The direct effects of plant polyploidy on the legume–rhizobia mutualism

Abstract

Background

Polyploidy is known to significantly alter plant genomes, phenotypes and interactions with the abiotic environment, yet the impacts of polyploidy on plant–biotic interactions are less well known. A particularly important plant–biotic interaction is the legume–rhizobia mutualism, in which rhizobia fix atmospheric nitrogen in exchange for carbon provided by legume hosts. This mutualism regulates nutrient cycles in natural ecosystems and provides nitrogen to agricultural environments. Despite the ecological, evolutionary and agricultural importance of plant polyploidy and the legume–rhizobia mutualism, it is not yet fully understood whether plant polyploidy directly alters mutualism traits or the consequences on plant growth.

Scope

The aim was to propose a conceptual framework to understand how polyploidy might directly enhance the quantity and quality of rhizobial symbionts hosted by legume plants, resulting in increased host access to fixed nitrogen (N). Mechanistic hypotheses have been devised to examine how polyploidy can directly alter traits that impact the quantity (e.g. nodule number, nodule size, terminal bacteroid differentiation) and quality of symbionts (e.g. nodule environment, partner choice, host sanctions). To evaluate these hypotheses, an exhaustive review of studies testing the effects of plant polyploidy on the mutualism was conducted. In doing so, overall trends were synthesized, highlighting the limited understanding of the mechanisms that underlie variation in results achieved thus far, revealing striking gaps in knowledge and uncovering areas ripe for future research.

Conclusions

Plant polyploidy can immediately alter nodule size, N fixation rate and the identity of rhizobial symbionts hosted by polyploid legumes, but many of the mechanistic hypotheses proposed here, such as bacteroid number and enhancements of the nodule environment, remain unexplored. Although current evidence supports a role of plant polyploidy in enhancing key aspects of the legume–rhizobia mutualism, the underlying mechanisms and effects on host benefit from the mutualism remain unresolved.

INTRODUCTION

Polyploidy (the condition of having more than two complete sets of chromosomes) is a major driver of evolutionary novelty and speciation in flowering plants (Levin, 2002; Soltis et al., 2014; Soltis and Soltis, 2016; Zhan et al., 2016). Although significant advances have been made in understanding how plant polyploidy affects genotypes, phenotypes and interactions with the abiotic environment (Balao et al., 2011; Husband et al., 2013; Soltis et al., 2014; Alix et al., 2017), much less is known about how it influences biotic interactions (Thompson et al., 2004; Segraves and Anneberg, 2016; Segraves, 2017). Although recent work has found that plant polyploidy can significantly alter plant–pollinator and plant–herbivore interactions (Segraves and Thompson, 1999; Nuismer and Cunningham, 2005; Arvanitis et al., 2008; Halverson et al., 2008), only a handful of published studies have explored the effects of plant polyploidy on their interactions with mutualistic soil microbes (Segraves and Anneberg, 2016; Segraves, 2017). Furthermore, most of these studies focus on mutualistic fungi (Tesitelova et al., 2013; Sudova et al., 2014), with relatively few testing the effects of polyploidy on mutualisms with rhizobia (but see Table 1).

Table 1.

Summary of published studies testing the effects of plant polyploidy on the legume–rhizobia mutualism. Studies are organized by A, Approach; B, Plant Taxa; and C, Polyploid Type. E, General Outcome for each study summarizes whether polyploids (P) have enhanced (>), reduced (<) or no difference (=) in nodulation traits and/or host benefit relative to diploids (D). F, Specific Outcomes for each study always compare polyploid plants to diploids and are organized by traits (I–VI) within the hierarchy

