08 ± 0.00 

0.86 ± 0.11 

0.05 ± 0.00 

0.75 ± 0.10 

0.09 ± 0.00 

0.52 ± 0.07 

0.06 ± 0.00 

0.24 ± 0.04 

0.11 ± 0.00 

0.08 ± 0.00 

0.06 ± 0.00 

0.08 ± 0.00 

4.93 ± 0.16 

0.62 ± 0.02 

7.57 ± 0.34 

0.41 ± 0.02 

2.23 ± 0.05 

0.60 ± 0.02 

9.74 ± 3.79 

0.43 ± 0.04 

5.04 ± 0.19 

0.41 ± 0.03 

0.35 ± 0.04 

0.57 ± 0.02 

3.46 ± 0.09 

0.38 ± 0.02 

3.05 ± 0.08 

0.38 ± 0.02 

0.05 ± 0.00 

0.50 

0.50 

0.70 

0.70 

0.94 ± 0.01 

0.86 ± 0.03 

β 

- 

0.50 

0.07 ± 0.00 

 
24 mg/L N 
experimentally 
estimated 
symmetric, without 
respiration 
symmetric, with 
respiration 
symmetric, with 
respiration, assume 
50% rhizobia 
average β, without 
respiration 

 

 

T 

 

aNP 

 

gP 

 

gR 

2.97 ± 0.09 

0.25 ± 0.01 

0.09 ± 0.00 

0.07 ± 0.01 

3.65 ± 0.15 

0.53 ± 0.03 

0.11 ± 0.01 

0.55 ± 0.08 

5.32 ± 0.30 

0.37 ± 0.03 

0.07 ± 0.00 

0.27 ± 0.06 

5.25 ± 0.20 

0.38 ± 0.02 

0.08 ± 0.00 

0.48 ± 0.08 

1.76 ± 0.06 

0.53 ± 0.03 

0.12 ± 0.01 

0.33 ± 0.05 

 
β 

- 

0.50 

0.50 

0.50 

0.70 

71 

 

Table 2.2 (cont’d) 

variable β, with respiration 

variable β, with respiration, 
assume 50% rhizobia 

 

40 mg/L N 

experimentally estimated 

symmetric, without 
respiration 

symmetric, with respiration 

symmetric, with respiration, 
assume 50% rhizobia 
average β, without 
respiration 

average β, with respiration 

variable β, without 
respiration 

variable β, with respiration 

variable β, with respiration, 
assume 50% rhizobia 

 

80 mg/L N 

experimentally estimated 

symmetric, without 
respiration 

symmetric, with respiration 

symmetric, with respiration, 
assume 50% rhizobia 
average β, without 
respiration 

average β, with respiration 

variable β, without 
respiration 

 

 

 

 

 

3.69 ± 
0.23 
3.18 ± 
0.17 

T 

3.06 ± 
0.19 
3.03 ± 
0.15 
4.80 ± 
0.29 
4.46 ± 
0.19 
1.51 ± 
0.06 
3.61 ± 
0.11 
1.62 ± 
0.77 
3.80 ± 
0.26 
3.28 ± 
0.22 

T 

2.72 ± 
0.07 
1.83 ± 
0.10 
3.09 ± 
0.15 
2.86 ± 
0.12 
0.98 ± 
0.05 
2.83 ± 
0.08 
2.24 ± 
0.54 

0.36 ± 
0.03 
0.37 ± 
0.03 

 

aNP 
0.26 ± 
0.01 
0.46 ± 
0.02 
0.35 ± 
0.02 
0.32 ± 
0.01 
0.45 ± 
0.02 
0.32 ± 
0.02 
0.45 ± 
0.02 
0.33 ± 
0.02 
0.33 ± 
0.02 

 

aNP 
0.24 ± 
0.01 
0.31 ± 
0.01 
0.24 ± 
0.01 
0.22 ± 
0.01 
0.31 ± 
0.01 
0.23 ± 
0.01 
0.31 ± 
0.01 

72 

 

0.08 ± 
0.00 
0.08 ± 
0.00 

 

gP 

0.09 ± 
0.00 
0.11 ± 
0.00 
0.08 ± 
0.00 
0.08 ± 
0.00 
0.11 ± 
0.00 
0.08 ± 
0.00 
0.11 ± 
0.00 
0.08 ± 
0.00 
0.08 ± 
0.00 

 

gP 

0.10 ± 
0.00 
0.11 ± 
0.00 
0.08 ± 
0.00 
0.08 ± 
0.00 
0.11 ± 
0.00 
0.08 ± 
0.00 
0.11 ± 
0.00 

0.07 ± 
0.00 
0.06 ± 
0.01 

 

gR 

0.08 ± 
0.01 
0.63 ± 
0.13 
0.26 ± 
0.07 
0.51 ± 
0.11 
0.38 ± 
0.08 
0.15 ± 
0.03 
0.08 ± 
0.01 
0.08 ± 
0.01 
0.08 ± 
0.01 

 

gR 

0.10 ± 
0.01 
2.72 ± 
1.10 
0.72 ± 
0.43 
1.22 ± 
0.57 
1.63 ± 
0.66 
0.77 ± 
0.93 
0.10 ± 
0.02 

0.75 ± 
0.06 
0.83 ± 
0.05 

 
β 

- 

0.50 

0.50 

0.50 

0.70 

0.70 

0.84 ± 
0.06 
0.62 ± 
0.08 
0.77 ± 
0.06 

 
β 

- 

0.50 

0.50 

0.50 

0.70 

0.70 

0.37 ± 
0.11 

Table 2.2 (cont’d) 
variable β, with 
respiration 
variable β, with 
respiration, assume 
50% rhizobia 

2.91 ± 0.12 

0.23 ± 0.01 

0.08 ± 0.00 

0.06 ± 0.03 

0.57 ± 0.11 

2.89 ± 0.12 

0.23 ± 0.01 

0.08 ± 0.00 

0.05 ± 0.02  0.63 ± 0.11 

 
 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

73 

 

Table 2.3 | Plant carbon and nitrogen elemental compositions. Carbon and nitrogen elemental 
compositions were measured in 4- and 6-week-old uninoculated (-) and nodulated (+) plants at 8, 
24, 40, and 80 mg L-1 N. Values are average ± standard deviation. 

Root %C 

Shoot %C 

Root %N 

Shoot %N 

Plant Yield (mg 

(%wt) 
45.89 ± 

1.90 

(%wt) 
39.18 ± 

0.58 

42.89 ± 

40.49 ± 

2.40 

0.76 

43.96 ± 

39.80 ± 

0.84 

0.18 

44.19 ± 

39.06 ± 

1.89 

1.06 

44.59 ± 

41.06 ± 

1.85 

0.13 

44.13 ± 

40.72 ± 

0.74 

1.42 

43.51 ± 

39.81 ± 

0.49 

0.43 

45.63 ± 

39.34 ± 

0.38 

1.21 

43.43 ± 

38.40 ± 

1.41 

0.78 

44.32 ± 

39.17 ± 

0.48 

0.43 

41.93 ± 

39.45 ± 

4.68 

0.65 

44.43 ± 

39.56 ± 

2.32 

1.02 

43.02 ± 

39.30 ± 

3.15 

1.54 

42.10 ± 

39.45 ± 

1.73 

0.56 

41.51 ± 

39.98 ± 

2.64 

0.29 

42.86 ± 

41.12 ± 

0.48 

0.10 

(%wt) 
2.19 ± 
0.14 
2.31 ± 
0.19 
2.31 ± 
0.26 
2.71 ± 
0.27 
2.34 ± 
0.23 
2.40 ± 
0.13 
2.82 ± 
0.08 
3.23 ± 
0.24 
2.00 ± 
0.36 
1.99 ± 
0.26 
2.08 ± 
0.21 
2.99 ± 
0.15 
2.28 ± 
0.09 
2.16 ± 
0.41 
2.42 ± 
0.21 
2.98 ± 
0.18 

(%wt) 
3.46 ± 
0.53 
3.34 ± 
0.13 
3.45 ± 
0.28 
4.57 ± 
0.55 
4.41 ± 
0.19 
4.26 ± 
0.08 
4.38 ± 
0.31 
4.75 ± 
0.35 
2.60 ± 
0.52 
2.52 ± 
0.09 
2.89 ± 
0.22 
4.26 ± 
0.83 
3.47 ± 
0.51 
3.67 ± 
0.31 
3.65 ± 
0.59 
3.79 ± 
0.06 

C mg-1 N) 

13.76 ± 1.69 

13.74 ± 0.69 

13.24 ± 1.20 

10.00 ± 1.65 

10.69 ± 0.46 

10.86 ± 0.36 

10.21 ± 0.59 

9.23 ± 0.52 

17.16 ± 3.12 

17.48 ± 1.14 

15.21 ± 1.47 

10.25 ± 1.64 

12.77 ± 1.90 

12.06 ± 1.02 

12.24 ± 1.68 

11.49 ± 0.17 

Age 
(wk) 

[N] 

(mg/L) 

Rhiz
obia 

8 

24 

40 

80 

8 

24 

40 

80 

8 

24 

40 

80 

8 

24 

40 

80 

- 

- 

- 

- 

+ 

+ 

+ 

+ 

- 

- 

- 

- 

+ 

+ 

+ 

+ 

4 

4 

4 

4 

4 

4 

4 

4 

6 

6 

6 

6 

6 

6 

6 

6 

 

 

 

 

 
 
 

74 

 

Table 2.4 | Nodule carbon and nitrogen elemental compositions. Carbon and nitrogen 
elemental compositions were measured in 4- and 6-week-old nodulated plant systems at 8, 24, 
40, and 80 mg L-1 N. Values are average ± standard deviation. 

Age 
(wk) 

[N] 

(mg/L) 

4 
4 
4 
4 
6 
6 
6 
6 

8 
24 
40 
80 
8 
24 
40 
80 

Nodule %C 

Nodule %N 

Nodule Yield (mg C mg-

(%wt) 

44.30 ± 0.52 
45.26 ± 0.06 
43.02 ± 1.14 
43.15 ± 0.00 
43.53 ± 1.14 
44.56 ± 1.10 
43.57 ± 1.07 
40.47 ± 0.92 

(%wt) 

8.10 ± 0.11 
7.85 ± 0.48 
7.50 ± 0.42 
7.17 ± 0.00 
6.52 ± 0.36 
6.72 ± 0.24 
6.89 ± 0.24 
6.43 ± 0.68 

1 N) 

5.47 ± 0.01 
5.78 ± 0.36 
5.74 ± 0.27 
6.02 ± 0.00 
6.69 ± 0.36 
6.64 ± 0.15 
6.32 ± 0.09 
6.32 ± 0.52 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

 

 

 

75 

 

 

 

 

 

 

 

 

 

 

 

 

 

REFERENCES 

 

 

 

 

 

 

 

 

 

 

76 

 

REFERENCES 

 

Akçay E, Roughgarden J. 2007. Negotiation of mutualism: rhizobia and legumes. Proceedings  
of the Royal Society of London B: Biological Sciences 274(1606): 25-32. 

Binmore K, Rubinstein A, Wolinsky A. 1986. The Nash bargaining solution in economic  
modelling. The RAND Journal of Economics: 176-188. 

Bloom AJ, Chapin III FS, Mooney HA. 1985. Resource limitation in plants-an economic  
analogy. Annual review of Ecology and Systematics 16(1): 363-392. 

Bronstein JL. 2015. Mutualism: Oxford University Press, USA. 

Busby PE, Soman C, Wagner MR, Friesen ML, Kremer J, Bennett A, Morsy M, Eisen JA, 
Leach JE, Dangl JL. 2017. Research priorities for harnessing plant microbiomes in sustainable  
agriculture.PLoS biology 15(3): e2001793. 

Chae S, Heidhues P. 2001.  Nash bargaining solution with coalitions and the joint bargaining  
paradox:WZB Discussion Paper. 

Chakraborty T. 2011. Bargaining and pricing in networked economic systems. University of 
Pennsylvania. 

Chakraborty T, Kearns M, Khanna S 2009. Network bargaining: algorithms and structural  
results. Proceedings of the 10th ACM conference on Electronic commerce: ACM. 159-168. 

Clark TJ, Friel CA, Grman E, Shachar‐Hill Y, Friesen ML. 2017. Modelling nutritional  
mutualisms:challenges and opportunities for data integration. Ecology letters 20(9): 1203-1215. 

Das, K., Prasanna, R., & Saxena, A. K. (2017). Rhizobia: a potential biocontrol agent for  
soilborne fungal pathogens. Folia microbiologica, 62(5), 425-435. 

Daubech B, Remigi P, de Moura GD, Marchetti M, Pouzet C, Auriac M-C, Gokhale CS, 
Masson-Boivin C, Capela D. 2017. Spatio-temporal control of mutualism in legumes helps 
spread symbiotic nitrogen fixation. Elife 6. 

Dean, J. M., Mescher, M. C., & De Moraes, C. M. (2014). Plant dependence on rhizobia for 
nitrogen influences induced plant defenses and herbivore performance. International journal of 
molecular sciences, 15(1), 1466-1480. 

Fahraeus G. 1957. The Infection of Clover Root Hairs by Nodule Bacteria Studied by a Simple  
Glass Slide Technique. Microbiology 16: 374–381. 

77 

 

Franklin O, Näsholm T, Högberg P, Högberg MN. 2014. Forests trapped in nitrogen 
limitation–an ecological market perspective on ectomycorrhizal symbiosis. New phytologist  
203(2): 657-666. 

Frederickson ME. 2013. Rethinking mutualism stability: cheaters and the evolution of  
sanctions. The Quarterly review of biology 88(4): 269-295. 

Ghoul, M., Griffin, A. S., & West, S. A. (2014). Toward an evolutionary definition of cheating.  
Evolution, 68(2), 318-331 

Goh CH, Nicotra AB, Mathesius U. 2016. The presence of nodules on legume root systems can 
alter phenotypic plasticity in response to internal nitrogen independent of nitrogen fixation.  
Plant, cell & environment 39(4): 883-896. 

Grman E, Robinson TM, Klausmeier CA. 2012. Ecological specialization and trade affect the  
outcome of negotiations in mutualism. The American Naturalist 179(5): 567-581. 

Harris D, Pacovsky R, Paul E. 1985. Carbon economy of soybean–Rhizobium–Glomus  
associations. New phytologist 101(3): 427-440. 

Herridge DF, Peoples MB, Boddey RM. 2008. Global inputs of biological nitrogen fixation in  
agricultural systems. Plant and Soil 311: 1–18. 

Jones EI, Afkhami ME, Akçay E, Bronstein JL, Bshary R, Frederickson ME, Heath KD,  
Hoeksema JD, Ness JH, Pankey MS, et al. 2015. Cheaters must prosper: reconciling 
theoretical and empirical perspectives on cheating in mutualism. Ecology letters 18(11): 1270- 
1284. 

Kawamori T. 2014. A noncooperative foundation of the asymmetric Nash bargaining solution.  
Journal of Mathematical Economics 52: 12-15. 

Kiers ET, Hutton MG, Denison RF. 2007. Human selection and the relaxation of legume 
defences against ineffective rhizobia. Proceedings of the Royal Society of London B: Biological  
Sciences 274(1629): 3119-3126. 

Lambers H, Szaniawski RK, Visser R. 1983. Respiration for growth, maintenance and ion 
uptake. An evaluation of concepts, methods, values and their significance. Physiologia  
plantarum 58(4): 556-563. 

Nash JF. 1950. The bargaining problem. Econometrica: Journal of the Econometric Society:  
155-162. 

Näsholm, T., Kielland, K., & Ganeteg, U. (2009). Uptake of organic nitrogen by plants. New  
phytologist, 182(1), 31-48. 

78 

 

Paponov I, Posepanov O, Lebedinskai S, Koshkin E. 2000. Growth and biomass allocation, 
with varying nitrogen availability, of near-isogenic pea lines with differing foliage structure.  
Annals of botany 85(4): 563-569. 

Phillips DA. 1980. Efficiency of symbiotic nitrogen fixation in legumes. Annual Review of Plant  
Physiology 31(1): 29-49. 

Ratcliff, W. C., Underbakke, K., & Denison, R. F. 2011. Measuring the fitness of symbiotic 
rhizobia. Symbiosis, 55(2), 85-90. 

Reeve W, Chain P, O’Hara G, Ardley J, Nandesena K, Bräu L, Tiwari R, Malfatti S, Kiss 
H, Lapidus A, et al. 2010. Complete genome sequence of the Medicago microsymbiont Ensifer  
(Sinorhizobium) medicae strain WSM419. Standards in genomic sciences 2(1): 77. 

Regus J, Wendlandt C, Bantay R, Gano-Cohen K, Gleason N, Hollowell A, O’Neill M, 
Shahin K, Sachs J. 2017. Nitrogen deposition decreases the benefits of symbiosis in a native 
legume. Plant and Soil 414(1-2): 159-170. 

Regus JU, Quides KW, O'Neill MR, Suzuki R, Savory EA, Chang JH, Sachs JL. 2017. Cell  
autonomous sanctions in legumes target ineffective rhizobia in nodules with mixed infections. 
American Journal of Botany 104(9): 1299-1312. 

Resendis-Antonio, O., Hernández, M., Salazar, E., Contreras, S., Batallar, G. M., Mora, Y., 
& Encarnación, S. 2011. Systems biology of bacterial nitrogen fixation: high-throughput 
technology and its integrative description with constraint-based modeling. BMC systems biology, 
5(1), 120. 

Rufty TW, Raper CD, Huber SC. 1984. Alterations in internal partitioning of carbon in  
soybean plants in response to nitrogen stress. Canadian Journal of Botany 62(3): 501-508. 

Ryle G, Arnott R, Powell C, Gordon A. 1984. N2 fixation and the respiratory costs of nodules, 
nitrogenase activity, and nodule growth and maintenance in Fiskeby soyabean. Journal of 
Experimental Botany 35(8): 1156-1165. 

Schwartz MW, Hoeksema JD. 1998. Specialization and resource trade: biological markets as a  
model of mutualisms. Ecology 79(3): 1029-1038. 

Simms, E. L., & Taylor, D. L. (2002). Partner choice in nitrogen-fixation mutualisms of 
legumes and rhizobia. Integrative and Comparative Biology, 42(2), 369-380. 

Udvardi M, Poole PS. 2013. Transport and metabolism in legume-rhizobia symbioses. Annual 
review of plant biology 64: 781-805. 

Valladares F, Villar-Salvador P, Domínguez S, Fernández-Pascual M, Peñuelas JL, 
Pugnaire FI. 2002. Enhancing the early performance of the leguminous shrub Retama 
sphaerocarpa (L.) Boiss.: fertilisation versus Rhizobium inoculation. Plant and Soil 240(2): 253- 

79 

 

262. 

Voisin A-S, Salon C, Munier-Jolain NG, Ney B. 2002. Effect of mineral nitrogen on nitrogen 
nutrition and biomass partitioning between the shoot and roots of pea (Pisum sativum L.). Plant  
and Soil 242(2): 251-262. 

Wang X, Pan Q, Chen F, Yan X, Liao H. 2011. Effects of co-inoculation with arbuscular 
mycorrhizal fungi and rhizobia on soybean growth as related to root architecture and availability  
of N and P. Mycorrhiza 21(3): 173-181. 

Wardlaw IF. 1990. Tansley Review No. 27 The control of carbon partitioning in plants. New  
phytologist 116(3): 341-381. 

Weese DJ, Heath KD, Dentinger B, Lau JA. 2015. Long‐term nitrogen addition causes the 
evolution of less‐cooperative mutualists. Evolution 69(3): 631-642. 