A. Approach	B. Plant taxa	C. Polyploid type	D. Ploidy levels tested	E. General outcome	F. Specific outcomes					Reference
					I	II	IV	V	VI
1. Natural Comparisons
	Arachis	Allopolyploid	2x–4x	P > D	Polyploids produce more nodules	Polyploids produce larger nodules	Polyploids fix N at a higher rate		Polyploids produce more biomass	Stalker et al. (1994)
	Glycine wightii (Neonotonia wightii)	Autopolyploid	2x–4x	P > D	Polyploids nodulate earlier and produce more nodules				Polyploids produce more biomass and have higher N content	Diatloff and Ferguson (1970)
	Medicago sativa	Autopolyploid	2x–4x	P > D	Polyploids produce more root biomass and nodules	Polyploids produce nodules with greater average biomass			Polyploids do not differ in shoot biomass	N. J. Forrester et al., unpubl. res.
	Stylosanthes hamata, S. seabrana	Allopolyploid	2x–4x	P > D				Polyploids form effective symbioses with unique and more strains		Date (2010)
	Trifolium ambiguum	Autopolyploid	2x–6x	P > D	Polyploids nodulate earlier and produce more nodules	Polyploids produce smaller nodules		Polyploids form effective symbioses with more strains	Polyploids produce more biomass	Hely (1957)
	Trifolium ambiguum	Autopolyploid	2x–4x–6x	P > D	Polyploids nodulate earlier and produce more nodules	Polyploids produce larger nodules			Polyploids produce more biomass	Evans and Jones (1966)
	Trifolium ambiguum, T. pratense, T. repens	Autopolyploid	2x–4x–6x	P > D				Polyploids form effective symbioses with more strains		Beauregard et al. (2004)
	Trifolium pratense	Autopolyploid	2x-4x	P ≤ D	Polyploids do not differ in timing of nodulation and produce fewer nodules				Polyploids do not differ in N content	Nilsson and Rydin (1954)
	Trifolium pratense	Autopolyploid	2x–4x	P > < D	Polyploids nodulate later and produce fewer nodules	Polyploids produce larger nodules			Polyploids produce more biomass	Weir (1961b)
	Trifolium pratense	Autopolyploid	2x–4x	P > D	Polyploids nodulate earlier, and produce larger roots and produce more nodules				Polyploids produce more biomass	Thilakarathna et al. (2012)
	Trifolium pratens, T. repens	Autopolyploid	2x–4x	P > < D	Polyploids produce more nodules (T. pratense) or fewer nodules (T. repens)	Polyploids produce larger nodules			Polyploids produce more (T. pratense) or less (T. repens) biomass	Weir (1961a)
	Trifolium repens	Allopolyploid	2x–4x	P ≤ D	Polyploids produce fewer nodules	Polyploids do not differ in nodule size			Polyploids produce less biomass	Weir (1964)
	Trifolium subterraneum	Autopolyploid	2x–4x	P ≤ D	Polyploids do not differ in timing of nodulation and produce fewer nodules				Polyploids do not differ in biomass	Nutman (1967)
2. Phylogenetically Informed Comparisons
	Glycine dolichocarpa, G. syndetica, G. tomentella	Allopolyploid	2x–4x	P ≥ D	Polyploids have higher rates of root hair deformation, produce more root biomass, and do not differ in nodule number	Polyploids produce larger nodules		Polyploids form effective symbioses with more strains	Polyploids produce more biomass	Powell and Doyle (2016)
3. Experimental Manipulations
	Medicago sativa	Autopolyploid	2x–4x–8x	P ≥ D	Polyploids do not differ in timing of nodulation or nodule number		Polyploids fix N at a higher rate		Polyploids have greater N content	Leps et al. (1980)
	Medicago sativa	Autopolyploid	2x–4x–8x	P ≥ D	Polyploids do not differ in nodule number	Polyploids do not differ in nodule size	Polyploids do not differ in N fixation rate		Polyploids produce more biomass	Pfeiffer et al. (1980)
	Phaseolus aureus	Autopolyploid	2x–3x–4x	P ≥ D	Polyploids nodulate earlier, produce larger tap roots, have higher rates of rhizobial infection, and produce fewer nodules	Polyploids produce larger nodules		Polyploids differ in nodule occupancy of rhizobial symbionts	Polyploids produce more biomass	Kabi and Bhaduri (1978)

The legume–rhizobia mutualism has significant impacts on global ecosystems as it is a key regulator of nitrogen (N) cycles in natural and agricultural environments (Herridge et al., 2008; Vitousek et al., 2013). Moreover, N is an essential and limiting resource for plants (Vitousek et al., 2002), and legumes associated with rhizobia have greater plant biomass and reproductive success (Daehler, 1998; Ndlovu et al., 2013). Although the effects of ancient whole genome duplication (WGD) on the legume–rhizobia mutualism have been well studied (Cannon et al., 2010, 2015; Doyle et al., 2011; Li et al., 2013), we do not fully understand the direct effects of plant polyploidy on key features of this interaction or the potential ecological and evolutionary consequences. The following paragraphs briefly summarize the salient features of the mutualism and consider novel ways in which polyploidy could directly alter it.

THE LEGUME–RHIZOBIA MUTUALISM

As a model system for studying mutualisms, rhizobia fix atmospheric N into ammonia, a compound usable by the plant hosts, in exchange for carbon and other photosynthetic resources from their host plant (Heath and Tiffin, 2007; Jones et al., 2007). Legume taxa exhibit variation in rhizobial infection method, nodulation type, products of N₂ fixation and other mutualism traits (Sprent, 2009). Root hair infection and differentiation of rhizobia within symbiosomes are two of the most common features among nodulating legume taxa (Sprent, 2009; Ferguson et al., 2010; Sprent et al., 2013) and will therefore be the focus of this review. For these legume taxa, the mutualism is initiated when legumes release flavonoids into the soil, triggering free-living rhizobia to produce signalling molecules, ‘Nod factors’ (Wang et al., 2012). Nod factors are perceived by Nod factor receptors of the plant host, stimulating root hair deformation and the development of nodules (Wang et al., 2012). Following successful initiation of the symbiosis, rhizobia enter the developing root nodules and differentiate into ‘bacteroids’ that fix atmospheric N (Wang et al., 2012). In legume taxa that produce indeterminate nodules, rhizobia terminally differentiate into bacteroids and lose the ability to reproduce, whereas in legume taxa that produce determinate nodules rhizobia retain the ability to reproduce (Kiers et al., 2003). Root nodules provide protective environments for N fixation to occur (Gage, 2004; Heath and Tiffin, 2007), and the amount of oxygen (O₂) within nodules is strictly regulated because O₂ is required for rhizobial respiration, yet also irreversibly inhibits nitrogenase and the amount of N fixed (Hunt and Layzell, 1993). Several factors regulate O₂ concentration within nodules, primarily nodule permeability and leghaemoglobin (Hunt and Layzell, 1993).

Because the process of symbiotic N fixation can be costly to plants and rhizobia, the interaction is finely regulated to ensure cooperation among partners (Kiers and Denison, 2008; Sachs et al., 2010). Although regulation can occur via multiple mechanisms, two primary ways in which legume hosts can stabilize cooperation with their rhizobial symbionts are partner choice, establishment of the symbiosis with beneficial rhizobial partners based on recognition signals (e.g. flavonoids, Nod factors), and host sanctions, the ability of a plant to assess nodule efficiency and invest more in efficient nodules than inefficient ones (Kiers and Denison, 2008). Despite our extensive understanding of the establishment and maintenance of the legume–rhizobia mutualism, little is known about how plant polyploidy directly affects mutualism traits, whether it immediately increases plant host access to fixed N and, if so, by what mechanism(s).

ANCIENT POLYPLOIDY AND THE LEGUME–RHIZOBIA MUTUALISM

Although the direct effects of plant polyploidy on the legume–rhizobia mutualism remain unresolved, studies evaluating the effects of ancient WGD on the evolution of nodulation suggest that polyploidy may have enhanced key aspects of the mutualism (Cannon et al., 2010; Doyle, 2011; Young et al., 2011; Li et al., 2013). Ancient WGD was not required for the evolution of nodulation, but the genetic material acquired and retained from a WGD event in the Papilionoideae is hypothesized to have led to enhanced and more complex interactions with rhizobia (Cannon et al., 2010; Young et al., 2011; Li et al., 2013). Notably, the Papilionoideae is the largest and most geographically widespread subfamily within the legumes and 90 % of taxa exhibit nodulation (Sprent, 2007, 2009).

Hypotheses about whether WGD led to enhancements of the legume–rhizobia mutualism focus primarily on gene copies retained during the papilionoid WGD event (~58 Mya) that function in establishment and maintenance of mutualism (Young et al., 2011; Li et al., 2013). Young et al. (2011) determined that several nodulation genes retained from the WGD event have undergone sub- or neofunctionalization in Medicago truncatula, thereby increasing the complexity of genes involved in rhizobial signalling (e.g. flavonoids, Nod factor receptors) and mutualism function (e.g. nodule-specific cysteine-rich peptides, leghaemoglobins). These patterns are also found across the Papilionoideae subfamily; Li et al. (2013) determined that a portion of duplicated genes retained from the papilionoid WGD event diverge in expression patterns and function in establishment (e.g. rhizobial signalling, nodule organogenesis, rhizobial infection) and maintenance of mutualism (e.g. nutrient exchange). Furthermore, Werner et al. (2015) suggest that genome duplications may reduce the rate of symbiotic loss and increase symbiotic persistence over evolutionary time.