Werner GD, Strassmann JE, Ivens AB, Engelmoer DJ, Verbruggen E, Queller DC, Noë R,  
Johnson NC, Hammerstein P, Kiers ET. 2014. Evolution of microbial markets. Proceedings of 
the National Academy of Sciences 111(4): 1237-1244. 

Weyl EG, Frederickson ME, Douglas WY, Pierce NE. 2010. Economic contract theory tests 
models of mutualism. Proceedings of the National Academy of Sciences 107(36): 15712-15716. 

Wolf, A. A., Funk, J. L., & Menge, D. N. (2017). The symbionts made me do it: legumes are 
not hardwired for high nitrogen concentrations but incorporate more nitrogen when inoculated. 
New Phytologist, 213(2), 690-699. 

Young ND, Debellé F, Oldroyd GE, Geurts R, Cannon SB, Udvardi MK, Benedito VA, 
Mayer KF, Gouzy J, Schoof H, et al. 2011. The Medicago genome provides insight into the 
evolution of rhizobial symbioses. Nature 480(7378): 520. 

Younginger, B. S., Sirová, D., Cruzan, M. B., & Ballhorn, D. J. 2017. Is biomass a reliable 
estimate of plant fitness?. Applications in plant sciences, 5(2), 1600094. 

Zhao, H., Li, M., Fang, K., Chen, W., & Wang, J. 2012. In silico insights into the symbiotic 
nitrogen fixation in Sinorhizobium meliloti via metabolic reconstruction. PLoS One, 7(2), 
e31287. 
 

 
 

 

 

 

80 

 

CHAPTER 3 

 

LEGUMES MODULATE ALLOCATION TO RHIZOBIAL NITROGEN FIXATION IN 

RESPONSE TO FACTORIAL RESOURCE MANIPULATION 

Abstract  

 

The costs and benefits that define gain from trade in resource mutualisms depend on 

resource availability. Optimal partitioning theory predicts that allocation to direct uptake versus 

trade will be determined by both the relative benefit of the resource acquired through trade, and 

the relative cost of the resource being traded away. The costs and benefits of carbon:nitrogen 

exchange in the legume-rhizobia symbiosis have been examined in depth with regards to mineral 

nitrogen availability. However, the effects of varying carbon costs have rarely been considered 

and we currently lack definitive empirical examples of legumes modulating symbiosis in 

response to carbon availability. Here, we report plant growth and symbiosis investment in the 

model legume Medicago truncatula and its symbiont Ensifer medicae across varying nitrogen 

and light environments. We demonstrate that plants modulate their allocation to roots and 

nodules as their return on investment varies according to external nitrogen and carbon 

availability. We find empirical evidence that plant allocation to nodules responds to carbon 

availability, but that this depends upon the nitrogen environment. In particular, at low nitrogen–

where rhizobia provided the majority of nitrogen for plant growth–relative nodule allocation 

increased when carbon limitation was alleviated with high light levels. This context-dependent 

modulation of resource allocation to rhizobia prevents this interaction from becoming parasitic 

even in low-light, high-nitrogen environments. 

 

 

81 

 

Introduction 

Interactions between plants and microbial symbionts are major drivers of global nutrient 

cycles and play vital roles in the productivity of natural and agricultural ecosystems (van der 

Heijden et al., 2008). Microbial symbionts can supply plants with both nitrogen and phosphorus, 

essential nutrients that commonly limit plant growth (Erisman et al., 2013). Similarly, soil 

bacteria known as rhizobia can colonize plant roots and induce the formation of nodules, inside 

of which the rhizobia fix an estimated 40 million tons of plant-inaccessible nitrogen from the 

atmosphere in exchange for photosynthetic carbon (Udvardi & Poole, 2013). While these 

interactions are generally regarded as mutualistic, with the symbiosis increasing fitness for both 

partners (Bronstein, 2015), theory predicts that symbioses exist along a gradient from mutualism 

to parasitism depending on the environmental context (Bronstein, 1994, 2001; Johnson et al., 

1997; Neuhauser & Fargione, 2004).   

One key factor that may shift an interaction along the mutualism-parasitism continuum is 

the availability of the traded resources in the environment. Symbionts are predicted to shift from 

mutualism towards parasitism when the resource they supply is abundant in the soil, and thus the 

benefit to the host of obtaining it from the symbiont is reduced (Bronstein, 1994; Neuhauser & 

Fargione, 2004). With high levels of nitrogen fertilizer use and nitrogen deposition across 

ecosystems (Foley et al., 2005), there is concern that nitrogen-fixing symbioses between plants 

and microbes such as rhizobia may break down. Even a single growing season of fertilizer 

application is sufficient to change the composition of the rhizobia population in soil (Simonsen et 

al., 2015). Long-term nitrogen addition experiments have shown that prolonged nitrogen 

fertilization leads to the evolution of less effective rhizobia in the Trifolium-rhizobium symbiosis 

(Weese et al., 2015). This partner quality decline may be linked to evolutionary differentiation in 

82 

 

the symbiotic plasmid (Klinger et al., 2016). However, in the presence of externally supplied 

nitrogen the plant may be able to minimize the costs associated with less beneficial rhizobia by 

reducing or eliminating its allocation of resources to the microbes.  

The way a plant allocates resources to microbial symbionts such as rhizobia can be 

considered in the framework of biological market theory (Schwartz & Hoeksema, 1998). The 

plant can allocate its resources in two distinct ways: it can increase root biomass to take up 

nitrogen directly from the soil or it can increase photosynthesis to acquire carbon to trade for 

nitrogen with rhizobia. The cost-benefit analysis of trade versus direct uptake depends on the 

availability of both traded resources. Optimal partitioning theory predicts that the plant will 

allocate biomass to the part of the plant that acquires the resource that is most limiting to the 

plant (Thornley, 1972; Bloom et al., 1985). In this case, each partner will specialize in acquiring 

the resource for which it has a comparative advantage and trade to acquire the other resources, 

and both partners will acquire more total resources than they would in isolation (Schwartz & 

Hoeksema, 1998). When trade is beneficial, an organism increases its own potential fitness by 

engaging in trade and mutualisms can readily evolve, essentially as by-product mutualisms. 

However, if the resource being traded away is not available in excess, or if the resource being 

traded for is abundant in the environment and/or cheap to obtain, the fitness gain from trade may 

become negative (Johnson et al., 1997; Schwartz & Hoeksema, 1998). In this context, optimal 

partitioning theory predicts that under high mineral nitrogen levels the plant will downregulate 

allocation to symbiosis–in extreme cases perhaps terminating the relationship entirely if they are 

able to. However, a plant’s optimal allocation to nitrogen uptake will also depend critically upon 

both the carbon cost of each uptake strategy as well as the carbon available to the plant. 

83 

 

There is a large body of empirical evidence for shifts in allocation and context-dependent 

benefits in response to nutrient availability for the nutrient supplied by the microbe in plant-

microbe nutritional symbioses, but very few studies have investigated the effects of factorially 

manipulating both traded resources. The negative effects of soil nitrate, the most commonly 

available form of soil nitrogen, on nodulation has long been reported in the literature (Streeter & 

Wong, 1988; Lucinski et al., 2002; Glyan’ko et al., 2009), though the magnitude of the effect of 

nitrate on nodulation may be strongly affected by genotype-by-genotype interactions (Heath et 

al., 2010).  

In contrast to the preponderance of literature exploring the effects of mineral nitrogen on 

legume-rhizobia interactions, relatively little is known about the effects of light on these 

interactions. The studies that do exist have found wildly varying results. Three species of 

Desmodium exhibit reduced plant biomass and total nodule number with shade, but no 

corresponding change in the ratio of nodule biomass to root biomass and plant nitrogen content 

(Houx et al., 2009). Trifolium repens reduced total nodule biomass in shaded conditions, but this 

was mostly explained by reduced root biomass (Chu & Robertson, 1974). Various studies in 

soybean have shown that shading increases nodule biomass and decreases efficiency (Santos et 

al., 1997), that shading decreases nodule biomass and increases efficiency (Araujo et al., 2018), 

and that shading decreases nodule biomass but does not affect efficiency (Hansen et al., 1990).  

The literature regarding interactions between nitrogen and carbon availability is even 

sparser and currently limited to determinate legume-rhizobium symbioses in which the rhizobia 

are non-terminally differentiated as bacteroids (Denison, 2000). Lau et al., (2012) manipulated 

N-P-K fertilizer and light levels and found that Bradyrhizobium japonicum nodulation on 

Glycine max (soybean) was significantly decreased by low light levels, but this study did not 

84 

 

detect an effect of fertilizer on nodulation. Bradyrhizobium increased plant biomass in the low 

nutrient, high light conditions, but had no effect on aboveground plant biomass in low nutrient, 

low light conditions or in any high nutrient conditions (Lau et al., 2012), making it difficult to 

interpret this study in the framework of market theory. Similarly, Regus et al., (2015) 

manipulated KNO3 application and light regime (by season of growth in the greenhouse) in the 

Lotus strigosus-Bradyrhizobium symbiosis. They found that nitrogen fertilization eliminated 

plant growth benefits from rhizobia in the fall when there light levels were lower, while nitrogen 

fertilization reduced but did not eliminated plant growth benefits from rhizobia in the winter 

when light levels were higher (Regus et al., 2015). Nitrogen decreased nodule biomass in both 

seasons, while it decreased nodule number in fall but not winter, consistent with the idea that low 

light makes carbon more expensive and thus reduces the overall investment in symbiosis (Regus 

et al., 2015). These results suggest that there may be interactions between the availability of 

carbon and nitrogen, though it is impossible to rule out other environmental factors that varied 

between seasons. Thus, these initial results highlight the need for a highly controlled analysis of 

the interactions between carbon and nitrogen availability and their effects on plant investment in 

trade with rhizobia.  

In this study, we examined changes in plant biomass allocation and trade with rhizobia in 

response to variation in both light level and soil mineral nitrogen. We used the model legume 

Medicago truncatula and its rhizobial partner Ensifer medicae WSM419 grown under controlled 

conditions with factorial light and nitrogen resource manipulation. We hypothesized that plants 

would allocate resources optimally according to optimal partitioning theory, leading to three 

predictions: 1) plants will allocate more resources to acquiring the limiting nutrient (nitrogen or 

carbon) as a function of external inputs, 2) plants will allocate relatively more resources to 

85 

 

nodules than to roots when soil nitrogen is low and thus the return on investment for root 

allocation is reduced relative to high soil nitrogen, and 3) when soil nitrogen is low, plants at 

high light will invest highly in nodules but carbon scarcity will modulate this allocation because 

when carbon is expensive the return on investment from roots will be higher than that from 

nodules. 

Materials and Methods 

Pot preparation 

 

We filled SC10 Cone-Tainers (Steuwe and Sons Inc., Corvallis, OR, USA) with medium 

vermiculite. The pots had a ⅜” cotton wick leading to an opaque 50 mL reservoir to ensure 

constant access to liquid and nutrients. Prior to planting, we added 25 mL of deionized water to 

each pot, autoclaved once for 60 minutes, then fertilized with 25 mL of Fahraeus solution with 8 

or 80 mg/L N (as NH4NO3). We then autoclaved the pots twice more for 60 minutes with 

approximately 24 hours between each run. 

  

Seedling preparation 

We scarified seeds of the model legume Medicago truncatula genotype A17 (Young et 

al., 2011) with 600 grit sandpaper and sterilized them in full strength commercial bleach (8.25% 

NaHClPO3) for 3 minutes, followed by 6 rinses with sterile deionized water. We incubated the 

seeds in sterile deionized water for 3 hours at room temperature, then re-sterilized imbibed seeds 

in 10% bleach for 30 seconds. After 6 rinses with sterile deionized water, seeds were incubated 

in sterile deionized water for 48 hours at 4°C. We changed the water once every twelve hours. 

After 48 hours, we transferred seeds to sterile petri dishes sealed with Parafilm and germinated 

them at room temperature in the dark for 48 hours. Seedlings with radicles at least 1 cm long 

86 

 

were transplanted into prepared pots. After planting, the plants were fully randomized and grown 

at 25°C with a 16 hour day/8 hour night cycle at approximately 150 µmol m-2s-1 to encourage 

uniform seedling establishment. Seedlings were misted daily with sterile deionized water to keep 

the radicles moist. After five days, plants were transferred to a high light growth chamber where 

half of the plants received full irradiation (400 µmol m-2s-1) and half were shaded with Sun Mesh 

Sunblock shade cloth to 200 µmol m-2s-1. Full irradiance plants were grown in one rack while 

shaded plants were grown in a separate rack. All plants were grown with equal spacing that 

prevented shading by other plants. All plants were grown at 25°C with a 16 hour day/8 hour 

night cycle.  

  

Rhizobia preparation 

Cultures of Ensifer medicae WSM419 (Reeve et al., 2010) were grown for 48 hours in ½ 

TY at 30°C shaking at 200 RPM. Cell density was determined by measuring the OD600 of the 

culture using a NanoDrop. After 3 days of acclimation to the new growth chamber conditions, 

rhizobia inoculated plants received 1 mL of 107 CFU/mL suspended in ½x phosphate buffered 

saline (PBS) and rhizobia-free plants received 1 mL of sterile ½x PBS. Sterility of the mock 

inoculum and cell count of the rhizobial inoculum were checked using spot plating and serial 

dilution on TY agar. 

  

Plant growth and harvest 

 

Plants were watered from below with 25 mL of sterile Fahraeus nutrient solution with 8 

or 80 mg/L nitrogen (as NH4NO3) whenever the reservoirs ran dry (at least twice a week). After 

4 weeks of growth, plants were harvested. Nodules were plucked and counted and tissue was 

87 

 

dried at 60°C for one week. We measured root, shoot, and nodule dry weight and nodule 

number. No uninoculated plants developed nodules, indicating that there was no rhizobial 

contamination of uninoculated plants.  

  

Statistical analysis 

To test the effects of soil nitrogen and light availability on nodule biomass and the ratio 

of nodule biomass to root biomass, we used a linear model ANOVA with Type II sum of squares 

(car package, R 3.4.3) with nitrogen and light main effects and nitrogen by light interactions as 

fixed effects. Nodule biomass and the nodule:root ratio were log transformed to improve 

normality. We tested the effects of nitrogen and light on nodule number using a generalized 

linear model with a poisson distribution and ANOVA with Type II sum of squares (car package, 

R 3.4.3).  

To test the relationship between nodule biomass and shoot biomass across nitrogen and 

light conditions, we used a linear model ANOVA with Type II sum of squares (car package, R 

3.4.3) with nitrogen, light, nodule biomass, and all their interactions as fixed effects. Shoot 

biomass and nodule biomass were log transformed to improve normality.  

To test the effects of nitrogen, light, and rhizobial inoculation on shoot biomass, root 

biomass, and the root:shoot ratio, we used a linear model ANOVA with Type II sum of squares 

(car package, R 3.4.3) with light, soil nitrogen, and rhizobial status and all interactions as fixed 

effects. Shoot biomass and root biomass were ln-transformed to improve normality. In all cases, 

to determine whether group means were significantly different, we conducted post-hoc testing 

with the Tukey test at a significance level of 0.05 (emmeans package, R 3.4.3).  

 

88 

 

Results 

Both light and nitrogen affect total investment in rhizobia 

High light increased nodule number by 32% overall (p = 4.93e-05), while high soil 

nitrogen decreased nodule number by 77% overall (p < 2e-16; Table 3.1). We did not detect a 

significant interactive effect of light and nitrogen on nodule number (p = 0.275; Table 3.1). This 

was surprising since high light significantly increases nodule number by 35% at low soil nitrogen 

(p = 0.0002) but non-significantly by 11% at high soil nitrogen (p = 0.906; Figure 3.1a). It seems 

likely that this statistical anomaly is caused by the relatively small effect size of light overall.  

We detected a significant interaction between the effects of light and nitrogen on nodule 

biomass (p = 6.54e-04; Table 3.1). High soil nitrogen significantly reduced nodule biomass by 

95% regardless of light level (p < 0.001 at both light levels; Figure 3.1b). However, at low soil 

nitrogen, high light increased nodule biomass by 103% (p < 0.001), while high light did not 

significantly increase nodule biomass at high soil nitrogen (p = 0.889; Figure 3.1b). 

 

 

 

 

 

 

 

 

 

 

89 

 

 

Figure 3.1 | Effects of light and nitrogen on total nodulation. Nodule biomass (a) and nodule 
number (b) in inoculated M. truncatula plants in response to changing soil nitrogen and light 
availability levels. Error bars represent +/- one standard error. Note that nodule biomass was ln-
transformed to improve normality in the ANOVA, but is represented here without transformation 
for ease of interpretation. Bars with the same letter within an individual panel do not 
significantly differ after Tukey post hoc testing (p > 0.05). 
 
 
Table 3.1 | ANOVAs summarizing the effects of light and nitrogen on nodulation. Bold 
indicates statistically significant effects (p < 0.05). 
 

Nodule number   Nodule biomass 

p 

  F(1,34) 

p 

1.19  0.275 

χ2 
  
16.5  4.93e-05    25.1  1.69e-05 
Light (L) 
Nitrogen (N)  375  < 2e-16    342  < 2.2e-16 
L*N 
  14.1  6.54e-04 
 
 
 
 
 
 
 
 
 
 
 

90 

 

Both light and nitrogen affect investment in rhizobia versus trade, but not return on investment 

 

High light decreased specific nodulation (nodule number per mg of root biomass) by 31% 

overall (p = 0.005; Table 3.2), though we could not detect significant differences in specific 

nodulation when all pairwise combinations were tested during Tukey testing (p = 0.355 at low 

nitrogen and p = 0.061 at high nitrogen; Figure 3.2a). High nitrogen decreased specific 

nodulation by 88% overall (p < 2.2e016; Table 3.2, Figure 3.2a). 

 

We detected a significant interaction between the effects of light and nitrogen on 

nodule:root biomass (mg nodule biomass per mg root biomass) (p = 0.022; Table 3.1). High 

nitrogen significantly decreased nodule:root biomass by approximately 97% regardless of light 

level (p < 0.001 for both light levels; Figure 3.1b). High light increased nodule:root biomass by 

16% at low nitrogen (p = 0.021) but did not have a significant effect on nodule:root biomass at 

high nitrogen (p = 0.977; Figure 3.1b) 

 

We detected a significant interaction between the effects of nitrogen and nodule biomass 

on shoot biomass gain from rhizobia, or shoot biomass minus mean biomass of control plants per 

condition (p = 0.004; Table 3.3). We did not detect a significant main effect of light, nitrogen, an 

interaction between the effects of light and nitrogen, or a three-way interaction between light, 

nitrogen, and nodule biomass (Table 3.3). The significant nitrogen x nodule biomass term 

indicated that the slope of the linear regression of shoot biomass gain on nodule biomass differed 

based on nitrogen treatment (Figure 3.2c). However, neither slope was significantly different 

from zero (p = 0.796 for low nitrogen and 0.393 for high nitrogen).  

 

91 

 

 

Figure 3.2 | Effects of light and nitrogen on investment in trade versus direct uptake and 
return on investment in trade. Specific nodulation (nodule number per mg root biomass) (a), 
nodule:root biomass (mg nodule biomass per mg root biomass) (b), and efficiency (shoot 
biomass gain, or shoot biomass minus mean shoot biomass for control plants of each treatment 
group versus nodule biomass: mg shoot gained per mg invested in rhizobia). Dashed lines 
indicate nonsignificant slopes.  
 
Table 3.2 | ANOVAs summarizing the effects of light and nitrogen on nodulation scaled by 
root biomass. Bold indicates statistically significant effects (p < 0.05). 
 