The relationship between ancient WGD and the evolution of nodulation in legumes is complex and warrants further investigation (Cannon et al., 2015; Werner et al., 2015); however, these studies support an overall role of polyploidy in enhancing key aspects of the legume–rhizobia mutualism. Evaluating the immediate and direct effects of plant polyploidy on the mutualism, in addition to the effects of ancient WGD, will clarify the relationship between polyploidy and improvements in nodulation as well as uncover underlying mechanisms.

THE DIRECT EFFECTS OF POLYPLOIDY ON THE LEGUME–RHIZOBIA MUTUALISM

Polyploidy could directly enhance the legume–rhizobia mutualism by increasing the quantity and/or quality of rhizobial symbionts hosted, which may occur by altering plant traits that function in the establishment and maintenance of the mutualism. The ways polyploidy can directly affect the legume–rhizobia mutualism can be organized into an effects hierarchy (Fig. 1) and the weight of current evidence for each node within this framework is then evaluated.

Fig. 1.

Framework of hypotheses for how plant polyploidy might directly enhance the legume–rhizobia mutualism. The framework is structured into a hierarchy with the predicted outcome that polyploid plants have greater access to fixed nitrogen via enhanced symbioses with rhizobia. Enhanced symbioses can broadly be categorized by improvements in the quantity and/or quality of rhizobial symbionts hosted. Specific mechanisms for how polyploidy can directly alter plant traits that affect the symbiosis are proposed. Hypotheses that have been never been tested are outlined in thin boxes, hypotheses that have been tested in one to five published studies are outlined in medium boxes, and hypotheses that have been tested in six or more published studies are outlined in thick boxes.

To do this, an exhaustive review of studies of the effects of plant ploidal level on one or more components of the framework was conducted. Specifically, ISI Web of Science was searched using the key words ‘polyploid* AND nodul*’ and ‘tetraploid* AND nodul*’ for studies published between 1900 and 2016; seven studies were identified. The references that cited these seven studies were then evaluated and nine additional published studies were found, as well as data from Forrester et al. (University of Pittsburgh, USA, unpubl. res.), for a total of 17 studies in this dataset (Table 1). Three approaches were used to test the effects of plant polyploidy on the mutualism (Table 1 A): (1) Natural Comparisons, in which traits are compared among natural diploids and polyploids within or among species (n = 13 studies), (2) Phylogenetically Informed Comparisons, in which polyploids are compared to their isogenic diploid progenitors with known time of WGD events (n = 1 study); and (3) Experimental Manipulations, in which polyploid plants are synthesized and compared to their diploid progenitors (n = 3 studies). Experimental Manipulation studies allow for separating the effects of polyploidy from the effects of hybridization and other evolutionary changes since the WGD event, but generating neopolyploids is challenging (Shi QH et al., 2015) and studies using this approach are limited (Table 1 A3). Although studies using Natural Comparisons and Phylogenetically Informed Comparisons approaches do not test the direct and immediate effects of polyploidy on the mutualism, they can inform whether polyploid plants have altered and/or enhanced relationships with rhizobia over evolutionary time.

Studies in the dataset tested ploidy effects in 24 species across six genera; however, over half (nine of 17 studies) used Trifolium species (Table 1 B). Furthermore, 13 of the 17 plant taxa are autopolyploids with only four using allopolyploid taxa (Table 1 C). The majority of studies (11 of 17) compared diploids and tetraploids, but several included other ploidy levels (triploids, hexaploids and octoploids; Table 1 D). All studies were conducted in pots, test tubes or jars in either glasshouses or growth chambers (Supplementary Data Table S1). These studies reveal long-standing interests in the effects of polyploidy on the legume–rhizobia mutualism, as 11 of the 17 studies were conducted between 1954 and 1980. They also reveal a striking gap in experimental studies addressing this question, especially given recent advancements in genetic and genomic techniques (Dufresne et al., 2014).