Specific nodulation   Nodule:root biomass 

F(1,34) 
  
Light (L) 
9.05 
Nitrogen (N)  318 
0.560 
N*L 

p 

0.005 
< 2e-16 
0.459 

  F(1,34) 
   3.80 
   858 
   5.77 

p 

0.060 

< 2.2e-16 

0.022 

92 

 

  Table 3.3 | ANCOVA summarizing the effects of light and nitrogen on the relationship 
between nodule biomass and shoot biomass gain. Bold indicates statistically significant effects 
(p < 0.05). 
 

Shoot biomass gain    
   
   
   
  
  
  
   
  

0.348 
0.163 
0.112 
0.469 
0.814 
0.004 
0.509 

p 

F(1,30) 
  
0.910 
Light (L) 
2.05 
Nitrogen (N) 
Nodule biomass (NB)  2.67 
0.540 
L*N 
0.056 
L*NB 
N*NB 
9.62 
L*N*NB 
0.446 
 
 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

93 

 

Nitrogen, light, and rhizobia affect relative allocation between root and shoot 

 

High light increases root:shoot ratio by 16% overall (p = 0,004; Table 3.4), though we 

did not detect a significant effect of light when all pairwise comparisons were tested during 

Tukey testing (Figure 3.3). We detected a significant interaction between the effects of nitrogen 

and rhizobia on root:shoot ratio (p = 1.29e-04; Table 3.4). Rhizobia decreased root:shoot ratio 

44% at low soil nitrogen (p < 0.001) but had no effect at high soil nitrogen (Figure 3.3). 

Similarly, increasing soil nitrogen decreased root:shoot ratio by 39% in the absence of rhizobia 

(p < 0.001), but had no significant effect in the presence of rhizobia (Figure 3.3). 

 

 

 

 

 

94 

 

Figure 3.3 | Effects of light, nitrogen, and rhizobia on root:shoot ratio. Error bars represent 
+/- one standard error. Note that root:shoot ratio was ln-transformed to improve normality in the 
ANOVA, but is represented here without transformation for ease of interpretation. Bars with the 
same letter do not significantly differ after post hoc testing with the Tukey test (p > 0.05). 

 

Root:shoot ratio 
F(1, 67) 
8.85 

0.004 

p 

 
 
Table 3.4 | ANOVA summarizing the effects of light, nitrogen, and rhizobia on root:shoot 
ratio. Bold indicates statistically significant effects (p < 0.05). 
  
  
Light (L) 
Nitrogen (N)  42.2  1.18e-08 
Rhizobia (R)  58.4  1.07e-10 
L*N 
L*R 
N*R 
L*N*R 
 

0.399  0.530 
0.872  0.354 
16.5  1.29e-04 
1.24 

0.270 

 

 

 

95 

 

Light, nitrogen, and rhizobia modulate plant performance 

We detected significant interactions between the effects of light and nitrogen on shoot 

biomass (p = 0.006, Table 3.5). Light increased shoot biomass by 35% at low soil nitrogen and 

by 149% at high soil nitrogen (Figure 3.4a). Similarly, increasing soil nitrogen increased shoot 

biomass by 183% at low light and by 423% at high light (Figure 3.4a). We also detected 

significant interactions between the effects of soil nitrogen and rhizobia (p = 3.50e-05, Table 

3.5). Rhizobia increased shoot biomass by 135% at low soil nitrogen but had no significant effect 

at high soil nitrogen (Figure 3.4a). Increasing soil nitrogen increased shoot biomass by 581% in 

the absence of rhizobia but only 207% in the presence of rhizobia (Figure 3.4a). 

We detected similar trends for root biomass. We detected a significant interaction 

between the effects of nitrogen and light (p = 0.001, Table 5). Increasing light increased root 

biomass by 62% at low soil nitrogen and by 192% at high soil nitrogen (Figure 3.4b). Increasing 

soil nitrogen increased root biomass by 121% at low light and by 299% at high light (Figure 

3.4b). We also detected significant interactions between the effects of rhizobia and nitrogen on 

root biomass (p = 0.011, Table 3.5). Rhizobia did not have a significant effect on root biomass, 

but it changed the magnitude of the effect of soil nitrogen: increasing soil nitrogen increased root 

biomass by 307% in the absence of rhizobia, but only by 175% in the presence of rhizobia 

(Figure 3.4b).  

 

 

96 

 

 
Figure 3.4 | Effects of light, nitrogen, and rhizobia on shoot and root biomass. a) Shoot 
biomass and b) root biomass in M. truncatula plants in response to changing soil nitrogen, light 
availability, and the presence or absence of rhizobia. Error bars represent +/- one standard error. 
Note that root and shoot biomass were ln-transformed to improve normality in the ANOVA, but 
are represented here without transformation for ease of interpretation. Bars with the same letter 
within an individual panel do not significantly differ after post hoc testing with the Tukey test (p 
> 0.05). 

 

 

 
Root biomass 
 
F(1, 67) 

p 

p 

Table 3.5 | ANOVAs summarizing the effects of light, nitrogen, and rhizobia on shoot and 
root biomass. Bold indicates statistically significant effects (p < 0.05). 
  
  
Light (L) 
 
Nitrogen (N)  194  < 2.2e-16 
 
Rhizobia (R)  27.9  1.45e-06 
 
L*N 
L*R 
N*R 
L*N*R 
 

Shoot biomass 
F(1, 67) 
41.6  1.43e-08   71.5  3.60e-12 
132  < 2.2e-16 
2.68  0.106 
11.5  0.001 
 0.945  0.334 
 
6.80  0.011 
 
0.646  0.424 

8.19  0.006 
0.204  0.653 
19.7  3.50e-05 
1.69  0.199 

 

97 

 

Discussion  

As predicted, shifting the resource environment altered the plant’s allocation of resources 

between their rhizobial trading partners and the acquisition of external nutrients. In the biological 

market framework, increasing soil nitrogen is expected to reduce allocation to trade with rhizobia 

because the plant is able to obtain a larger proportion of its nitrogen needs through direct 

nitrogen uptake, which is less energetically costly than nitrogen fixation (Voisin et al., 2002). 

This is exactly what we observe in response to increased soil nitrogen: plants sharply decreased 

both their total investment in nodulation (Figure 3.1) and their relative investment to nodules 

versus roots (Figure 3.2). This means that increasing soil nitrogen decreases both nodule 

initiation and the amount of biomass being allocated to existing nodules. In contrast, Lotus 

japonicus has been shown to regulate nodulation in response to nitrogen mainly through changes 

in nodule size (Regus et al., 2015), and soybean has been shown to regulate nodulation in 

response to nitrogen mainly through changes in nodule number (Lau et al., 2012). The effects of 

soil nitrogen on nodulation in M. truncatula have been shown to vary depending on genotype x 

genotype interactions between host and rhizobia, suggesting that these effects are highly specific 

(Heath et al., 2010).  

In contrast to trends in the abundance of soil nitrogen, increasing light availability 

increases the plant’s potential for carbon fixation and thus presumably its carbon supply. This 

should result in increasing investment into nitrogen acquisition, though the breakdown between 

direct nitrogen update and trade would be based on the relative cost of each, which is determined 

by the soil nitrogen level (Schwartz & Hoeksema, 1998). We found that increasing light 

availability increased both total nodule number and total nodule biomass, and this effect was 

much more pronounced at low soil nitrogen—the only condition in which rhizobia significantly 

increased shoot biomass (Figure 3.1, Figure 3.4). In addition, light increased the nodule:root 

98 

 

biomass ratio only at low soil nitrogen (Figure 3.2). This suggests that plants are investing 

relatively more biomass into existing nodules with increasing light, but only under nitrogen 

conditions in which the rhizobia are beneficial to the plant. In contrast, light decreased specific 

nodulation regardless of nitrogen level (Figure 3.2), which indicates that the increase in total 

nodule number with increasing light (Figure 3.1) is being driven by an increase in root biomass 

that overcomes a reduced level of nodule initiation per unit of root biomass. In total, these 

complex effects of light suggest, since the magnitude and direction of the effect of light is 

different for nodule number and nodule biomass, light may be acting through multiple pathways 

to regulate nodulation. Furthermore, these results highlight the importance of examining 

nodulation relative to root biomass when other experimental treatments are expected to alter 

plant size. While unscaled nodule number and biomass are important predictors of rhizobial 

fitness (Ratcliff et al., 2012), assessing only unscaled nodulation may lead to misleading 

conclusions about plant allocation. The wide variation in the effects of light on nodulation 

reported in the literature may be due to variation in genotype x genotype responses to light, 

similar to the varying effects of nitrogen detected by Heath et al., 2010, but it may also be 

explained by differences in the type of nodulation measures reported (i.e. nodule:root biomass in 

Houx et al., 2009 but unscaled nodule biomass in Santos et al., 1997). 

The benefit that plants get from trade with rhizobia is the product of their investment in 

rhizobia and the return on investment, or efficiency of the nodules. We measured efficiency as 

the increase in shoot biomass relative to the control, divided by nodule biomass. We did not 

detect a significant effect of nodule biomass on shoot biomass gain, but there was generally a 

positive relationship between the two (Figure 3.2). Figure 3.2 also suggests that nodules have a 

much higher efficiency at high nitrogen than at low, but this is likely an artefact of the 

99 

 

visualization. When the same plot was visualized for the four-week-old plants from Chapter 1, 

where only soil nitrogen was varied, the 80 mg/L N plants appeared to have a much higher slope 

than low nitrogen plants (Figure 3.5), but there was no difference in the carbon:nitrogen trade 

ratio for those plants (Figure 2.5). Thus, these apparent trends are most likely caused by much 

larger variation in shoot biomass combined with very low nodule biomass at high soil nitrogen, 

rather than any real relationship.  

Regardless, there is no support for an effect of light on nodule efficiency as was reported 

in soybean (Santos et al., 1997; Araujo et al., 2018). There are several possible explanations for 

this discrepancy, including differences in the magnitude of the change in light, the way 

efficiency was measured, and the fact that soybeans form determinate nodules while M. 

truncatula forms indeterminate nodules (Oono et al., 2009). We were only able to measure 

efficiency as shoot biomass gain per unit of nodule biomass, but Santos et al., (1997) and Araujo 

et al., (2018) measured it as mg of nitrogen fixed per mg of nodule. Neither measure is perfect 

because nodule biomass does not account for all of the carbon allocated to rhizobia (Rainbird et 

al., 1984), and the amount of nitrogen fixed may not directly translate into plant fitness benefit if 

the biomass yield per nitrogen or relative nitrogen allocation changes. However, combining our 

results here with the results from Chapter 1 that the carbon:nitrogen exchange ratio does not 

change with changing soil nitrogen, it appears that the efficiency of nodules in M. truncatula is 

not affected by nutrient conditions. This suggests that M. truncatula has the regulation of 

rhizobial nitrogen fixation very tightly controlled and is always operating it at maximum 

efficiency, likely due to the energy intensity of nitrogen fixation (Silsbury, 1977; Andrews et al., 

2009).  

100 

 

The final aspect of plant biomass allocation is the balance of biomass between the roots 

(for nitrogen acquisition) and shoots (for carbon acquisition). When we assessed these allocation 

patterns by measuring the root:shoot ratio, we found strong interactions between the effects of 

nitrogen and rhizobia. Nitrogen only affected root:shoot ratio in uninoculated plants, and 

rhizobia only affected root:shoot ratio at low nitrogen (Figure 3.3). Low nitrogen inoculated 

plants, high nitrogen uninoculated plants, and high nitrogen inoculated plants all had statistically 

indistinguishable root:shoot ratios (Figure 3.3), even though the low nitrogen plants were 

significantly smaller than the high nitrogen plants (Figure 3.4). Thus, the rhizobial effect on 

root:shoot ratio is not fully explained by changes in plant nitrogen status. This suggests that 

plants are changing their allocation strategy to acquire more carbon to trade for nitrogen instead 

of directly acquiring nitrogen with root biomass. Goh et al., (2016) used a non-fixing nodD 

mutant to show that rhizobial effects on root:shoot ratio in M. truncatula appear to depend only 

on nodule initiation, not on nitrogen fixation by the rhizobia. Thus, this trend may show that 

rhizobia are able to manipulate plants into allocating more resources to trade even when it is not 

beneficial. 

Finally, these allocation decisions and environmental conditions resulted in interesting 

trends in plant biomass. We detected significant interactions between nitrogen and rhizobia 

effects and between nitrogen and light effects on shoot biomass, but not between light and 

rhizobia (Table 3.5). Rhizobia are not beneficial to the plant at high soil nitrogen levels because 

of the high cost of nitrogen fixation (Silsbury, 1977; Voisin et al., 2002; Andrews et al., 2009) 

and the ability of the plant to obtain sufficient “cheap” nitrogen directly from the soil. The 

interaction of nitrogen and light can be explained by the plant being strongly nitrogen limited at 

low soil nitrogen levels, so that increasing light does not allow more growth because the limiting 

101 

 

resource is nitrogen. However, at high soil nitrogen and low light, carbon is limiting, so 

increasing light availability allows for a much larger nitrogen effect. At first it seems 

counterintuitive that investment in rhizobia is increased by increasing light, and that efficiency 

does not change depending on conditions, but the benefit from rhizobia does not change 

depending on light. This is likely due to the high cost of nitrogen fixation: since each unit of 

nitrogen fixed by rhizobia costs a relatively high amount of carbon, this change in allocation 

does not have statistically significant effects on shoot biomass. In addition, the carbon costs of 

rhizobia may be at least partially counteracted by the stimulation of photosynthesis by nodules, 

either by improved leaf nitrogen status (Kaschuk et al., 2009) or through the extreme sink 

strength of nodules (Brown & Bethlenfalvay, 1987; Kaschuk et al., 2009, 2010, 2012), meaning 

that these mechanisms may counteract the carbon dependency of benefit from nodules. 

It is important to note that due to limitations of the facilities and equipment available, the 

light levels used in this experiment were a relatively small fraction of the maximum light 

intensity a plant might experience in full sun (200-400 µmol m-2s-1  compared to 2000 µmol m-2s-

1; (Korczynski. et al., 1991). However, these levels are within the range that legumes may face 

when being shaded by other plants (Burkey & Wells, 1991). Light level had a significant impact 

on plant biomass accumulation (Figure 3.5, Table 3.5), suggesting that light is limiting under 

these conditions. One important caveat of this work is the fact that manipulating light levels has 

broader effects beyond simply altering total carbon available to the plant. Changes in light 

availability, such as what plants in vegetation canopies experience, induces changes in traits such 

as shoot architecture (Cescatti & Niinemets, 2004; Poorter et al., 2009) and chlorophyll content 

(Evans, 1993). The reduction in photosynthesis and thus growth due to low light reduces demand 

for soil nutrients (Cui & Caldwell, 1997). However, manipulating light levels offers a highly 

102 

 

feasible, ecologically relevant method of altering nutrient availability to examine its effects on 

mutualisms. Understanding the effect of shading on mutualisms is agriculturally important for 

the use of legumes in agroforestry (Houx et al., 2009) and for determining optimum planting 

density in commercial soybean crops (Pons & Pearcy, 1994). 

Because of the relative scarcity of literature regarding the impacts of both traded nutrients 

on plant-microbe nutritional symbioses, it is obvious that further research is required. It is 

particularly important to conduct rigorous testing of differences in optimal allocation patterns 

across different types of rhizobial nodules and different plant-microbe symbioses. This field 

offers a unique opportunity to use mathematical modeling and empirical testing to further our 

understanding of the stability of evolutionarily ancient symbioses (Clark et al., 2017).  In 

addition, this avenue of research will have important implications for understanding how 

anthropogenic nutrient deposition may affect plant-microbe symbioses both in agricultural and 

natural environments.  

 

 
 
 

 

 

 

 

 

 

 

 

 

103 

 

 

 

 

 

 

 

 

 

 

 

 

APPENDIX 

 

 

 

 

 

 

 

 

 

104 

 

 

 

SUPPLEMENTAL INFORMATION TO CHAPTER 3 

 

Figure 3.5 | Nodule efficiency in response to soil nitrogen. Shoot biomass gain (shoot biomass 
minus mean shoot biomass from uninoculated controls for each treatment) plotted against nodule 
biomass for four-week-old plants from Chapter 1. See Chapter 1 for full experimental details 
regarding these plants.  

 

 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

105 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

REFERENCES 

 

 

 

 

 

 

 

106 

 

REFERENCES 

 

Andrews M, Lea PJ, Raven JA, Azevedo RA. 2009. Nitrogen use efficiency. 3. Nitrogen  
fixation: genes and costs. Annals of Applied Biology 155: 1–13. 

Araujo KEC, Vergara C, Guimarães AP, Rouws JRC, Jantalia CP, Urquiaga S, Alves BJR, 
Boddey RM. 2018. Changes in 15N natural abundance of biologically fixed N2 in soybean due  
to shading, rhizobium strain and plant growth stage. Plant and Soil: 1–16. 

Bloom AJ, Chapin FS, Mooney HA. 1985. Resource limitation in plants-an economic analogy.  
Ann. Rev. Ecol. Sysl 16: 363–92. 

Bronstein JL. 1994. Conditional outcomes in mutualistic interactions. Trends in Ecology &  
Evolution 9: 214–7. 

Bronstein JL. 2001. The exploitation of mutualisms. Ecology Letters 4: 277–287. 

Bronstein JL (Ed.). 2015. Mutualism. Oxford: Oxford University Press. 

Brown MS, Bethlenfalvay GJ. 1987. Glycine-Glomus-Rhizobium Symbiosis : VI. 
Photosynthesis in Nodulated, Mycorrhizal, or N- and P-Fertilized Soybean Plants. Plant  
physiology 85: 120–123. 

Burkey KO, Wells R. 1991. Response of soybean photosynthesis and chloroplast membrane  
function to canopy development and mutual shading. Plant physiology 97: 245–52. 

Cescatti A, Niinemets Ü. 2004. Leaf to Landscape. In: Springer New York, 42–85. 

Chu ACP, Robertson AG. 1974. The effects of shading and defoliation on nodulation and  
nitrogen fixation by white clover. Plant and Soil 41: 509–519. 

Clark TJ, Friel CA, Grman E, Shachar-Hill Y, Friesen ML. 2017. Modelling nutritional 
mutualisms: challenges and opportunities for data integration (M van Baalen, Ed.). Ecology  
Letters 20: 1203–1215. 

Cui M, Caldwell MM. 1997. Shading reduces exploitation of soil nitrate and phosphate by 
Agropyron desertorum and Artemisia tridentata from soils with patchy and uniform nutrient  
distributions. Oecologia 109: 177–183. 

Denison RF. 2000. Legume Sanctions and the Evolution of Symbiotic Cooperation by Rhizobia.  
The American Naturalist 156: 567–576. 

Erisman JW, Galloway JN, Seitzinger S, Bleeker A, Dise NB, Petrescu AMR, Leach AM, 
de Vries W. 2013. Consequences of human modification of the global nitrogen cycle. 

107 

 

Philosophical transactions of the Royal Society of London. Series B, Biological sciences 368:  
20130116. 

Evans J. 1993. Photosynthetic Acclimation and Nitrogen Partitioning Within a Lucerne Canopy. 
II. Stability Through Time and Comparison With a Theoretical Optimum. Australian Journal of  
Plant Physiology 20: 69. 

Foley JA, Defries R, Asner GP, Barford C, Bonan G, Carpenter SR, Chapin FS, Coe MT, 
Daily GC, Gibbs HK, et al. 2005. Global consequences of land use. Science (New York, N.Y.)  
309: 570–4. 

Glyan’ko AK, Vasil’eva GG, Mitanova NB, Ishchenko AA. 2009. The influence of mineral  
nitrogen on legume-rhizobium symbiosis. Biology Bulletin 36: 250–258. 

Goh C-H, Nicotra AB, Mathesius U. 2016. The presence of nodules on legume root systems 
can alter phenotypic plasticity in response to internal nitrogen independent of nitrogen fixation.  
Plant, Cell & Environment 39: 883–896. 

Hansen AP, Gresshoff PM, Pate JS, Day DA. 1990. Interactions between Irradiance Levels, 
Nodulation and Nitrogenase Activity of Soybean cv. Bragg and a Supernodulating Mutant.  
Journal of Plant Physiology 136: 172–179. Journal of Plant Physiology 136: 172–179. 

Heath KD, Stock AJ, Stinchcombe JR. 2010. Mutualism variation in the nodulation response  
to nitrate. Journal of Evolutionary Biology 23: 2494–2500. 

van der Heijden MGA, Bardgett RD, van Straalen NM. 2008. The unseen majority: soil 
microbes as drivers of plant diversity and productivity in terrestrial ecosystems. Ecology Letters  
11: 296–310. 

Houx JH, Mcgraw RL, Fritschi FB, Navarrete-Tindall NE. 2009. Effects of shade on growth 
and nodulation of three native legumes with potential for use in agroforestry. Native Plants  
Journal 10: 232–238. 

Johnson NC, Graham JH, Smith FA. 1997. Functioning of mycorrhizal associations along the  
mutualism-parasitism continuum. New Phytologist 135: 575–585. 

Kaschuk G, Kuyper TW, Leffelaar PA, Hungria M, Giller KE. 2009. Are the rates of 
photosynthesis stimulated by the carbon sink strength of rhizobial and arbuscular mycorrhizal  
symbioses? Soil Biology and Biochemistry 41: 1233–1244. 

Kaschuk G, Leffelaar PA, Giller KE, Alberton O, Hungria M, Kuyper TW. 2010. 
Responses of legumes to rhizobia and arbuscular mycorrhizal fungi: A meta-analysis of potential  
photosynthate limitation of symbioses. Soil Biology and Biochemistry 42: 125–127. 

Kaschuk G, Yin X, Hungria M, Leffelaar PA, Giller KE, Kuyper TW. 2012. Photosynthetic 
adaptation of soybean due to varying effectiveness of N2 fixation by two distinct  

108 

 

Bradyrhizobium japonicum strains. Environmental and Experimental Botany 76: 1–6. 

Klinger CR, Lau JA, Heath KD. 2016. Ecological genomics of mutualism decline in nitrogen-
fixing bacteria. Proceedings of the Royal Society of London B: Biological Sciences 283:  
20152563. 

Korczynski. PC, Logan J, Faust JE. 1991. Mapping Monthly Distribution of Daily Light  
Integrals across the Contiguous United States. HortTechnology 12: 12–16. 

Lau JA, Bowling EJ, Gentry LE, Glasser PA, Monarch EA, Olesen WM, Waxmonsky J, 
Young RT. 2012. Direct and interactive effects of light and nutrients on the legume-rhizobia  
mutualism. Acta Oecologica 39: 80–86. 

Lucinski R, Polcyn W, Ratajczak L. 2002. Nitrate reduction and nitrogen fixation in symbiotic  
association Rhizobium - legumes. ACTA BIOCHIMICA POLONICA 49: 537–546. 

Neuhauser C, Fargione JE. 2004. A mutualism-parasitism continuum model and its application  
to plant-mycorrhizae interactions. Ecological Modelling 177: 337–352. 

Oono R, Denison RF, Kiers ET. 2009. Controlling the reproductive fate of rhizobia: how  
universal are legume sanctions? The New phytologist 183: 967–79. 

Pons TL, Pearcy RW. 1994. Nitrogen reallocation and photosynthetic acclimation in response  
to partial shading in soybean plants. Physiologia Plantarum 92: 636–644. 

Poorter H, Niinemets Ü, Poorter L, Wright IJ, Villar R. 2009. Causes and consequences of  
variation in leaf mass per area (LMA): a meta-analysis. New Phytologist 182: 565–588. 

Rainbird RM, Hitz WD, Hardy RW. 1984. Experimental determination of the respiration 
associated with soybean/rhizobium nitrogenase function, nodule maintenance, and total nodule  
nitrogen fixation. Plant physiology 75: 49–53. 

Ratcliff WC, Underbakke K, Denison RF. 2012. Measuring the fitness of symbiotic rhizobia. 
Symbiosis 55: 85–90. 

Reeve W, Chain P, O’Hara G, Ardley J, Nandesena K, Bräu L, Tiwari R, Malfatti S, Kiss 
H, Lapidus A, et al. 2010. Complete genome sequence of the Medicago microsymbiont Ensifer  
(Sinorhizobium) medicae strain WSM419. Standards in Genomic Sciences 2: 77–86. 

Regus JU, Gano KA, Hollowell AC, Sofish V, Sachs JL. 2015. Lotus hosts delimit the 
mutualism-parasitism continuum of Bradyrhizobium. Journal of Evolutionary Biology 28: 447– 
456. 

Santos VAF, Neves MCP, Rumjanek NG. 1997. Differential symbiotic efficiency by shading 
of soybean nodulated by B. japonicum and B. elkanii strains. Soil Biology and Biochemistry 29:  
1015–1018. 

109 

 

Schwartz MW, Hoeksema JD. 1998. Specialization and resource trade: Biological markets as a  
model of mutualisms. Ecology 79: 1029–1038. 

Silsbury JH. 1977. Energy requirement for symbiotic nitrogen fixation. Nature 267: 149–150. 

Simonsen AK, Han S, Rekret P, Rentschler CS, Heath KD, Stinchcombe JR. 2015. Short-
term fertilizer application alters phenotypic traits of symbiotic nitrogen fixing bacteria. PeerJ 3:  
e1291. 

Streeter J, Wong PP. 1988. Inhibition of legume nodule formation and N 2 fixation by nitrate.  
Critical Reviews in Plant Sciences 7: 1–23. 

Thornley JHM. 1972. A Model to Describe the Partitioning of Photosynthate during Vegetative  
Plant Growth. Annals of Botany 36: 419–430. 

Udvardi M, Poole PS. 2013. Transport and metabolism in legume-rhizobia symbioses. Annual  
review of plant biology 64: 781–805. 

Voisin A-S, Salon C, Munier-Jolain NG, Ney B. 2002. Effect of mineral nitrogen on nitrogen 
nutrition and biomass partitioning between the shoot and roots of pea (Pisum sativum L.). Plant  
and Soil 242: 251–262. 

Weese DJ, Heath KD, Dentinger BTM, Lau JA. 2015. Long-term nitrogen addition causes the 
evolution of less-cooperative mutualists. Evolution; international journal of organic evolution  
69: 631–42. 

Young ND, Debellé F, Oldroyd GED, Geurts R, Cannon SB, Udvardi MK, Benedito VA, 
Mayer KFX, Gouzy J, Schoof H, et al. 2011. The Medicago genome provides insight into the  
evolution of rhizobial symbioses. Nature 480: 520–524. 

 
 
 
 
 
 
 
 
 
 

110 

 

 
 
 
 
 
 

CHAPTER 4 

 

HOST SPECIFICITY IN THE TRIFOLIUM-RHIZOBIUM SYMBIOSIS 

Abstract 

 

 

Interactions with microbial mutualists may be able to promote plant coexistence if these 

interactions allow plants to occupy different niches. Plants may also be able to directly affect the 

composition of the soil microbial community, which then feeds back into plant fitness in a 

process known as plant-soil feedback. There is limited empirical evidence for the role of 

microbial mutualists in the promotion of plant coexistence, and it is focused mainly on plant-

fungus interactions. Here, we assess the role rhizobia (soil bacteria that colonize plant roots and 

fix nitrogen in exchange for photosynthetic carbon) may play in the promotion of coexistence 

between 8 native species of Trifolium clover at our field site at Bodega Bay, California. We find 

that while both plants and rhizobia from this ecosystem are specialists, there is no clear pattern of 

increased benefit from or investment into conspecific rhizobia (those isolated from the same 

species as the test host), which we expected to perform better due to co-adaptation with their 

preferred hosts. We find that Trifolium species vary both in the amount they invest into rhizobia 

overall as well as the efficiency with which they extract benefit from rhizobia, leading to large 

differences in overall response to rhizobia. These results have important implications for our 

understanding of rhizobia’s role in this ecosystem and the development of effective crop 

inoculants.  

 

 

111 

 

Introduction 

Biodiversity in an ecosystem, both at the genetic and species level, is vital for ecosystem 

resilience to disturbance, response to environmental change, and productivity (Hughes & 

Stachowicz, 2004; Cardinale et al., 2006, 2009; Fridley et al., 2007; Hughes et al., 2008; Cadotte 

et al., 2009). Diversity at any level is maintained by a given species or allele increasing in 

relative abundance when it becomes rare in the system (Levene, 1953; Levine & 

HilleRisLambers, 2009). The ability to increase when rare depends on niche differentiation, or 

differences in the way organisms interact with or affect their environment (Chesson, 2000; Adler 

et al., 2010; Turnbull et al., 2013). Niche differentiation means that a given species will limit 

itself more than its competitors (Chesson, 2000). Plants may be able to engage in niche 

differentiation through interactions with beneficial soil microbes that allow them to specialize in 

various ways of interacting with the biotic and abiotic environment (Peay, 2016).   

Ecological theory has presented two theories for how interactions with mutualists may 

promote coexistence. Parker, (1999) suggested that co-adaptation between host and symbiont can 

lead to a stable mosaic pattern. In this scenario, different physical locations are occupied by 

stable groups of co-adapted plants and microbes that resist invasion by outsiders, maintaining 

diversity through patchy spatial distribution (Parker, 1999). Bever, (1999) suggested that 

diversity can be maintained by microbial mutualists when the microbe that receives the most 

benefit from plant species A provides the most benefit to plant species B, resulting in negative 

feedback between plant species A and the microbe to which plant species A gives the most 

benefit (Bever, 1999). This negative feedback then maintains diversity of the plant species.  

Niche differentiation on microbial partners requires the ability of plants to specialize in 

their interactions with microbes. If there is a disconnect between the amount of benefit a microbe 

112 

 

receives from its plant partner and the amount of resources it extracts from that plant, then these 

negative plant-soil feedbacks can maintain diversity between plant species as discussed above 

(Bever, 1999). In contrast, if a plant can select for its most beneficial microbes and preferentially 

amplify those microbes, then the plant may be able to engage in positive plant-soil feedbacks. In 

this case, the plant must condition the soil or alter the composition of the microbial community 

in such a way that makes that community more beneficial to that plant species than to other plant 

species (Bever, 2003; Hodge & Fitter, 2013). In these cases, plants are engaging in niche 

differentiation through their impacts on the environment, a process known as niche construction 

(van der Putten et al., 2013). Plant-soil feedback can have major impacts on plant competition, 

altering whether conspecifics or heterospecifics are more limiting competitors (Pendergast et al., 

2013). Most examples of plant-soil feedback tend to be negative (Hodge & Fitter, 2013), 

resulting from the accumulation of pathogens (e.g. Klironomos, 2002). However, positive plant-

soil feedbacks have been reported between neotropical trees and their mycorrhizal fungi 

(Mangan et al., 2010).  

There is relatively little empirical research on the effects of plant-microbe mutualisms on 

niche differentiation and host coexistence, and most of this research has focused on plant-fungus 

interactions (Peay, 2016). Fungal endophytes reduce the impact of drought stress on the grass 

Bromus laevipes and allow it to occupy habitats that are too dry for uninfected individuals 

(Afkhami et al., 2014). In contrast, a fungal endophyte of the grass Poa leptacoma limits its host 

range at early life stages and promotes growth at later life stages (Kazenel et al., 2015). Diverse 

communities of arbuscular mycorrhizal fungi can reduce interspecific competition between 

plants by reducing competition for soil resource (Wagg et al., 2011). However, the role of other 

plant microbial symbionts in niche differentiation is unclear. 

 

113 

 

The legume-rhizobia symbiosis represents a thus-far untapped resource for exploring the 

effects of microbial symbionts on niche differentiation and plant-soil feedbacks. In this 

interaction, nitrogen-fixing rhizobia induce the formation of nodules on legume roots. Inside the 

nodules, the rhizobial bacteria fix atmospheric dinitrogen into ammonia for export to the plant in 

exchange for photosynthetically fixed carbon (Poole et al., 2018). Legumes are a highly diverse 

group, and multiple legume species often co-occur (Sprent et al., 2017). Co-occurring legumes 

can share symbionts that may vary in effectiveness between hosts (Ehinger et al., 2014). There 

are examples of both highly specialized and promiscuous symbionts (Lira et al., 2015). There is 

also evidence of legumes being able to differentiate between cooperative and uncooperative 

rhizobia (Kiers et al., 2003; Yates et al., 2005, 2008; Gubry-Rangin et al., 2010). Thus, the 

legume-rhizobia symbiosis fulfills all the requirements for niche differentiation and/or plant-soil 

feedbacks as described above. The ability of legumes to differentiate between cooperative and 

uncooperative rhizobia suggest that Parker (1999)’s theory of co-adaptation is more likely than 

Bever (1999)’s to hold true in this system.  

 

Our field site at Bodega Bay Marine Reserve in Bodega Bay, CA provides an excellent 

model to explore rhizobial axes of niche differentiation and test for plant-soil feedbacks. We 

have more than eight years of field data documenting the coexistence of eight native species of 

Trifolium in 330 2x2 m plots at this field site. All rhizobia that have been isolated from Trifolium 

nodules belong to Rhizobium leguminosarum bv. trifolii (Jordan, 1984). Members of this species 

exhibit geographic specificity--strains isolated from Africa, Europe, and North America 

generally cannot effectively colonize plants from different continents (Howieson et al., 2005). In 

addition, strains from perennial species cannot effectively colonize annual species and vice versa 

(Howieson et al., 2005). When strains from one Trifolium host can colonize a different host, 

114 

 

there is wide variation in the fitness benefits provided by a given strain across different hosts 

(Holland & Parker, 1966; Holland, 1970; Carlsson et al., 2006). All of this suggests that 

interactions with rhizobia are a key aspect of niche differentiation in Trifolium, and that genetic 

diversity in the Rhizobium population at our field site may support plant diversity. 

We explored the role that rhizobia may play in niche differentiation and plant-soil 

feedback at Bodega Bay by testing the benefits gained from and resources invested into 32 field-

isolated rhizobial strains across all 8 Trifolium species. The rhizobia strains were isolated from 

the 7 Trifolium species with sufficient population size to destructively sample in the field. Thus, 

for these 7 species, we were able to contrast the effects of interacting with conspecific rhizobia 

(rhizobia for which the isolation host and test host are the same plant species) versus 

heterospecific rhizobia (rhizobia for which the isolation host and test host are different plant 

species). This setup allowed us to test the following hypotheses: 

1.  The plants and rhizobia in this experiment will tend to be specialists. Specialists will 

benefit more from their partners than generalists since they are co-adapted for higher 

levels of benefit.  

2.  If Trifolium specialize on rhizobia and are able to preferentially amplify their most 

beneficial strains, plant performance and investment in rhizobia should be higher when 

plants are inoculated with conspecific rhizobia than with heterospecific rhizobia.  

3.   Trifolium species that are more common in the field will have had a chancer to 

preferentially amplify their most beneficial strains in a larger proportion of the microbial 

community than rare species. Thus, common species are expected to extract more benefit 

from and invest more resources into rhizobia than rare species.  

115 

 

The results from this experiment will provide insight into the role of rhizobia in promoting plant 

diversity and shaping the community composition of ecosystems that include legumes (Keller, 

2014). In addition, rhizobial specialization and specificity is an important phenomenon to 

understand for the development of effective rhizobial crop inoculants, which are frequently 

outcompeted by promiscuous and ineffective native rhizobia (e.g. Gerding et al., 2014). 

 

Materials and Methods 

Pot preparation 

We filled SC10 Cone-Tainers (Steuwe and Sons Inc., Corvallis, OR, USA) with a 1:1:1 

mixture (by volume) of medium vermiculite, greenhouse sand, and Turface. We triple autoclaved 

the pots for 60 minutes with 24 hours between cycles to sterilize. 

  

Seedling preparation 

We scarified seeds of greenhouse-grown Trifolium barbigerum, T. bifidum, T. fucatum, T. 

macraei, T. microdon, T. microcephalum, and T. willdenovii by nicking the seed coat with a 

clean razor blade. We sterilized the seeds with chlorine gas by placing the seeds in a dessicator 

with a beaker of 100 mL of full strength commercial bleach (8.25% NaHClPO3), to which we 

added 3 mL of HCl. We immediately sealed the dessicator and incubated the seeds in the gas for 

4 hours. After 4 hours, we transferred seeds to sterile water agar plates sealed with Parafilm and 

incubated them in the dark at 4°C for 48 hours. We then germinated the seeds at room 

temperature in the dark for 24-48 hours. Seedlings with radicles at least 1 cm long were 

transplanted into prepared pots. After planting, the plants were fully randomized and grown in 

116 

 

the greenhouse at Michigan State University from [date] to [date]. Artificial lighting maintained 

a 16 hour day/8 hour night cycle.    

Rhizobia isolation 

We isolated rhizobia from nodules of field-collected Trifolium plants from our field site 

in Bodega Bay, California. We randomly sampled plants from across the field site and nodules 

were selected to represent the range of nodule size on a given plant. We washed the root systems 

with dH2O on ice to remove soil and other debris. We sterilized each root system in 10% 

commercial bleach (0.825% NaHClPO3) for 2 minutes, then rinsed 6 times with cold sterile 

dH2O. We plucked individual nodules with flame sterilized forceps and crushed them with a 

plastic pestle in 150 uL of sterile crushing buffer (a 1:1 mixture of ½x phosphate buffered saline 

(PBS) and 80% glycerol). We streaked the resulting suspension onto tryptone yeast (TY) plates 

and incubated the plates at 30°C for 48-72 hours. Individual colonies were restreaked 2-3 times. 

A single colony was plucked and grown in liquid TY and frozen in 40% glycerol at -80°C for 

future use.  

Rhizobia inoculation 

We streaked 35 strains from the strain collection into a lawn onto replicate TY plates and 

grown at 30°C for 48 hours. These 35 strains represented 5 strains isolated from each of 7 plant 

species. We were unable to sample rhizobia from T. microcephalum plants because so few plants 

are found in the field that this destructive sampling would have major effects on their population 

numbers. Three strains did not grow sufficiently for inoculation, so the experiment was 

conducted with 5 strains each from T. barbigerum, T. fucatum, T. gracilentum, and T. willdenovii 

and 4 strains each from T. bifidum, T. macraei , and T. microdon. These 32 strains were 

inoculated factorially across all 8 plant species with 4 replicates per condition. The bacterial 

117 

 

lawn was gently scraped from the plate and resuspended in 1 mL of  ½x PBS.  We determined 

cell density by measuring the OD600 of the culture using a NanoDrop. After one week of 

growth, plants received 1 mL of 107 CFU/mL rhizobial cells suspended in ½x phosphate 

buffered saline (PBS) and rhizobia-free control plants received 1 mL of sterile ½x PBS. We 

checked the sterility of the mock inoculum and viability of the rhizobial inoculum using spot 

plating on TY agar. 