Although the relatively small number of studies limits quantitative analyses, results from these studies can be synthesized using several approaches to gain insight into the direct effects of plant polyploidy on the legume–rhizobia mutualism. First, a qualitative synthesis of general outcomes across studies allows for identifying broad patterns of the effects of plant polyploidy on the mutualism (Table 1 E). Variation in approach, origin of polyploid plants, nodulation traits and experimental methods among studies may lead to idiosyncratic or species-specific outcomes when synthesizing results across the dataset; thus, in-depth details about each study are provided in Table S1. Second, evaluating specific outcomes of polyploidy on key mutualism traits (e.g. nodule number, plant N content; Table 1 F) aids in determining specific mechanisms by which polyploidy alters the mutualism. To assess the weight of current evidence for specific traits in the hierarchy, results from all studies in the dataset were organized by each trait in Supplementary Data Table S2. Third, considering case studies using the Experimental Manipulation approach provides insight into the direct and immediate effects of polyploidy on mutualism traits. In addition to these perspectives, this paper discusses data from studies of the effects of ancient WGD on the mutualism as well as studies of synthetic neopolyploid plants alone to test the immediate effects of WGD on plant traits (e.g. flavonoid composition, photosynthetic rate) to predict their effects on the legume–rhizobia mutualism.

OVERVIEW OF FRAMEWORK

Fundamental features of polyploidy such as increased cell size and alterations to genetic content and activity (Song et al., 1995; Levin, 2002; Beaulieu et al., 2008; Shi X et al., 2015) could directly and immediately enhance the legume–rhizobia mutualism, thereby allowing plants to access more fixed N, ultimately increasing plant growth and reproductive success (Fig. 1 VI; Parker, 1995; Heath and Tiffin, 2007; Munoz et al., 2016). Such enhancements could result from increases in the quantity or the quality of rhizobial symbionts hosted by legume plants.

First, enhancements in the quantity of rhizobial symbionts could be achieved if polyploid plants host more bacteroids than diploids. Direct changes in root architecture resulting from polyploidy (e.g. increase in root length and volume; Kulkarni and Borse, 2009) could enhance infection rate by rhizobia, thereby increasing the total number of nodules produced and bacteroids hosted (Fig. 1 I; Nutman, 1967; Kabi and Bhaduri, 1978). Enlarged cell size immediately resulting from polyploidy may increase nodule size and subsequently the number of bacteroids contained within nodules (Fig. 1 II; Kondorosi et al., 2000; Beaulieu et al., 2008; Maroti and Kondorosi, 2014). Additionally, WGD may directly alter plant host factors that control terminal bacteroid differentiation, thereby increasing the number and symbiotic efficiency of bacteroids hosted by polyploid plants relative to diploids (Fig. 1 III; Mergaert et al., 2006; Oono and Denison, 2010; Van de Velde et al., 2010; Kondorosi et al., 2013).

Second, enhancements in the quality of rhizobial symbionts hosted by legume plants could be achieved through two additional pathways: improving the nodule environment for rhizobia or by altering the identity of rhizobial symbionts engaged in the mutualism. In terms of the nodule environment, polyploidy might immediately change O₂ and nutrient diffusion rates into nodules and leghaemoglobin quantity and functions relative to diploids, thereby providing a more efficient environment for N fixation to occur (Fig. 1 IV; Robson and Postgate, 1980; Denison and Layzell, 1991; Hunt and Layzell, 1993; Warner and Edwards, 1993; Levin, 2002). Moreover, polyploid plants may have more photosynthetic resources to allocate to nodules than diploid plants (Warner and Edwards, 1993; Levin, 2002; Ramsey and Schemske, 2002). In addition, changes in plant chemistry resulting from polyploidy could affect the identity of rhizobial symbionts via partner choice and host sanctioning mechanisms (Fig. 1 V; Levy, 1976; Levin, 2002; Powell and Doyle, 2015).

Despite the numerous pathways by which polyploidy could enhance the legume–rhizobia symbiosis, few studies have tested any specific mechanisms. The thickness of boxes within the hierarchy in Fig. 1 reflects the number of published studies that have explicitly tested each hypothesis. Hypotheses that have been never been tested are outlined in thin boxes, hypotheses that have been tested in one to five published studies are outlined in medium boxes, and hypotheses that have been tested in six or more published studies are outlined in thick boxes. The following paragraphs formalize hypotheses for how polyploidy might directly enhance the quantity and quality of mutualists hosted by legumes, these are confronted with current evidence, and areas in great need of empirical work are thus highlighted.

I. QUANTITY OF SYMBIONTS: NODULE NUMBER

Increased cell size and genomic changes resulting from polyploidy may alter root and nodule traits, leading to the production of more nodules that can accommodate more rhizobial symbionts than diploids (Fig. 1 I; Kondorosi et al., 2000; Levin, 2002; Beaulieu et al., 2008; Melino et al., 2012).