Plant growth and harvest 

 

We planted the experiment February 13-16, 2017 and harvested all plants April 17-20, 

2017. Each plant was given 2 pellets of Osmocote 14-14-14 slow release fertilizer on March 24, 

2017 to provide a small amount of nutrients to maintain plant growth without disrupting nitrogen 

fixation. Plants were watered as needed by automatic misting from above. Misting was adjusted 

throughout the growth period to ensure adequate moisture in the pots without over-watering. At 

harvest, we separated the root, shoot, and flower tissue and counted and separated nodules from 

the root system using forceps. Tissue was dried at 60°C for one week and weighed. Only one 

uninoculated control developed nodules, and this plant was excluded from further analysis.   

Statistical analysis 

To assess the degree of specialization of the plants and rhizobia, we calculated the paired 

difference index (PDI; bipartite package, R 3.4.3). PDI is a highly robust and informative 

measure of specialization that ranges from 0-1, with larger values indicating increasing 

specialization (Poisot et al., 2012). PDI is calculated use the formula  

 

 

118 

 

Where P1 is the largest link strength (strength of the interaction with a given resource), Pi is the 

link strength with the ith resource, and R is the number of resources (Poisot et al., 2012). From 

the plant perspective, each rhizobial strain is a resource, and each plant species is a resource for 

the rhizobia. We used scaled shoot biomass, specific nodulation, and nodule:root biomass ratio 

as link strengths to calculate three PDI values for each plant species and rhizobial strain. Scaled 

shoot biomass PDI (sPDI) represents specificity in the amount of benefit received or given, 

specific nodulation PDI (snPDI) represents specificity in the initiation of nodules, and 

nodule:root biomass PDI (nbPDI) represents specificity in the total amount of biomass allocated 

to/received by rhizobia. We used a linear model ANOVA to test whether isolation host affected 

sPDI, snPDI, and nbPDI. We also used linear model ANOVA to test whether any PDI measure 

affected mean and maximum scaled shoot biomass, specific nodulation, and nodule:root 

biomass. Maximum responses were calculated in a two-step process: first we calculated the mean 

response for each strain-plant species combination. We then selected the combination with the 

highest mean for each plant species or rhizobial strain. This allowed us to assess maximum 

possible investment while avoiding using a single plant as a data point, which could lead to 

excessive variation. The response variables in all of these models were log-transformed to 

improve normality and model fit. The p-values for the effects of various PDI measures on plant 

responses were Bonferroni-corrected to account for multiple testing. 

We tested the effects of conspecific versus heterospecific rhizobia on plant performance 

using scaled shoot biomass (shoot biomass of an individual divided by the mean shoot biomass 

of the uninoculated controls of that species), flowering success (a binary yes/no measure of 

whether plants flowered or not) and flower number as proxies for plant fitness. In all cases, we 

included main effects of interaction type (conspecific versus heterospecific), test species, and 

119 

 

their interaction as predictors in the model. We used a linear model ANOVA for scaled shoot 

biomass and log transformed scaled shoot biomass to improve normality. We used a generalized 

linear model ANOVA with a binomial distribution for flowering success and a generalized linear 

model ANOVA with a Poisson distribution for flower number. Since flowering success was only 

dependent on plant species (Table 1), so we only assessed effects on flower number in plant 

species that had at least 40% of individuals flower (Table 4.9).   

We also tested whether plants invest most in conspecific or heterospecific rhizobia using 

nodule number and nodule biomass. In both cases, we included main effects of interaction type, 

test species, and nodule counter (the identity of the researcher who plucked and counted 

nodules), and the interaction type x test species interaction term as predictors. We used a 

generalized linear model ANOVA with a Poisson distribution for nodule number and linear 

model ANOVA for nodule biomass. We log-transformed nodule biomass to improve normality.  

We confirmed that patterns of investment between conspecific and heterospecific 

rhizobia were due to specificity in the interaction by testing the effects of isolation host (the plant 

species from which a rhizobial strain was isolated) on nodule number and biomass. In these 

models, we used the main effects of isolation host, test species, and nodule counter, and the 

interaction between isolation host and test species as predictors. We used a generalized linear 

model ANOVA with a Poisson distribution for nodule number and a linear model ANOVA for 

biomass. We log-transformed nodule biomass to improve normality.  

To assess plant response to rhizobia, we measured scaled shoot biomass, specific 

nodulation (nodule number per g root biomass), and nodule:root biomass (mg nodule biomass 

per g root biomass) for each plant species averaged across all rhizobia treatments. For these 

models, we used a general linear model ANOVA with plant species as a predictor. We log-

120 

 

transformed all of the response variables to improve normality. We tested whether response to 

rhizobia increases with commonness in the field with a linear model of each response as 

predicted by the fraction of plots in the field in which each species is present. We log-

transformed the response variables to improve normality and model fit. 

We assessed the efficiency of the rhizobia on each plant species in two different ways. To 

assess raw efficiency, we used the raw increase in shoot biomass (shoot biomass of the 

individual minus the mean shoot biomass of uninoculated controls of that plant species) as the 

response. We also assessed scaled efficiency by using scaled shoot biomass as the response. We 

used general linear model ANOVA with main effects of nodule biomass and test species and the 

nodule biomass x test species interaction as predictors. Since we were particularly interested in 

the linear relationship between nodule biomass and shoot biomass, we did not log-transform the 

response variables to improve normality. While coefficients of course varied between models 

with log-transformed and untransformed responses, there was no difference in whether each term 

in the model was significant. To visualize the relationship between nodule biomass and shoot 

biomass across species, we bootstrapped the nodule biomass coefficient for each plant species 

and displayed the mean and 95% CI (20,000 replicates; bias corrected and accelerated (Bca) 

output, boot package, R 3.4.3). We also assessed the linear relationship between nodule biomass 

and the raw increase in shoot biomass for each plant species by testing for correlation with 

Spearman’s correlation coefficient (cor.test function, R 3.4.3). P values for the correlation tests 

were Bonferroni-corrected to account for multiple testing.  

All linear models used the lm function (R 3.4.3), all generalized linear models used the 

glm function (R 3.4.3), and all ANOVAs were conducted with Type II sum of squares using the 

121 

 

Anova function (car package, R 3.4.3). Post-hoc testing was conducted with Tukey tests with the 

emmeans function (emmeans package, R 3.4.3). 

 

Results 

Plants and rhizobia in this ecosystem are mostly specialists 

We found that all plant species were specialists (PDI > 0.5) when assessed by scaled 

shoot biomass PDI (sPDI) and specific nodulation PDI (snPDI) (Figure 4.1 a,c). When assessed 

by nbPDI, T. fucatum was found to be slightly generalist (nbPDI of 0.47), while all other plant 

species were specialists (nbPDI > 0.5; Figure 4.1e). 

 

We found that most rhizobia were specialists, but the pattern depended on the type of PDI 

being calculated. For sPDI, 13/32 (40%) of strains were generalists (sPDI < 0.5), while 19/32 

(60%) were specialists (sPDI > 0.5; Figure 4.1b). For snPDI, 5/32 (16%) of strains were 

generalists (snPDI < 0.5), while 27/32 (84%) of strains were specialists (snPDI > 0.5; Figure 

4.1d). For nbPDI, all rhizobia were specialists (nbPDI > 0.5; Figure 4.1f). We did not detect a 

significant effect of isolation host on any measure of rhizobial PDI (Table 4.1). These results 

should be interpreted with caution because we were unable to calculate 95% confidence intervals 

for the PDI values and thus cannot determine how confident we are in the division of organisms 

between specialist and generalist. In addition, it is unlikely that there is a meaningful biological 

difference between an organism with a PDI of 0.49 and 0.51, though one would be classified as a 

weak generalist and one as a weak specialist. However, PDI is still useful for providing an 

overview of the patterns of generalization and specialization.  

 

 

122 

 

 

Figure 4.1 | Specialization of plants and rhizobia as measured by Paired Difference Index 
(PDI). PDI for each plant species calculated with a) scaled shoot biomass, c) specific nodulation, 
and e) nodule:root biomass. PDI for each rhizobia strain calculated with b) scaled shoot biomass, 
d) specific nodulation, and f) nodule:root biomass. 
 
Table 4.1 | ANOVAs summarizing the effect of isolation host on rhizobial PDI.  sPDI 
indicates scaled shoot biomass PDI, snPDI indicates specific nodulation PDI, and nbPDI 
indicates nodule biomass PDI.  
  
  
Isolation host 1.83 0.133   1.73 0.157   1.35 0.274 
 

   snPDI 
  F(6,25)  p 

sPDI 
F(6,25)  p 

   nbPDI 
  F(6,25)  p 

123 

 

No effect of plant specialization on investment in or benefit from rhizobia 

We did not detect a significant effect of any plant PDI measure on mean or maximum 

scaled shoot biomass, specific nodulation, or nodule:root biomass (Figure 4.2). See Table 4.2 for 

a summary of all p-values.  

Effects of rhizobial specialization depend on the response being measured 

We detected significant negative effects of rhizobial sPDI, snPDI, and nbPDI on 

nodule:root, but did not detect a significant effect of these PDIs on mean scaled shoot biomass, 

maximum specific nodulation, or nodule:root biomass (Figure 4.3, Table 4.3). Other effects of 

rhizobial specialization depend on the response being measured. For instance, we detected a 

significant positive effect of rhizobial sPDI but not snPDI or nbPDI on maximum scaled shoot 

biomass. Furthermore, we did not detect a significant effect of rhizobial sPDI on mean specific 

nodulation, but we did detect significant negative effects of rhizobial snPDI and nbPDI. See 

Table 4.3 for a summary of all p-values.  

 

 

 

 

 

 

 

 

 

124 

 

 

Figure 4.2 | Effect of plant specialization on response to rhizobia.  Effect size of various plant 
PDI measures (y axis) on plant responses (facets)in a linear model of the form log(response) ~ 
PDI. A negative value in this figure indicates a negative effect of PDI on the response, and a 
positive value indicates a positive effect of PDI on the response. Since regressions were 
performed with log transformed responses, the mean of the response is multiplied by exp(effect 
size) when PDI increases by one unit. Triangles with dotted lines indicate a nonsignificant effect 
of the predictor on the response (p > 0.05 after Bonferroni correction) and circles with solid lines 
indicate a significant effects of the predictor on the response (p < 0.05 after Bonferroni 
correction). See Figure 4.9 and 4.10 for scatterplots. 
 

Table 4.2 | ANOVAs summarizing the effects of plant PDI measures on plant responses to 
rhizobia. Note that predictors are the rows and responses the column; a separate ANOVA was 
conducted for each predictor-response combination, while in other tables each row contains a 
predictor that was included in a single model for each response. P values were Bonferroni 
corrected to account for multiple testing.  
  

 
 
Mean specific 

Mean scaled 

Max scaled 

Mean 

 

 

 

Max specific 
nodulation 

shoot 

nodulation 

nodule:root 

shoot biomass 

biomass 
p 
F(1,6) 
  
sPDI 
0.315  <1 
snPDI  0.139  <1 
nbPDI  0.501  <1 
 

 
 
 
 

 
 
 
 

p 
F(1,6) 
0.134  <1 
0.198  <1 
0.446  <1 

biomass 
p 

F(1,6) 
0.152  0.977 
3.34 
3.66  0.937 

<1 

 
 
 
 

p 

F(1,6) 
3.59  0.961 
0.175 
0.997 

>1 
>1 

 
 
 
 

p 

F(1,6) 
6.52  0.390 
4.97  0.606 
6.37  0.405 

125 

 

Max 

nodule:root 

biomass 
F(1,6) 
0.394 
0.283 
0.193 

p 
>1 
>1 
>1 

 
 
 
 

 

Figure 4.3 | Effect of rhizobial specialization on plant responses. Effect size of rhizobia PDI 
measures(y axis) on plant responses (facets)  in a linear model of the form log(response) ~ PDI. 
A negative value in this figure indicates a negative effect of PDI on the response, and a positive 
value indicates a positive effect of PDI on the response. Since regressions were performed with 
log transformed responses, the mean of the response is multiplied by exp(effect size) when PDI 
increases by one unit. Triangles with dotted lines indicate a nonsignificant effect of the predictor 
on the response (p > 0.05 after Bonferroni correction) and circles with solid lines indicates a 
significant effect of the predictor on the response (p < 0.05 after Bonferroni correction). See Fig 
4.11 and 4.12 for scatterplots.  
 
Table 4.3 | ANOVAs summarizing the effects of rhizobial PDI measures on plant responses 
to rhizobia. Note that predictors are the rows and responses the column; a separate ANOVA was 
conducted for each predictor-response combination, while in other tables each row contains a 
predictor that was included in a single model for each response. P values were Bonferroni 
corrected to account for multiple testing.  
  

 
 
Mean specific 

Mean scaled 

Mean 

Max scaled 

 

 

 

Max specific 
nodulation 

shoot 

nodulation 

nodule:root 

shoot biomass 

biomass 
p 
F(1,30) 
0.213  >1 

  
sPDI 

snPDI  0.831  >1 
nbPDI  1.51 
>1 
 
 

 
 

 
 

p 

F(1,30) 
8.70  0.055 

14.50  0.006 
13.1  0.010 

 
 

 
 

biomass 
p 

F(1,30) 
11.99  0.015 

12.0  0.015 
13.9  0.007 

 
 

 
 

p 

F(1,30) 
41.7  3.51e-

06 
>1 
>1 

0.372 
2.22 

 
 

 
 

p 

F(1,30) 
6.55  0.142 

>1 

1.73 
4.45  0.391 

 
 

 
 

126 

 

Max 

nodule:root 

biomass 
p 

F(1,30) 
8.36  0.064 

5.91  0.191 
3.92  0.514 

Positive effect of conspecific rhizobia on flower number but not shoot biomass 

We detected a significant effect of test species on scaled shoot biomass (p < 2e-16), but 

we did not detect a significant main effect of interaction type (p = 0.890) or an interaction 

between interaction type and test species (p = 0.731; Table 1, Figure 4.4a, Figure 4.14a). We 

detected a significant effect of test species on flowering success (p < 2e-16), but we did not 

detect a significant effect of interaction type (p = 0.989) or a significant interaction between 

interaction type and test species (p = 0.849). Because the probability of flowering was only 

affected by plant species, and some species had such low flowering rates as to prevent 

meaningful analysis, we limited flower number analysis to the three plant species that had at 

least 40% of individuals flower (Table 4.9). Mean flower number in plants inoculated with 

heterospecific rhizobia was approximately 54% of mean flower number in plants inoculated with 

conspecific rhizobia (p = 0.010). We also detected a significant effect of plant species (p < 2e-

16) on flower number in this reduced dataset, but we did not detect a significant interaction 

between interaction type and plant species (p = 0.523; Table 1, Figure 4.4b, Figure 4.14b).  

 

 

 

 

 

 

 

 

 

127 

 

Table 4.4 | ANOVAs summarizing the effects of interaction type and test species on plant 
performance. A linear model was used to assess scaled shoot biomass, while a generalized 
linear model with a binomial distribution was used to assess flowering probability and a 
generalized linear model with a Poisson distribution was used to assess flower number. Note that 
the full dataset was used for scaled shoot biomass and flowering probability models, while only 
T. fucatum, T. macraei, and T. microdon were included in the model for flower number since the 
other plant species exhibited very low flowering rates.  
  
  
  
Interaction type (IT) F(1,1039) = 0.019  0.890    χ2
F(7,1039) = 13.6  < 2e-16   χ2
Test species (TS) 
F(6,1039) = 0.600  0.731    χ2
IT*TS 
 

Scaled shoot biomass    Flowering probability    Flower number 
p 

(1,405) =6.63  0.010 
(2,405)= 97.6 <2e-16 
(2,405) =1.30  0.523 

χ2 

χ2 

p 

  
(1,1039) =0.989  0.320    χ2
(7,1039)= 250.1 <2e-16   χ2
(6,1039) =2.67  0.849    χ2

F 

p 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

128 

 

Species-dependent differences in nodulation between conspecific and heterospecific rhizobia 

Interaction type, test species, and their interaction significantly affected nodule number (p 

= 7.78e-08, p < 2.2e-16, and p < 2.2e-16, respectively; Table 4.5). We did not detect a significant 

main effect of interaction type on nodule biomass (p = 0.320), but we did detect a significant 

effect of test species (p < 2.2e-16) and a significant interaction between interaction type and test 

species (p = 2.41e-04; Table 4.5). After Tukey testing, we did not detect a significant difference 

in nodule number or nodule biomass in T. barbigerum plants inoculated with conspecific versus 

heterospecific rhizobia ((p = 0.482 and p = 0.839, respectively; Figure 4.4). T. bidifum, T. 

microdon, and T. willdenovii make significantly fewer nodules with conspecific rhizobia than 

with heterospecific rhizobia (p < 0.001, p = 0.014, and p = 0.024, respectively; Figure 4.4a). We 

did not detect a significant difference in nodules biomass between plants inoculated with 

conspecific versus heterospecific rhizobia for these three species (p = 0.288, p = 0.718, and p = 

0.815, respectively; Figure 4.4b). T. fucatum, T. gracilentum, and T. macraei made significantly 

more nodules with conspecific rhizobia than with heterospecific rhizobia (p < 0.001, p = < 

0.001, p = 0.003, respectively; Figure 4.4a). T. fucatum had significantly higher nodule biomass 

when inoculated with conspecific rhizobia than with heterospecific rhizobia (p < 0.001), but we 

did not detect a significant difference in nodule biomass between T. gracilentum and T. macraei 

inoculated with conspecific versus heterospecific rhizobia (p = 0.630 and p = 0.815, 

respectively; Fig 4.4b).  

 

 

 

 

129 

 

 

 
Figure 4.4 | Plant performance and nodulation with conspecific versus heterospecific 
rhizobia. a) Mean scaled shoot biomass across all test species. b) Mean flower number across 
the three test species analyzed for flowering. In a) and b) error bars represent +/- 1 SE. Bars with 
the same letter in a given panel are not statistically different (p > 0.05). c) Ratio of estimated 
marginal mean nodule number and d) Ratio of estimated marginal mean nodule biomass of 
plants inoculated with conspecific versus heterospecific rhizobia. The point range represents +/- 
1 SE. Dotted lines with triangles indicate nonsignificant differences between mean responses of 
plants inoculated with conspecific and heterospecific rhizobia after Tukey testing (p > 0.05), 
while solid lines with circles indicate significant differences between mean responses of plants 
inoculated with conspecific and heterospecific rhizobia after Tukey testing (p < 0.05). 
 

Table 4.5 ANOVAs summarizing the effects of interaction type and test species on 
nodulation. Results of generalized linear model ANOVA (nodule number) and linear model 
ANOVA (nodule biomass) summarizing the effects of interaction type and plant species on 
nodulation. Nodule counter was included to control for variation in nodule counting accuracy 
among researchers but is not of experimental interest. 
  
  
Interaction type (IT) 
Test species (TS) 
IT*TS 
Nodule counter 
 

χ2
χ2
(7, 1030) =2.80e04    <2.2e-16   F(7,1030)=210.5 <2.2e-16 
χ2
(6, 1030) =171.9     <2.2e-16   F(6,1030)=4.35  2.41e-04 
χ2
(3, 1030) =143.7  <2.2e-16   F(3,1030)=2.62  0.050 

7.78e-08   F(1,1030)=3.01  0.083 

Nodule number 

   Nodule biomass 
p 
  

F 

p 

χ2 

(1,1030) =28.9 

130 

 

Specificity of interactions 
 

To confirm that detected nodulation trends were due to specificity in the interactions and 

not because of uniformly poor colonization by any particular strain, we tested the effect of 

isolation host on nodulation measures. We detected a significant main effect of isolation host and 

test species (p < 2.2e-16 for both) on nodule number as well as a significant interaction between 

isolation host and test species (p < 2.2e-16; Table 4.6). We then determined the number of test 

species for which a given isolation host is the worst isolation host (i.e., that isolation host is 

statistically indistinguishable via Tukey testing from the isolation host with the smallest mean) 

and the best isolation host (i.e., that isolation host is statistically indistinguishable via Tukey 

testing from the isolation host with the largest mean). No isolation host was either the best nor 

the worst isolation host for all 8 test species, suggesting that their nodulation trends on 

conspecific hosts are not a universal characteristic (Table 4.7, Figure 4.5). In the most extreme 

example, T. bifidum was the worst isolation host for 7 out of 8 test species but was the best 

isolation host for the remaining test species (Table 4.7). Thus, even though these strains are 

generally poor colonizers, they are capable of relatively high colonization on at least one host.  

 

We also assessed the effect of isolation host on nodule biomass. We detected significant 

main effects of isolation host and test species on nodule biomass and a significant interaction of 

the two (p =2.2e-07, p < 2.2e-16, and p =6.30e-05, respectively; Table 4.6). After Tukey testing, 

there were no significant differences in nodule biomass between isolation hosts for any other test 

species than T. fucatum, confirming that the high nodule biomass in T. fucatum individuals 

inoculated with conspecific strains is a specific interaction and not a universal feature of the 

rhizobia (Figure 4.6).  

 

131 

 

 
Figure 4.5 | Effects of isolation host on nodule number across all test species. Nodule number 
of each combination of isolation host (x axis; the plant species from which a given strain was 
isolated) and test host (facet; the plant species on which a strain was inoculated). Within a given 
facet, bars with the same letter do not significantly differ after Tukey testing (p > 0.05). Note that 
each facet has a different y axis range.   
 
Table 4.6 ANOVAs summarizing the effects isolation host and test species on 
nodulation. Results of generalized linear model ANOVA (nodule number) and linear model 
ANOVA (nodule biomass) summarizing the effects of isolation host and test species on 
nodulation. Nodule counter was included to control for variation in nodule counting accuracy 
among researchers but is not of experimental interest. 
 

  
Isolation host (IH) 
Test species (TS)  χ2
IH*TS 
Nodule counter 

χ2

(6,989) =727 

Nodule number 

χ2 

   Nodule biomass 
p 
  

F 

p 

<2.2e-16   F(6,989)=7.07  2.2e-07 
(7, 989) =2.81e04    <2.2e-16   F(7,989)=224  <2.2e-16 
χ2
(42, 989) =969     <2.2e-16   F(42,989)=2.11 6.30e-05 
χ2

(3, 989) =135  <2.2e-16   F(3,989)=2.15  0.092 

132 

 

4 
1 
5 
2 
5 
0 
2 

Table 4 .7 | Summary of number of test species for which each isolation host was the worst 
isolation host or best isolation host. Worst isolation host indicates mean nodule number was 
statistically indistinguishable from the lowest mean nodule number, p > 0.05 and best isolation 
host indicates mean nodule number was statistically indistinguishable from the highest mean 
nodule number, p > 0.05. See Figure 4.5 for full data.  
  
Worst isolation host Best isolation host 
T. barbigerum 
T. bifidum 
T. fucatum 
T. gracilentum 
T. macraei 
T. microdon 
T. willdenovii 
 
 
 
 
 
 
 
 
 
 
 

1 
7 
1 
2 
0 
5 
2 

133 

 

Figure 4.6 | Effects of isolation host on nodule biomass across all test species. Nodule 
biomass of each combination of isolation host (x axis; the plant species from which a given strain 
was isolated) and test host (facet; the plant species onto which a strain was inoculated). Within a 
given facet, bars with the same letter do not significantly differ after Tukey testing (p > 0.05). 
Note that each facet has a different y axis range.   
 

 

 

 

 

 

 

 

134 

 

Trifolium species vary in their responsiveness to rhizobia 

 

We detected a significant effect of plant species on scaled shoot dry weight (p < 2 e-16; 

Table 4.4). To assess the net effect of rhizobia on scaled shoot biomass, we determined whether 

the 95% CI of each species scaled shoot biomass overlapped with 1. Most species performed 

better when inoculated than without rhizobia: inoculated T. barbigerum, T. fucatum, T. 

gracilentum, T. macraei, T. microdon, and T. willdenovii were on average 2.0 (95% CI 1.7-2.5), 

1.6 (95% CI 1.5-1.8), 1.4 (95% CI 1.5-1.8) 1.5 (95% CI 1.3-1.8), 2.0 (95% CI 1.7-2.4), and 1.3 

(95% CI 1.1-1.5), respectively, times larger than control plants (Figure 4.7a). The remaining two 

species did not perform better with rhizobia than without: inoculated T. bifidum and T. 

microcephalum were on average 1.1 (95% CI 0.9-1.3) and 1.0 (95% CI 0.8-1.3) times the size of 

control plants (Figure 4.7a).  

 

We also assessed response to rhizobia in terms of investment through specific nodulation 

and nodule:root biomass. There was relatively little variation in specific nodulation (Figure 

4.7b). T. fucatum displayed significantly higher nodule:root biomass than any other species 

(Figure 4.7c).  

 

We detected a marginally significant positive effect of commonness on scaled shoot 

biomass (p = 0.078; Fig 4.13). We did not detect a significant effect of commonness on specific 

nodulation (p = 0.338), but there was a general positive trend (Fig 4.13). We did not detect a 

significant effect of commonness on nodule:root biomass (p < 1 after Bonferroni correction) and 

there was no clear relationship between the two (Fig 4.13). It is important to note that with only 

eight points in this regression, the lack of significant effects could be due to lack of power due to 

small sample number, but the overall trends are still informative.  

 

135 

 

 

Figure 4.7 | Response to rhizobia for each plant species averaged across all rhizobia 
treatments. a) Mean scaled shoot dry weight, b) mean specific nodulation (nodule number per 
gram root biomass), and c) mean nodule:root biomass ratio (mg nodule per g root biomass) for 
each plant species across all inoculated treatments  Error bars represent the 95% CI in a) and +/- 
1 SEM in b) and c). Bars with the same letter in a given panel are not significantly different after 
Tukey testing (p > 0.05). 
 

 

 

 

136 

 

Trifolium species vary in return on investment in rhizobia 

 

We also measured raw efficiency (g of shoot biomass minus mean control biomass for 

that species, per mg of nodule biomass) and scaled efficiency (g of shoot biomass divided by 

mean control biomass for that species, per mg of nodule biomass). We detected a significant 

correlations and positive correlation coefficients between nodule biomass and shoot biomass for 

all plant species (Table 4.10, Figure 4.8). The magnitude of the correlation coefficient varied by 

plant species. T. fucatum and T. willdenovii had the highest correlation coefficients (0.759 and 

0.710, respectively), while the 95% confidence intervals of the remaining species all overlapped 

(Figure 4.8). Note that scaled shoot biomass and raw efficiency are interchangeable in a 

correlation for a single species because they have a linear relationship.  

 

We also assessed efficiency through linear regression. We detected significant main 

effects of plant species (p < 2.2e-16) and nodule biomass (p = 1.85 e-11) and a significant plant 

species x nodule biomass interaction (p = 0.01) on shoot biomass minus mean control biomass 

(raw efficiency; Table 4.8). Similarly, we detected significant main effects of plant species (p < 

2.2e-16) and nodule biomass (p < 2.2e-16) and a significant plant species x nodule biomass 

interaction (p < 2.2e-16) on shoot biomass minus mean control biomass (scaled efficiency; Table 

4.8). Raw efficiency ranged from approximately 0.02 g shoot/mg nodule to 0.06 g shoot/mg 

nodule (Figure 4.8a). T. gracilentum had the highest raw efficiency, while T. fucatum had the 

lowest raw efficiency (Figure 4.8a). Scaled efficiency ranged from approximately 0.1x mean 

control biomass to 4x mean control biomass per mg nodule. T. barbigerum and T. microdon had 

the highest relative efficiency, followed by T. microcephalum (Figure 4.8b). T. fucatum had by 

far the lowest relative efficiency (Figure 4.8b). 

 

137 

 

Figure 4.8 | Correlation coefficient and effect size for efficiency across all test species. Mean 
and 95% confidence interval for a) the correlation coefficient for nodule biomass and scaled 
shoot biomass, b) the slope of raw shoot biomass gain (shoot biomass minus mean control shoot 
biomass for each plant species) plotted against nodule biomass, i.e., g shoot gained per mg 
invested in nodule biomass, and c) the slope of scaled shoot biomass plotted against nodule 
biomass, i.e., proportion of control biomass gained per mg invested in nodule biomass. 
 

 

Table 4.8 | ANCOVA summarizing the effects of nodule biomass and test species on shoot 
biomass gain.  Raw biomass gain was calculated as shoot biomass minus the mean shoot 
biomass of uninoculated controls, while scaled biomass gain was calculated as shoot biomass 
divided by the mean shoot biomass of uninoculated controls.  
  Scaled biomass gain 
  
  
  
 
Nodule biomass (NB) F(1,1032)=1.17e03 <2.2e-16 
F(1,1032)=82.0 <2.2e-16 
F(7,1032)=9.49  1.85e-11   F(7,1037)=14.6 <2.2e-16 
Plant species (PS) 
NB*PS 
F(7,1032)=2.53 
  F(3,1037)=31.9 <2.2e-16 
 

Raw biomass gain 
p 

0.014 

F 

F 

p 

 

 

 

138 

 

Discussion 

If it is assumed that interacting with any partner is better for a symbiont than interacting 

with no partners at all, then one would expect that generalist symbionts would have higher 

fitness because they have a wider range of potential partners available to them (Futuyma & 

Moreno, 1988; Wilson & Yoshimura, 1994). However, the jack of all trades, master of none 

hypothesis suggests that specialist symbionts exist because there is heterogeneity in hosts, and 

specialists have higher fitness on their preferred host than a generalist would have on that host 

(Wilson & Yoshimura, 1994; Burdon et al., 1999). This theory assumes that there is a trade-off 

between traits that promote adaptation to various hosts (Futuyma & Moreno, 1988; Kassen, 

2002), which has been demonstrated in the legume-rhizobia symbiosis (Ehinger et al., 2014). In 

this experiment, plants were almost always classified as specialists based on PDI and most 

rhizobia were classified as specialists based on PDI (Figure 4.1). It is difficult to clarify the 

relationship between plant specialization and plant benefit based on our data. We did not detect a 

significant effect of any measure of PDI on plant benefit from rhizobia, though it is difficult to 

say whether this is simply due to the small number of points in the regression (Figure 4.2). Since 

an individual plant can interact with many rhizobial partners at once and are predicted to be able 

to preferentially associate with their most beneficial rhizobia, we expect that the maximum plant 

responses to rhizobia in these single strain inoculations are more representative of the benefit 

plants would get from their more beneficial partners in the field. The effects of all measures of 

PDI on plant scaled shoot biomass are positive, suggesting that plants have higher potential 

fitness in the field as they become more specialized (Figure 4.2). This agrees with Ehinger et al., 

(2014), who found that the specialist native Californian legume Acmispon strigosus obtained 

more benefit from and invested more resources into rhizobia than the generalist native 

139 

 

Californian legume Lupinus bicolor. However, the results from this study should be approached 

cautiously due to the lack of power of these tests.  

Our results for the correlation between specialization and benefit gained by rhizobia 

sharply contrast to the plant results. We detected significant negative effects of rhizobial 

specialization on mean investment in rhizobia, both in terms of nodule number and nodule 

biomass (Figure 4.3). However, we did not detect a significant effect of specialization on 

maximum investment in rhizobia, suggesting that maximum benefit from symbiosis does not 

increase with specialization for the rhizobia as it does for the plant. Unlike plants, a single 

rhizobial cell can only colonize one plant, so the relationship between mean responses and 

rhizobial PDI is probably more representative of what occurs in the field. In this case, there is 

fairly strong negative selection pressure against being a specialist, since mean allocation to 

rhizobia decreases with increasing rhizobial specialization (Figure 4.3). However, a large 

proportion of the rhizobia we tested are still specialists (Figure 4.1), suggesting that there must 

be something counteracting this negative selection. Perhaps existing in a more realistic field 

setting with multiple strains of rhizobia competing for nodule space would change the relative 

benefits of specialization. 

We had hypothesized that if Trifolium species were specialized in their interactions with 

rhizobia, and were able to amplify their most beneficial rhizobia, they would perform better 

when inoculated with conspecific versus heterospecific rhizobia. Our results do not provide 

strong support for this hypothesis because interaction type did not significantly affect scaled 

shoot biomass (Table 4.4). However, in the small subset of plant species in which we could 

assess flower number, we detected a small positive effect of conspecific rhizobia (Table 4.4, 

Figure 4.14). Flower number is a more direct measure of plant fitness than shoot biomass, and it 

140 

 

is theoretically possible for a plant to produce more flowers without a significant increase in 

shoot biomass. Thus, the positive effect of conspecific rhizobia on flower number in three 

species may extend to the remaining species, if they had been given time to flower. However, we 

cannot be sure of this without further experimentation, and thus evidence of more benefit for 

plants from conspecific rhizobia must be taken with caution. A number of wild legumes have 

been shown to perform better with conspecific rhizobia than with heterospecific rhizobia 

(Wilkinson & Parker, 1996; Thrall et al., 2000; Murray et al., 2001; Ehinger et al., 2014), 

making this result particularly surprising. 

There are two possible explanations for the lack of difference between benefits received 

from conspecific and heterospecific rhizobia. The first is that there is no difference in the level of 

benefit provided by the 32 strains tested in this experiment. Indeed, we did not detect a 

significant effect of inoculum (a factor with 32 levels, one for each different rhizobial strain) or a 

significant inoculum x test species interaction on scaled shoot biomass (Table 4.11). We did, 

however, detect a significant main effect of inoculum and a significant inoculum x test species 

interaction on flower number in the reduced flower number dataset (Table 4.11). Thus, we once 

again have only weak evidence for variation in benefit gained from each individual strain that 

may or may not extend to all 8 Trifolium species. While we do not have data for this particular 

field site, field sites with Trifolium have been reported to contain 106 cells of R. leguminosarum 

bv. trifolii per gram (Hirsch et al., 1993). In addition, 72 distinct strains of R. leguminosarum 

were isolated from a 1 m2 plot in York, United Kingdom (Kumar et al., 2015). This high level of 

diversity coupled with high levels of specificity in plant-rhizobia interactions in Trifolium 

(Andrews & Andrews, 2017) make it unlikely that there is simply no variation in benefit.  

141 

 

The second potential explanation is that if there is variation in the benefit gained from 

these rhizobial strains, it is not clearly divided by the conspecific and heterospecific groupings 

that we tested. Since we only sampled 4-5 strains per isolation host, the mean benefit from the 

conspecific strains may be skewed by the inclusion of just a few strains that were not preferred 

symbionts for that host but managed to form the single nodule we sampled. The benefit provided 

by rhizobia also does not appear to be partitioned by isolation host (Figure 4.15 & 4.16), though 

this assessment required a large number of pairwise tests, so the power to detect individual 

differences is fairly low. Finally, it is also possible that there is variation in benefit provided by 

rhizobial strains, and it is distributed throughout these 32 strains without regard to the interaction 

type or isolation host. There may be underlying genetic patterns in these strains that explain 

variation in benefit that we cannot detect without sequencing the rhizobia. This sequencing is in 

progress currently and will provide insight into the population structure of the rhizobia and the 

genetic factors that may shape their effectiveness. Such comparative genomic studies of other 

rhizobia strain collections have revealed novel insights about population structure and the myriad 

ways the rhizobia adapted to their native soil and host plants  (Sugawara et al., 2013; Kumar et 

al., 2015; Porter et al., 2016). 

 

Our results were also unclear regarding the hypothesis that conspecific rhizobia would 

benefit more from their hosts. T. barbigerum showed no differences in nodule number or 

biomass between conspecific and heterospecific rhizobia, while three species made fewer 

nodules with conspecific rhizobia with no change in nodule biomass, two species made more 

nodules with conspecific rhizobia with no change in nodule biomass, and T. fucatum made more 

nodules and had higher nodule biomass with conspecific rhizobia than with heterospecific 

rhizobia (Figure 4.4). This trend suggests that there is significant variation in the way Trifolium 

142 

 

species adjust their allocation to rhizobia, and that there is a disconnect between nodule number 

and nodule biomass. 

Legumes can engage in partner choice, or a pre-infection selection of partners that relies 

on some sort of signal communicating partner quality (Simms & Taylor, 2002; Sachs et al., 

2004). Plants can also engage in sanctions/preferential allocation, which occurs post-infection 

and involves the plant punishing poor quality partners and/or allocating more resources to more 

beneficial partners (Kiers et al., 2002, 2003). Many plant species changed nodule number 

without changing average or total nodule biomass, suggesting that they tend to engage in partner 

choice rather than sanctions/preferential allocations. However, this experiment cannot be 

considered a direct test of partner choice since the plants did not have multiple strains of rhizobia 

to choose from. Further experimentation would be required to confirm this trend. Partner choice 

has been demonstrated in Trifolium polymorhpum and Trifolium purpureum (Yates et al., 2005, 

2008), though this was an extreme case with rhizobia from different continents and a choice 

between an effective fixer and a strain that did not fix any nitrogen(Yates et al., 2005, 2008). The 

closely related model legume Medicago truncatula has been shown to engage in partner choice 

(Gubry-Rangin et al., 2010). These trends suggest that the Trifolium species in this experiment 

may be able to engage in partner choice. However, since autoregulation of nodulation is tied to 

nitrogen status (Mortier et al., 2012), an increase in nodule number may also be explained by the 

plant initiating more nodules in hopes of finding a high quality rhizobium if all of its current 

partners are poor nitrogen fixers. In contrast, T. fucatum makes more and larger nodules with its 

conspecific rhizobia, suggesting that it may engage in both partner choice and 

sanctions/preferential allocation. The congener Trifolium pratense preferentially allocates more 

resources to more cooperative arbuscular mycorrhizal fungi (Argüello et al., 2016), and M. 

143 

 

trucatula has been shown to employ changes in nodule number and average nodule biomass 

simultaneously, though the changes in nodule biomass did not affect viable rhizobia cells per 

nodule (Gubry-Rangin et al., 2010). However, without a clear connection to increased plant 

benefit, it is unclear how effective these forms of partner policing are. The overall trend of 

changes in nodulation patterns without resulting changes in plant performance have been 

detected in Medicago lupulina, and may indicate that the plant is increasings its allocation of 

resources to rhizobia to compensate for less beneficial partners (Harrison et al., 2017). 

Interpreting the effects of these trends on rhizobial fitness is complicated, since nodule 

number and nodule biomass are both fitness components for rhizobia (Ratcliff et al., 2012). The 

Trifolium-Rhizobium symbiosis produces indeterminate nodules in which N-fixing bacteroids are 

terminally differentiated (Poole et al., 2018). Thus, the only rhizobia that are able to reproduce 

once the nodule senesces at the end of the growing season are the undifferentiated cells that 

remain near the nodule meristem and do not fix nitrogen (Oono et al., 2009). Crushing and 

plating experiments would be required to track the ratio of live cells to nodule number and 

nodule biomass to determine direct effects on rhizobial fitness. These ratios may also vary by 

rhizobial strain, making things even more complicated. In addition, nodule number in single 

strain inoculations can be correlated to nodulation competitiveness in multi-strain inoculations 

(Friesen, 2012). Thus, nodule number differences in these single strain inoculations may have 

important implications in more natural settings where multiple strains of rhizobia are competing 

to nodulate plant roots. However, a direct understanding of how our nodulation results translate 

into rhizobial fitness would require a large number of tests that assessed which rhizobia strains 

nodulated which plant species in a natural soil environment, and how many rhizobial offspring 

arose from each nodule. 