Nodule number is partially influenced by timing of nodulation, root size and architecture, and autoregulation of nodulation (Nutman, 1967; Diatloff and Ferguson, 1970; Kabi and Bhaduri, 1978; Reid et al., 2011; Thilakarathna et al., 2012). Reduced time to nodulation might occur if polyploidy alters plant signalling molecules that function in mutualism establishment (e.g. flavonoids, Nod factor receptors; Powell and Doyle, 2015). Plants that nodulate earlier have more time to develop root nodules, which could increase nodule production and result in a greater quantity of bacteroids hosted by the plant (Hely, 1957; Evans and Jones, 1966; Diatloff and Ferguson, 1970). Early effective nodulation is thought to be particularly important for plant survival and fitness in N-limited environments (Diatloff and Ferguson, 1970).

Across all studies, there were no consistent effects of polyploidy on time to nodulation: five studies found that polyploid plants nodulated earlier and four studies found the opposite (Table 1 FI). However, synthetic neotetraploids of Phaseolus aureus produced nodules significantly earlier than its diploid progenitors, suggesting that plant polyploidy immediately reduces time to nodulation. In contrast, synthetic neotetraploid and neooctoploid Medicago sativa plants did not differ from their diploid progenitors in time to nodulation (Table 1 A3 FI). While plant polyploidy might not directly and consistently reduce time to nodulation, these results might also reflect variation in experimental approach across the studies (Supplementary Data Table S1). Moreover, five of the nine studies either did not conduct statistical analyses or only report anecdotally that diploids and tetraploids differ in time to nodulation (Nilsson and Rydin, 1954; Hely, 1957; Weir, 1961b; Evans and Jones, 1966; Diatloff and Ferguson, 1970).

Enhancements in root length and lateral root production due to polyploidy can enhance rhizobial infection rate and subsequently increase nodule production per plant (Nutman, 1948, 1967; Kabi and Bhaduri, 1978). Across all studies, four of four found that polyploid plants produced roots with greater size or biomass than diploid plants (Table 1 FI). Although not a legume, in Capsicum annuum, synthetic neotetraploid plants produced longer primary roots and more lateral roots than diploids, suggesting an immediate effect of polyploidy on root size and morphology (Kulkarni and Borse, 2009). In P. aureus, synthetic neotetraploid plants had significantly greater volumes of tap and lateral roots, and a higher infection rate by rhizobia (Table 1 A3 FI; Kabi and Bhaduri, 1978). Consistent with this, Powell and Doyle (2016) found a higher rate of root hair deformation in allopolyploid Glycine dolichocarpa relative to its diploid progenitors (Table 1 A2 FI). Combined, these results show positive, direct effects of plant polyploidy on root size and architecture, and rhizobial infection rate.

However, increases in root morphology and rhizobial infection rate of polyploid plants did not lead to increases in nodule production. Across all studies, polyploid plants did not consistently produce more nodules than diploids: seven studies found that polyploids produced more nodules than diploids and eight found the opposite (Table 1 FI). Allopolyploid G. dolichocarpa did not differ in nodule production compared to its diploid progenitors (Table 1 A2 FI; Powell and Doyle, 2016). Of the three studies that used synthetic polyploids to evaluate the direct effects of polyploidy on nodule production, one found the neotetraploids produced fewer nodules than diploids (Kabi and Bhaduri, 1978), while the other two found no significant differences among diploids and polyploids (Table 1 A3 FI; Leps et al., 1980; Pfeiffer et al., 1980).

Lack of ploidy effects on nodule production may be due to variation in methods among experimental studies (Supplementary Data Table S1) or any number of factors that are known to affect nodule number (e.g. plant biomass, rhizobium genotype, environmental conditions; Heath and Tiffin, 2007; Regus et al. 2015). Because production and investment in nodules can be energetically costly to plant hosts, nodule production is regulated via autoregulation of nodulation (Caetano-Anolles and Gresshoff, 1991; Reid et al., 2011). Autoregulation of nodulation may function similarly in diploid and polyploid plants and explain the lack of ploidy effects on nodule production. This process occurs in response to host infection condition and soil N availability and is characterized by a nodulation phenotype in which nodules form near the crown of the roots and decrease along the root surface (Reid et al., 2011). Autoregulation of nodulation occurs systemically and involves a signalling circuit between root and shoot tissue, ultimately restricting the production of additional nodules (Reid et al., 2011). Although the molecular basis of this process is relatively well understood (Ferguson et al., 2010; Reid et al., 2011), it remains unclear whether and how plant polyploidy directly alters it and how it may constrain differences in nodule production in diploid and polyploid plants (Fig. 1 I, thin boxes).