144 

 

While we did not detect a significant effect of commonness in the field on plant response 

to rhizobia, the general trend is positive as predicted, and the lack of significant relationship may 

be due to lack of power from only having 8 points (Figure 4.13). A species that is more common 

in the field is more likely to have co-adapted to a given rhizobial strain since they would have 

had more opportunities to interact over evolutionary time, all other factors equal. While specific 

nodulation also had a positive relationship with commonness, nodule:root biomass did not, 

highlighting the disconnect between nodule number and nodule biomass (Figure 4.13). Plant 

growth response to rhizobia is determined both by the amount of investment and the return on 

that investment, or the nodule efficiency. T. barbigerum and T. microdon are the most common 

species in the field that also show the largest growth response to rhizobia (Figure 4.7). These 

species have fairly average biomass investment in rhizobia, but have much higher scaled 

efficiency than any other species, explaining their large response (Figure 4.7). On the other hand, 

T. fucatum invests a huge proportion of biomass in nodules that have a relatively low raw 

efficiency (Figure 4.7). This relatively low raw efficiency translates into an exceptionally low 

scaled efficiency because T. fucatum is on average significantly larger than any other species. T. 

fucatum makes the largest average biomass per nodule by far, suggesting that perhaps there are 

inefficiencies in gas exchange or nutrient exchange in excessively large nodules.   

It is important to consider these results in light of a few caveats about the setup of this 

experiment. First, we only sampled 4-5 rhizobial strains per plant species, which is likely to be a 

very small subset of the strains available in the field. We also artificially forced a single 

Trifolium to interact with a highly concentrated inoculum of a single strain of rhizobia. In the 

field, Trifolium would have to contend with inter- and intraspecific competition, abiotic stresses, 

and a full microbial community including other strains of rhizobia and pathogens. These factors 

145 

 

can affect the dynamics of nutrient trade between the plant and rhizobia. The presence of other 

rhizobial strains is particularly important, since this introduces competition for nodulation, which 

has serious impacts on colonization success, and it gives the plant the ability to exert partner 

choice (Gerding et al., 2014). In addition, the colonization success and/or benefit provided by a 

given strain may depend on other microbes that were not included in this experiment (Martínez-

Hidalgo & Hirsch, 2017). These trends could be explored through inoculating Trifolium with a 

mix of rhizobial strains and tracking which strains nodulate and fix nitrogen best on the different 

plant species in a more complex environment.  

In conclusion, the results of this experiment do not fit with either of the theoretical 

explanations of how mutualisms may promote plant coexistence. If Parker (1999)’s theory of co-

adaptation leading to coexistence through spatial patchiness held true, we would have expected 

to see positive plant-rhizobia feedbacks, with plants benefiting more from and investing more 

into conspecific rhizobia. Alternatively, if Bever (1999)’s theory of coexistence through negative 

feedbacks between rhizobia and plant were true, we would have expected that plants would 

perform worse with conspecific rhizobia and invest more into heterospecific rhizobia. We did not 

see either of these trends, suggesting that some other trend is driving coexistence between 

Trifolium species. Both the plants and rhizobia were usually specialists, suggesting that there is 

specialization, but maximum benefit is not clearly partitioned by interaction type or by isolation 

host. Future work will include sequencing the genomes of the rhizobial strains, which should 

help to clarify the genomic patterns that will hopefully explain the patterns of benefit. These and 

future results will have important implications for understanding the role of rhizobia in natural 

ecosystems and the development of suitably effective and specialized rhizobial crop inoculants. 

 

146 

 

 

 

 

 

 

 

 

 

 

 

 

APPENDIX 

 

 

 

 

 

 

 

 

 

 

 

147 

 

SUPPLEMENTAL INFORMATION TO CHAPTER 4 

 

Figure 4.9 | Scatterplots of mean plant responses to rhizobia versus plant PDI measures. 
Mean scaled shoot biomass per plant species plotted against a) scaled shoot biomass PDI, d) 
specific nodulation PDI, and g) nodule:root biomass PDI. Mean plant specific nodulation per 
plant species plotted against a) scaled shoot biomass PDI, d) specific nodulation PDI, and g) 
nodule root biomass PDI. Mean plant nodule:root biomass plotted against a) scaled shoot 
biomass PDI, d) specific nodulation PDI, and g) nodule root biomass PDI. Dotted lines indicate a 
nonsignificant effect of x on y (p > 0.05 after Bonferroni correction) in a linear model of the 
form log(response) ~ x. 
 
 
 

148 

 

 

Figure 4.10 | Scatterplots of maximum plant responses to rhizobia versus plant PDI 
measures. Maximum scaled shoot biomass per plant species plotted against a) scaled shoot 
biomass PDI, d) specific nodulation PDI, and g) nodule:root biomass PDI. Maximum specific 
nodulation per plant species plotted against a) scaled shoot biomass PDI, d) specific nodulation 
PDI, and g) nodule root biomass PDI. Maximum nodule:root biomass per plant species plotted 
against a) scaled shoot biomass PDI, d) specific nodulation PDI, and g) nodule root biomass PDI. 
Dotted lines indicate a nonsignificant effect of x on y (p > 0.05 after Bonferroni correction) in a 
linear model of the form log(response) ~ PDI. Solid lines indicate a significant effect of x on y (p 
< 0.05 after Bonferroni correction). Maximum values of scaled shoot biomass, specific 
nodulation, and nodule:root ratio were calculated by finding the mean of each response for each 
combination of inoculum and plant species, and then selecting the maximum mean for each plant 
species. 
 

149 

 

 

 

Figure 4.11 | Scatterplots of mean plant responses to rhizobia versus rhizobial PDI 
measures. Mean scaled shoot biomass per rhizobial strain plotted against a) scaled shoot 
biomass PDI, d) specific nodulation PDI, and g) nodule:root biomass PDI. Mean plant specific 
nodulation per plant species plotted against a) scaled shoot biomass PDI, d) specific nodulation 
PDI, and g) nodule root biomass PDI. Mean plant nodule:root biomass plotted against a) scaled 
shoot biomass PDI, d) specific nodulation PDI, and g) nodule root biomass PDI. Dotted lines 
indicate a nonsignificant effect of x on y (p > 0.05 after Bonferroni correction) in a linear model 
of the form log(response) ~ PDI. Solid lines indicate a significant effect of x on y (p < 0.05 after 
Bonferroni correction). 
 
 
 

150 

 

 

Figure 4.12 | Scatterplots of maximum plant responses to rhizobia versus rhizobial PDI 
measures. Maximum scaled shoot biomass per rhizobial strain plotted against a) scaled shoot 
biomass PDI, d) specific nodulation PDI, and g) nodule:root biomass PDI. Maximum specific 
nodulation per rhizobial strain plotted against a) scaled shoot biomass PDI, d) specific nodulation 
PDI, and g) nodule root biomass PDI. Maximum nodule:root biomass per rhizobial strain plotted 
against a) scaled shoot biomass PDI, d) specific nodulation PDI, and g) nodule root biomass PDI. 
Dotted lines indicate a nonsignificant effect of x on y (p > 0.05) in a linear model. Maximum 
values of scaled shoot biomass, specific nodulation, and nodule:root ratio were calculated by 
finding the mean of each response for each combination of inoculum and plant species, and then 
selecting the maximum mean for each plant species. 
 
 
 

151 

 

Figure 4.13 | Response to rhizobia versus commonness in the field. a) Scaled shoot biomass, 
b) specific nodulation, and c) nodule:root biomass plotted against the fraction of plots in the field 
in which a species is present. We detected a marginally significant effect of commonness on 
scaled shoot biomass (F(1,6)=8.57, p = 0.078 after Bonferroni correction). We did not detect a 
significant effect of commonness on specific nodulation F(1,6)=3.45, p = 0.338 after Bonferroni 
correction). We did not detect a significant effect of commonness on specific nodulation 
(F(1,6)=0.213, p < 1 after Bonferroni correction). 
 
 

 

152 

 

 

Figure 4.14 | Mean scaled shoot biomass and flower number for each test species inoculated 
with conspecific or heterospecific rhizobia.  a) Mean scaled shoot biomass for each test species 
inoculated with conspecific or heterospecific rhizobia. We did not detect a significant effect of 
interaction type or an interaction between interaction type and test species on scaled shoot 
biomass (p = 0.890 and p = 0.731, respectively). b) Mean flower number for the reduced set of 
test species analyzed for flower number. Plants inoculated with conspecific rhizobia made 
significantly more flowers than plants inoculated with heterospecific rhizobia overall (p = 0.010), 
but we did not detect an interaction between flower number and plant species (p = 0.523). 
 

153 

 

 
Figure 4.15 | Mean scaled shoot biomass for each plant species inoculated with rhizobia 
from each isolation host. We detected a significant effect of test host species (F7,998=17.3, p < 
2e-16), but we did not detect a significant effect of isolation host (F6,998=1.54, p = 0.161) or a 
significant interaction between isolation host and test host species (F42,998=1.33, p = 0.078). 
 

154 

 

(2,390)=97.2, p < 2e-

(6,390)=10.4, p = 0.108) or a 

(12,390)=18.7, p = 0.097). 

 
Figure 4.16 | Mean flower number for each plant species inoculated with rhizobia from 
each isolation host. We detected a significant effect of test host species (χ2
16), but we did not detect a significant effect of isolation host (χ2
significant interaction between isolation host and test host species (χ2
 
 
 
 
 
 
 
 
 
Table 4.9 | Percentage of inoculated individuals of each plant species that flowered.  
Plant species 
T.barbigerum 
T. bifidum 
T. fucatum 
T. gracilentum 
T. macraei 
T. microdon 
T. microcephalum 
T. willdenovii 
 
 
 

2.54 
19.1 
40.7 
23.1 
48.9 
50.8 
2.59 
1.44 

Flowering percentage 

155 

 

t 

p 

Table 4.10 | Correlation between nodule biomass and scaled shoot biomass for each test 
species. Pearson’s correlation coefficient, the upper and lower bound of the 95% confidence 
interval for the coefficient, and t and Bonferroni-corrected p values for the correlation test 
between nodule biomass and scaled shoot biomass for each plant species. 
 Correlation coefficient 2.5% 97.5% 
  
T. barbigerum 
 
T. bifidum 
 
T. fucatum 
 
T. gracilentum 
 
T. macraei 
 
T. microdon 
 
T. microcephalum  
T. willdenovii 
 
 
 
 
 
Table 4.11 | ANOVA summarizing the effects of inoculum and test species on plant 
performance.  Raw biomass gain was calculated as shoot biomass minus the mean shoot 
biomass of uninoculated controls, while scaled biomass gain was calculated as shoot biomass 
divided by the mean shoot biomass of uninoculated controls.  
 

0.241 0.548 4.72  6.03e-05 
0.431 0.664 7.79  1.44e-11 
0.678 0.821 13.6 < 1.76e-15 
0.270 0.551 5.32  3.81e-06 
0.093  0.40  3.10  0.021 
0.337 0.604 6.22  5.25e-08 
0.273 0.571 5.13  9.43e-06 
0.616 0.784 11.8 < 1.76e-15 

0.406 
0.559 
0.759 
0.420 
0.256 
0.482 
0.433 
0.710 

  

Flower number 

χ2 

(31,315) =63.0 
(2,315)= 95.2 
(62,315) =83.3 

p 

5.87e-04 
<2e-16 
0.037 

p 

F 

Scaled shoot biomass    
    
  
F(31,798) = 0.921  0.592      χ2
Inoculum (I) 
Test species (TS)  F(7,798) = 17.5  < 2e-16     χ2
F(217,798) = 1.11  0.152      χ2
I*TS 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

156 

 

 

 

 

 

 

 

 

 

 

 

 

REFERENCES 

 

 

 

 

 

 

 

 

 

 

 

157 

 

 

REFERENCES 

Adler PB, Ellner SP, Levine JM. 2010. Coexistence of perennial plants: an embarrassment of  
niches. Ecology Letters 13: 1019–1029. 

Afkhami ME, McIntyre PJ, Strauss SY. 2014. Mutualist-mediated effects on species’ range  
limits across large geographic scales. Ecology Letters 17: 1265–1273. 

Andrews M, Andrews ME. 2017. Specificity in Legume-Rhizobia Symbioses. International  
Journal of Molecular Sciences 18: 705. 

Argüello A, O’Brien MJ, van der Heijden MGA, Wiemken A, Schmid B, Niklaus PA. 2016. 
Options of partners improve carbon for phosphorus trade in the arbuscular mycorrhizal 
mutualism (H Maherali, Ed.). Ecology Letters 19: 648–656. 

Bever JD. 1999. Dynamics within mutualism and the maintenance of diversity: inference from a  
model of interguild frequency dependence. Ecology Letters 2: 52–61. 

Bever JD. 2003. Soil community feedback and the coexistence of competitors: conceptual  
frameworks and empirical tests. New Phytologist 157: 465–473. 

Burdon JJ, Gibson AH, Searle SD, Woods MJ, Brockwell J. 1999. Variation in the 
effectiveness of symbiotic associations between native rhizobia and temperate Australian Acacia:  
within-species interactions. Journal of Applied Ecology 36: 398–408. 

Cadotte MW, Cavender-Bares J, Tilman D, Oakley TH. 2009. Using Phylogenetic, 
Functional and Trait Diversity to Understand Patterns of Plant Community Productivity (RP  
Freckleton, Ed.). PLoS ONE 4: e5695. 

Cardinale BJ, Srivastava DS, Duffy JE, Wright JP, Downing AL, Sankaran M, Jouseau C, 
Cadotte MW, Carroll IT, Weis JJ, et al. 2009. Effects of biodiversity on the functioning of 
ecosystems: a summary of 164 experimental manipulations of species richness (WK Michener,  
Ed.). Ecology 90: 854–854. 

Cardinale BJ, Srivastava DS, Emmett Duffy J, Wright JP, Downing AL, Sankaran M, 
Jouseau C. 2006. Effects of biodiversity on the functioning of trophic groups and ecosystems. 
Nature 443: 989–992. 

Carlsson G, Palmborg C, Huss-Danell K. 2006. Discrimination against 15 N in three N 2 -
fixing Trifolium species as influenced by Rhizobium strain and plant age. Acta Agriculturae  
Scandinavica, Section B - Plant Soil Science 56: 31–38. 

Chesson P. 2000. Mechanisms of Maintenance of Species Diversity. Annual Review of Ecology  
and Systematics 31: 343–366. 

158 

 

Ehinger M, Mohr TJ, Starcevich JB, Sachs JL, Porter SS, Simms EL. 2014. Specialization-
generalization trade-off in a Bradyrhizobium symbiosis with wild legume hosts. BMC Ecology  
14: 8. 

Fridley JD, Grime JP, Bilton M. 2007. Genetic identity of interspecific neighbours mediates 
plant responses to competition and environmental variation in a species-rich grassland. Journal  
of Ecology 95: 908–915. 

Friesen ML. 2012. Widespread fitness alignment in the legume-rhizobium symbiosis. The New  
phytologist 194: 1096–111. 

Futuyma DJ, Moreno G. 1988. The evolution of ecological specialization. Ann. Rev. Ecol. Syst  
19: 207–240. 

Gerding M, O’Hara GW, Howieson JG, Bräu L. 2014. Overcoming non-selective nodulation 
of Lessertia by soil-borne rhizobium in the presence of inoculant mesorhizobium. Plant and Soil 
380: 117–132. 

Gubry-Rangin C, Garcia M, Béna G. 2010. Partner choice in Medicago truncatula- 
Sinorhizobium symbiosis. Proceedings. Biological sciences / The Royal Society 277: 1947–51. 

Harrison TL, Wood CW, Borges IL, Stinchcombe JR. 2017. No evidence for adaptation to  
local rhizobial mutualists in the legume Medicago lupulina. Ecology and Evolution 00: 1–10. 

Hirsch PR, Jones MJ, McGrath SP, Giller KE. 1993. Heavy metals from past applications of 
sewage sludge decrease the genetic diversity of rhizobium leguminosarum biovar trifolii  
populations. Soil Biology and Biochemistry 25: 1485–1490. 

Hodge A, Fitter AH. 2013. Microbial mediation of plant competition and community structure  
(D Robinson, Ed.). Functional Ecology 27: 865–875. 

Holland AA. 1970. Competition between soil- and seed-borneRhizobium trifolii in nodulation of  
introducedTrifolium subterraneum. Plant and Soil 32: 293–302. 

Holland AA, Parker CA. 1966. Studies on microbial antagonism in the establishment of clover  
pasture. Plant and Soil 25: 329–340. 

Howieson JG, Yates RJ, O’Hara GW, Ryder M, Real D. 2005. The interactions of Rhizobium 
leguminosarum biovar trifolii in nodulation of annual and perennial Trifolium spp. from diverse  
centres of origin. Australian Journal of Experimental Agriculture 45: 199. 

Hughes AR, Inouye BD, Johnson MTJ, Underwood N, Vellend M. 2008. Ecological  
consequences of genetic diversity. Ecology Letters 11: 609–623. 

Hughes AR, Stachowicz JJ. 2004. Genetic diversity enhances the resistance of a seagrass 
ecosystem to disturbance. Proceedings of the National Academy of Sciences of the United States  
of America 101: 8998–9002. 

159 

 

Jordan D. 1984. Family 111. Rhizobiaceae. In: Krieg N, JG H, eds. Bergey’s Manual of  
Systematic Bacteriology. Baltimore, MD: The Williams & Wilkins Co., 234–244. 

Kassen R. 2002. The experimental evolution of specialists, generalists, and the maintenance of 
diversity. Journal of Evolutionary Biology 15: 173–190. 

Kazenel MR, Debban CL, Ranelli L, Hendricks WQ, Chung YA, Pendergast TH, Charlton 
ND, Young CA, Rudgers JA. 2015. A mutualistic endophyte alters the niche dimensions of its 
host plant. AoB PLANTS 7: plv005-plv005. 

Keller KR. 2014. Mutualistic rhizobia reduce plant diversity and alter community composition. 
Oecologia 176: 1101–1109. 

Kiers ET, Rousseau RA, West SA, Denison RF. 2003. Host sanctions and the legume– 
rhizobium mutualism. Nature 425: 78–81. 

Kiers ET, West SA, Denison RF. 2002. Mediating mutualisms: farm management practices and  
evolutionary changes in symbiont co-operation. Journal of Applied Ecology 39: 745–754. 

Klironomos JN. 2002. Feedback with soil biota contributes to plant rarity and invasiveness in 
communities. Nature 417: 67–70. 

Kumar N, Lad G, Giuntini E, Kaye ME, Udomwong P, Shamsani NJ, Young JPW, Bailly 
X. 2015. Bacterial genospecies that are not ecologically coherent: population genomics of 
Rhizobium leguminosarum. Open Biology 5: 140133–140133. 

Levene H. 1953. Genetic Equilibrium When More Than One Ecological Niche is Available. The 
American Naturalist 87: 331–333. 

Levine JM, HilleRisLambers J. 2009. The importance of niches for the maintenance of species 
diversity. Nature 461: 254–257. 

Lira MA, Nascimento LRS, Fracetto GGM. 2015. Legume-rhizobia signal exchange: 
promiscuity and environmental effects. Frontiers in Microbiology 6: 945. 

Mangan SA, Herre EA, Bever JD. 2010. Specificity between Neotropical tree seedlings and  
their fungal mutualists leads to plant–soil feedback. Ecology 91: 2594–2603. 

Martínez-Hidalgo P, Hirsch AM. 2017. The Nodule Microbiome: N 2 -Fixing Rhizobia Do  
Not Live Alone. Phytobiomes: PBIOMES-12-16-0. 

Mortier V, Holsters M, Goormachtig S. 2012. Never too many? How legumes control nodule 
numbers. Plant, Cell & Environment 35: 245–258. 

Murray BBR, Thrall PH, Woods MJ. 2001. Acacia species and rhizobial interactions: 
Implications for restoration of native vegetation. Ecological Management and Restoration 2:  

160 

 

213–219. 

Oono R, Denison RF, Kiers ET. 2009. Controlling the reproductive fate of rhizobia: how 
universal are legume sanctions? The New phytologist 183: 967–79. 

Parker MA. 1999. Mutualism in Metapopulations of Legumes and Rhizobia. The American 
naturalist 153: S48–S60. 

Peay KG. 2016. The Mutualistic Niche: Mycorrhizal Symbiosis and Community Dynamics. 
Annual Review of Ecology, Evolution, and Systematics 47: 143–164. 

Pendergast TH, Burke DJ, Carson WP. 2013. Belowground biotic complexity drives 
aboveground dynamics: a test of the soil community feedback model. New Phytologist 197: 
1300–1310. 

Poisot T, Canard E, Mouquet N, Hochberg ME. 2012. A comparative study of ecological 
specialization estimators. Methods in Ecology and Evolution 3: 537–544. 

Poole P, Ramachandran V, Terpolilli J. 2018. Rhizobia: from saprophytes to endosymbionts. 
Nature Reviews Microbiology 16: 291–303. 

Porter SS, Chang PL, Conow CA, Dunham JP, Friesen ML. 2016. Association mapping 
reveals novel serpentine adaptation gene clusters in a population of symbiotic Mesorhizobium.  
The ISME Journal. 

van der Putten WH, Bardgett RD, Bever JD, Bezemer TM, Casper BB, Fukami T, Kardol 
P, Klironomos JN, Kulmatiski A, Schweitzer JA, et al. 2013. Plant-soil feedbacks: the past,  
the present and future challenges (M Hutchings, Ed.). Journal of Ecology 101: 265–276. 

Ratcliff WC, Underbakke K, Denison RF. 2012. Measuring the fitness of symbiotic rhizobia.  
Symbiosis 55: 85–90. 

Sachs JL, Mueller UG, Wilcox TP, Bull JJ. 2004. The Evolution of Cooperation. The 
Quarterly Review of Biology 79: 135–160. 

Simms EL, Taylor DL. 2002. Partner choice in nitrogen-fixation mutualisms of legumes and 
rhizobia. Integrative and comparative biology 42: 369–80. 

Sugawara M, Epstein B, Badgley BD, Unno T, Xu L, Reese J, Gyaneshwar P, Denny R, 
Mudge J, Bharti AK, et al. 2013. Comparative genomics of the core and accessory genomes of 
48 Sinorhizobium strains comprising five genospecies. Genome Biology 14: R17. 

Thrall PH, Burdon JJ, Woods MJ. 2000. Variation in the effectiveness of symbiotic 
associations between native rhizobia and temperate Australian legumes: interactions within and 
between genera. Journal of Applied Ecology 37: 52–65. 

161 

 

Turnbull LA, Levine JM, Loreau M, Hector A. 2013. Coexistence, niches and biodiversity 
effects on ecosystem functioning (JHR Lambers, Ed.). Ecology Letters 16: 116–127. 

Wagg C, Jansa J, Stadler M, Schmid B, van der Heijden MGA. 2011. Mycorrhizal fungal 
identity and diversity relaxes plant–plant competition. Ecology 92: 1303–1313. 

Wilkinson HH, Parker MA. 1996. Symbiotic specialization and the potential for genotypic 
coexistence in a plant-bacterial mutualism. Oecologia 108: 361–367. 

Wilson DS, Yoshimura J. 1994. On the Coexistence of Specialists and Generalists. The 
American Naturalist 144: 692–707. 

Yates RJ, Howieson JG, Real D, Reeve WG, Vivas-Marfisi A, O’Hara GW. 2005. Evidence 
of selection for effective nodulation in the Trifolium spp. saymbiosis with Rhizobium 
leguminosarum biovar trifolii. Australian Journal of Experimental Agriculture 45: 189. 

Yates RJ, Howieson JG, Reeve WG, Brau L, Speijers J, Nandasena K, Real D, Sezmis E, 
O’Hara GW. 2008. Host–strain mediated selection for an effective nitrogen-fixing symbiosis 
between Trifolium spp. and Rhizobium leguminosarum biovar trifolii. Soil Biology and 
Biochemistry 40: 822–833. 
  
 
 

 

 

 

 

 

 

 
 
 
 
 
 
 
 

 

162 

 

CHAPTER 5 

 

CONCLUSIONS AND FUTURE DIRECTIONS 

 

 

The interaction between legumes and rhizobia is a vital interaction that has shaped 

natural and agricultural ecosystems and has huge implications for ecosystem function and 

agricultural sustainability (Peoples et al., 2009; Denison & Kiers, 2011). This interaction can be 

considered in the framework of a biological market (Noë & Hammerstein, 1995; Schwartz & 

Hoeksema, 1998). In this simplified market of carbon and nitrogen, the plant must decide how 

much of its resources to allocate to direct nitrogen uptake and how much to allocate to trade. 

This balance of resource allocation can be driven by the external abiotic environment, as 

discussed in Chapters 2 and 3. In Chapter 2, I collaborated with an interdisciplinary group of 

ecologists and metabolic modelers to explore how nitrogen affects a model of resource trade 

based on the Nash bargaining solution, a concept adopted from economics (Nash & Jr., 1950; 

Nash, 1953; Grman et al., 2012). Our data did not support the assumption of symmetric 

bargaining power of the Nash bargaining solution. Instead, plants have more bargaining power 

than rhizobia, and, unexpectedly, this differential is more pronounced at low soil nitrogen. 

Importantly, while plants alter the amount of resources they invest in rhizobia, the “price” of 

carbon (the C:N trade ratio) does not change.  

 

This result of changing allocation but not return on investment was carried over into 

Chapter 3. In this chapter, I altered market conditions in two ways by changing both light and 

nitrogen at the same time. These manipulations should change both the amount of carbon 

available to the plant to allocate as well as the availability and relative cost of nitrogen uptake 

163 

 

from the soil (Schwartz & Hoeksema, 1998; Kiers & van der Heijden, 2006; Sachs & Simms, 

2006). We detected interactions between the effects of light and nitrogen on investment in 

rhizobia: light generally only altered total investment in rhizobia at low nitrogen when rhizobia 

were beneficial to the plant. At high nitrogen, plant allocation was very low and tended not to 

change with light. This chapter highlighted the complicated set of inputs that a plant in the field 

must integrate to determine optimal allocation to rhizobia and suggested that the effects of one 

input will depend on the levels of other nutrients. On a technical note, this chapter also illustrated 

the importance of carefully considering the method in which nodulation is measured. Increasing 

light had opposite effects on unscaled nodule number and specific nodulation, i.e., nodule 

number per unit of root biomass. Many studies that report the effects of external nutrients on 

nodulation report only unscaled nodulation measures. Unscaled nodulation is important since it 

provides information about the fitness benefits a rhizobium receives from its host (Ratcliff et al., 

2012), but may lead to misleading conclusions about plant allocation patterns.  

 

There are a number of future directions that would synthesize the results and approaches 

of Chapters 2 and 3. Our results in Chapter 3 suggest that nitrogen and light have interactive 

effects on certain aspects of the legume-rhizobia symbiosis. Parameterizing the model from 

Chapter 2 with factorial manipulation of nitrogen and light would allow us to explore these 

interactions in more detail. Increasing light has the opposite effect on investment in rhizobia that 

increasing soil nitrogen does. Thus, it would be interesting to explore whether light has a similar 

magnitude and direction of effect on bargaining power as nitrogen does.  

 

It would also be interesting to extend Chapters 2 and 3 to other legume-rhizobia pairs. 

Research suggests that the response of the legume-rhizobia interaction to nitrogen depends on 

the specific pairing of the legume and plant even within the same plant species (Heath et al., 

164 

 

2010). Thus, it is vital to test our results in a variety of systems to see how universal they are 

within the legume-rhizobia symbiosis, and perhaps even in other plant-microbe mutualisms. One 

option for extension is to compare model results when M. truncatula is inoculated with a set of 

strains that range from uncooperative (fixing very little nitrogen) to highly cooperative (fixing 

large amounts of nitrogen). If we consider a less effective strain that induces the formation of the 

same amount of nodule tissue but provides the plant with less growth benefit, that strain 

presumably has negotiated a higher C:N exchange ratio (i.e., they receive more carbon for every 

unit of nitrogen). This higher price for nitrogen suggests that the plant may have lower 

bargaining power in this situation.  

 

There are many possible extensions to this question in other legume-rhizobia systems. 

Some rhizobia accumulate poly-β-hydroxybutyrate (PHB) (Trainer & Charles, 2006). PHB can 

be viewed as a way to “hoard” resources from the plant to fuel later rhizobial reproduction 

instead of being funneled into nitrogen fixation for plant benefit (Ratcliff et al., 2008). This 

resource hoarding would likely affect the price a plant must pay for each unit of nitrogen, while 

also setting up a fitness conflict between the plant and rhizobium, which may alter the balance of 

bargaining power. It would also be interesting to explore C:N trade dynamics in legumes that 

form determinate nodules. M. truncatula forms indeterminate nodules, with persistent meristems 

and terminally differentiated swollen bacteroids (Masson-Boivin et al., 2009). Other legumes 

form determinate nodules, with transient meristems and non-terminally differentiated bacteroids 

that can reproduce after nodule senescence (Masson-Boivin & Sachs, 2018). Indeterminate 

nodules may be more efficient than determinate nodules, but this is difficult to test since no plant 

forms both types of nodule (Oono & Denison, 2010). These differences have many important 

implications for C:N trade dynamics. The reproductive ability of nitrogen fixing bacteroids may 

165 

 

translate into a more direct correlation between plant and rhizobial fitness, while higher 

efficiency may reduce the C:N ratio.  

 

There are a number of important technical considerations that would be required to carry 

out these extensions. First, it would be valuable to expand the range of nitrogen and light 

conditions tested so that they range from the minimum level that supports uninoculated plant 

survival through the experiment to a growth-saturating maximum. These minimum and 

maximum levels would vary for each plant species tested. This setup could also be used to test 

hypotheses about resource availability, trade dynamics, and plant control of their rhizobial 

partners. For example, you could test whether plants that can generally enforce more favorable 

C:N trade ratios and/or tend to have higher bargaining power with their rhizobia, and whether 

nutrient availability affects the strength of partner choice or sanctions, such as in (Grillo et al., 

2016).   

 

Returning to the market framework for legume-rhizobia resource trade, in Chapter 4 I 

manipulated the partners available for trade on the market. Different partners come with different 

returns on investment and thus different optimal investment in those partners (Noë & 

Hammerstein, 1994; Werner et al., 2014). This chapter also produced results that challenged our 

assumptions based in ecological theory. While both the plants and rhizobia were specialists, 

plants did not receive more benefit from or invest more resources into conspecific rhizobia than 

heterospecific rhizobia. Benefit from and investment in rhizobia was also not clearly partitioned 

by isolation host. This suggests that there is some other underlying population structure that we 

were not able to determine with our knowledge of the isolation host for each rhizobial strain. 

Sequencing the genomes of the rhizobia involved in this experiment, as well as the hundreds of 

other strains we have isolated from our field site, will go a long way to address this question. 

166 

 

These results will provide information about the genetic diversity of the rhizobial population at 

Bodega Bay and will help us assess whether there are patterns in the genome that explain the 

pattern of benefit and investment we detected in this experiment. We can also examine how 

rhizobia have co-adapted to their plant hosts by checking for genes in which the phylogeny of the 

rhizobia track the phylogeny of the plants. This analysis will reveal genes that may be important 

for the specificity of the interaction.  

 

An important follow-up involves increasing the complexity of the biotic environment in 

which the interactions occur. The next logical step would be performing pairwise inoculations 

with a subset of the plant species to test the correlation between nodulation and benefit in single-

strain inoculations and in multi-strain inoculations. This would require marking the strains in 

some way, such as gusA and cellB, which allows easy visualization of nodule occupancy after 

staining the root system (Sessitsch et al., 1996). Assessment of nodulation success and rhizobial 

fitness in more complex inoculations would require DNA barcoding and sequencing to track 

population sizes. Alternatively, it would be both fascinating and challenging to track rhizobial 

populations in the soil at the field site and correlate their population levels in space with that of 

the plants.  

 

 

A plant participating in the biological carbon:nitrogen market must integrate a large 

number of signals about resource availability and partner quality to make continuous adjustments 

to its nitrogen acquisition strategy. This dissertation has focused on exploring these questions at 

an organismal level, but I am particularly interested in understanding the molecular mechanisms 

that underlie these trends.  This is of course a tremendous task, since the pathways that integrate 

such a large number of signals to regulate many plant processes are likely to be complex and 

intertwined. Research on autoregulation of nodulation has been ongoing for decades, but there 

167 

 

are still several pieces of the pathway that we still do not understand (Nishida & Suzaki, 2018). 

There are some experiments that may provide initial insight into these pathways. Two 

complementary approaches would include transcriptomics and mutant analysis. Transcriptomics 

could be used to compare in broad strokes which pathways plants use to downregulate allocation 

to nodules in response to increasing soil nitrogen, decreasing light, or poor rhizobial cooperation. 

The specifics of the experiment would be challenging: one would need to compare across many 

tissues, such as nodule to uninfected root, and effective nodule to ineffective nodule. However, 

since the shoot is involved in autoregulation of nodulation, it would be informative to assess 

shoot transcriptomic changes as well. Alternatively, one could test nodulation responses to one 

input in a mutant deficient in response to a different input, such as testing nodulation responses 

to light in a mutant deficient in autoregulation of nodulation (Bacanamwo & Harper, 1997). This 

would provide preliminary evidence about the involvement of known pathways in novel 

responses.  

 

It is important to note the limitations of the work described in this dissertation. It would 

be ideal to repeat these experiments to determine the variability in the trends observed. In 

addition, this work employed a combination of growth chamber and greenhouse studies. Growth 

chambers and greenhouses are reductionist by nature, since they eliminate stresses from biotic 

and abiotic sources that plants would face in the field. Including relevant stresses from the field, 

such as nutrient limitation, pathogen stress, or plant-plant competition could help increase the 

relevance of these results to field conditions. However, increasing the complexity of the 

experimental setup will make interpretation of results more difficult.  

 

In conclusion, this dissertation assessed the effects of the abiotic environment and 

genotype by genotype interactions on resource allocation and trade in the legume-rhizobia 

168 

 

interactions. These projects have helped me understand the value of combining ecological theory 

and modeling with rigorous empirical testing (Clark, et al., 2017). It has also taught me the value 

of questioning assumptions and carefully considering the meaning of the data reported. These 

results provide an excellent first step for continuing to explore the regulation and ecology of the 

legume-rhizobia symbiosis.  

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

169 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

REFERENCES 

 

 

 

 

 

 

 

 

 

170 

 

REFERENCES 

 

Bacanamwo M, Harper JE. 1997. Response of a hypernodulating soybean mutant to increased  
photosynthate supply. Plant Science 124: 119–129. 

Denison RF, Kiers ET. 2011. Life histories of symbiotic rhizobia and mycorrhizal fungi.  
Current biology : CB 21: R775-85. 

Clark TJ, Friel CA, Grman E, Shachar-Hill Y, Friesen ML. 2017. Modelling nutritional 
mutualisms: challenges and opportunities for data integration (M van Baalen, Ed.). Ecology  
Letters 20: 1203–1215. 

Grillo MA, Stinchcombe JR, Heath KD. 2016. Nitrogen addition does not influence pre-
infection partner choice in the legume-rhizobium symbiosis. American journal of botany 103:  
1763–1770. 

Grman E, Robinson TMP, Klausmeier CA. 2012. Ecological specialization and trade affect 
the outcome of negotiations in mutualism. The American naturalist 179: 567–81. 

Heath KD, Stock AJ, Stinchcombe JR. 2010. Mutualism variation in the nodulation response  
to nitrate. Journal of Evolutionary Biology 23: 2494–2500. 

Kiers ET, van der Heijden MGA. 2006. Mutualistic stability in the the arbuscular mycorrhizal 
symbiosis: exploring hypotheses of evolutionary cooperation. Ecology 87: 1627–1636. 

Masson-Boivin C, Giraud E, Perret X, Batut J. 2009. Establishing nitrogen-fixing symbiosis 
with legumes: how many rhizobium recipes? Trends in Microbiology 17: 458–466. 

Masson-Boivin C, Sachs JL. 2018. Symbiotic nitrogen fixation by rhizobia — the roots of a 
success story. Current Opinion in Plant Biology 44. 

Nash J. 1953. Two-Person Cooperative Games. Econometrica 21: 128. 

Nash JF, Jr. 1950. The Bargaining Problem. Econometrica 18: 155. 

Nishida H, Suzaki T. 2018. Nitrate-mediated control of root nodule symbiosis. Current Opinion  
in Plant Biology 44: 129–136. 

Noë R, Hammerstein P. 1994. Biological markets: supply and demand determine the effect of 
partner choice in cooperation, mutualism and mating. Behavioral Ecology and Sociobiology 35: 
1–11. 

Noë R, Hammerstein P. 1995. Biological markets. Trends in Ecology & Evolution 10: 336–339. 

171 

 

Oono R, Denison RF. 2010. Comparing symbiotic efficiency between swollen versus  
nonswollen rhizobial bacteroids. Plant physiology 154: 1541–8. 

Peoples MB, Brockwell J, Herridge DF, Rochester IJ, Alves BJR, Urquiaga S, Boddey RM, 
Dakora FD, Bhattarai S, Maskey SL, et al. 2009. The contributions of nitrogen-fixing crop 
legumes to the productivity of agricultural systems. Symbiosis 48: 1–17. 

Ratcliff WC, Kadam S V., Denison RF. 2008. Poly-3-hydroxybutyrate (PHB) supports survival 
and reproduction in starving rhizobia. FEMS Microbiology Ecology 65: 391–399. 

Ratcliff WC, Underbakke K, Denison RF. 2012. Measuring the fitness of symbiotic rhizobia. 
Symbiosis 55: 85–90. 

Sachs JL, Simms EL. 2006. Pathways to mutualism breakdown. Trends in Ecology & Evolution 
21: 585–592. 

Schwartz MW, Hoeksema JD. 1998. Specialization and resource trade: Biological markets as a 
model of mutualisms. Ecology 79: 1029–1038. 

Sessitsch A, Wilson K, Akkermans A, de Vos W. 1996. Simultaneous detection of different 
Rhizobium strains marked with either the Escherichia coli gusA gene or the Pyrococcus furiosus  
celB gene. Appl. Envir. Microbiol. 62: 4191–4194. 

Trainer MA, Charles TC. 2006. The role of PHB metabolism in the symbiosis of rhizobia with 
legumes. Applied Microbiology and Biotechnology 71: 377–386. 

Werner GDA, Strassmann JE, Ivens ABF, Engelmoer DJP, Verbruggen E, Queller DC, 
Noë R, Johnson NC, Hammerstein P, Kiers ET. 2014. Evolution of microbial markets. 
Proceedings of the National Academy of Sciences of the United States of America 111: 1237–44. 
 

 
 
 
 
 

 
 

 

172