II. QUANTITY OF SYMBIONTS: NODULE SIZE AND BIOMASS

Even if polyploidy does not directly alter nodule number, it may increase nodule size, resulting in a greater quantity of symbionts hosted by polyploid plants relative to diploids (Fig. 1 II; Heath and Tiffin, 2007; Regus et al., 2015). Genome size is strongly correlated with cell size across 101 angiosperm species (Beaulieu et al., 2008) and polyploidy directly increases cell size in synthetically produced Capiscum annuum, Chamerion angustifolium, Vicia cracca and other plant taxa (Maherali et al., 2009; Kulkarni and Borse, 2009; Munzbergova, 2017). Therefore, polyploids may be predisposed to hosting large numbers of bacteroids because larger cells are needed to accommodate N-fixing bacteroids; indeed, many legumes undergo endoreduplication in nodule tissue to achieve greater cell sizes (Mergaert et al., 2006; Kondorosi et al., 2013; Maroti and Kondorosi, 2014). Endoreduplication of nodule tissue has been detected in legume species with indeterminate, determinate and lupinoid nodules (González-Sama et al., 2006; Kondorosi et al., 2013). However, the content and distribution of polyploid nuclei vary across legume hosts, and some legume taxa (e.g. Glycine) do not undergo endoreduplication of nodule tissue at all (Schwent et al., 1983; González-Sama et al., 2006).

Of the taxa that do undergo endoreduplication of nodule tissue, polyploid plants may produce nodule cells with higher ploidal levels than diploids, and thereby accommodate a greater quantity of bacteroids. Consistent with this, in nodules of isogenic diploid, tetraploid and octoploid M. sativa plants, diploid plants produced nodules with mostly tetraploid nuclei, tetraploid plants produced nodules with tetraploid and octoploid nuclei, and octoploid plants produced nodules with mostly octoploid nuclei (Shanklin and Schrader, 1986). Alternatively, even if ploidal level of nodule cells is the same for diploid and polyploid plants, nodule cells of polyploids may reach their maximum ploidal level (e.g. 32C or 64C) faster than nodule cells of diploids because they experience fewer cycles of endoreduplication (Kondorosi et al., 2000; González-Sama et al., 2006).

Across all studies, seven of ten found that polyploids produced larger nodules or nodules with greater biomass than diploids (Table 1 FII). Allopolyploid G. dolichocarpa produced nodules with greater biomass than both of its diploid progenitors (Table 1 A2 FII; Powell and Doyle, 2016). Synthetic neotetraploid plants of P. aureus also produced larger nodules than its diploid progenitor, supporting an immediate and direct effect of polyploidy on nodule size (Table 1 A3 FII; Kabi and Bhaduri, 1978). However, nodule size did not differ between diploid and synthetic neotetraploid M. sativa plants (Pfeiffer et al., 1980). These studies suggest that polyploidy directly increases nodule size and biomass, although these effects may depend on host taxa, symbiont taxa or both.

While these results suggest polyploid plants ought to host more rhizobial symbionts per plant than diploids, additional work is needed to evaluate this hypothesis as well as the potential underlying mechanisms (Fig. 1 I and II, thin boxes). Although the hypotheses regarding endoreduplication and ploidal level of nodule cells in diploid and polyploid plants are theoretically possible, empirical tests are lacking and therefore we can only speculate about potential effects of plant polyploidy on these traits. Experimental Manipulation studies evaluating the direct effects of polyploidy on the timing of nodulation, root architecture and nodule number will be imperative to tease apart the direct effects of polyploidy from other evolutionary changes that have occurred since the WGD event. Moreover, to our knowledge, no published studies have measured bacteroid quantity within nodules of diploid and polyploid plants. Such experiments are essential for determining whether polyploid plants host more bacteroids than diploids, thereby increasing access to fixed N and host benefit from the mutualism.

III. QUANTITY AND QUALITY OF SYMBIONTS: TERMINAL BACTEROID DIFFERENTIATION

In addition to changes in the quantity of the mutualism via nodule number and size, plant polyploidy might directly alter plant host factors regulating terminal bacteroid differentiation, thereby increasing the number and symbiotic efficiency of bacteroids hosted by polyploid plants relative to diploids (Fig. 1 III; Oono & Denison, 2010). Terminal bacteroid differentiation occurs when rhizobia enter the plant host cell and undergo cell expansion, genome endoreduplication and membrane permeabilization (Van de Velde et al., 2010; Kereszt et al., 2011; Maroti et al., 2011; Alunni and Gourion, 2016). Terminal bacteroid differentiation is regulated by plant antimicrobial peptides, ‘nodule-specific cysteine-rich peptides’ (NCRs; Van de Velde et al., 2010). NCRs were identified in legumes of the inverted repeat-lacking clade and functional homologues of NCRs were recently found in the genus Aeschynomene, but terminal bacteroid differentiation does not occur in all legume taxa (Van de Velde et al., 2010; Maroti et al., 2011; Alunni and Gourion, 2016). Notably, plants that impose terminal bacteroid differentiation on rhizobia have more symbiotically efficient bacteroids and benefit more from the mutualism than plants that do not (Oono and Denison, 2010).

NCRs exhibit extensive diversity (e.g. M. truncatula contains over 600 NCRs) and are hypothesized to differ in function, modes of action and bacterial targets; yet many specific functions remain unresolved (Farkas et al., 2014; Maroti and Kondorosi, 2014; Horvath et al., 2015). While no studies have explicitly tested the direct effects of plant polyploidy on the composition and function of NCRs, genome duplication is known to alter expression patterns of polypeptides and peptide transporters in Brassica and Utricularia, respectively (Albertin et al., 2006; Lan et al., 2017). Moreover, in M. truncatula, ancient WGD is hypothesized to have enhanced the legume–rhizobia mutualism because many amplified gene families, including the NCR gene family, have nodule-specific functions (Young et al., 2011). If polyploidy directly enhances the diversity and functions of NCRs, then polyploid plants may have a greater ability to regulate terminal bacteroid differentiation, thereby increasing bacteroid quantity and symbiotic efficiency. However, no published studies have tested whether plant polyploidy directly alters the composition and functions of NCRs or the process of terminal bacteroid differentiation (Fig. 1 III, thin boxes); thus, there is insufficient data to decisively conclude whether plant polyploidy affects these mutualism traits.

IV. QUALITY OF SYMBIONTS: NODULE ENVIRONMENT

The other primary pathway by which polyploidy could enhance the legume–rhizobia mutualism is by improving the quality of the symbiosis, and this could be achieved by enhancing the nodule environment (Fig. 1 IV). Improvements in the nodule environment may allow for finer regulation of O₂ content within nodules, which is critical for rhizobial growth and nitrogenase function (Robson and Postgate, 1980; Sheehy et al., 1983; Kiers et al., 2003). Nodule permeability and leghaemoglobin are two key factors that regulate O₂ concentration within nodules (Robson and Postgate, 1980; Denison and Layzell, 1991; Hunt and Layzell, 1993), both of which could be altered by polyploidy (Warner and Edwards, 1993; Kondorosi et al., 2000; Levin, 2002).

Nodule permeability is primarily limited by one or more layers of densely packed cells that comprise the nodule inner cortex (Denison and Layzell, 1991; Hunt and Layzell, 1993; Denison, 2015). Enlargements in cortex cell size due to polyploidy might increase cortex thickness or adjust the size and distribution of intercellular spaces in the inner cortex layer, which could either decrease or increase nodule permeability relative to nodules produced by diploid plants.

Leghaemoglobin is an O₂-binding protein that facilitates O₂ diffusion to respiring bacteroids (Appleby, 1984; Hunt and Layzell, 1993), and its content within nodules is correlated with N-fixing ability (Appleby, 1984). Polyploidy can have drastic effects on plant genomes by altering gene expression patterns and sub- and neofunctionalization of duplicated gene copies (Levin, 2002; Doyle et al., 2008; Li et al., 2013; Shi X et al., 2015); thus, polyploid plants may produce more leghaemoglobin or have modified functions of leghaemoglobin gene copies compared to diploids. Consistent with this, Young et al. (2011) found that the leghaemoglobin gene family was amplified in the M. truncatula genome following WGD and contains nine symbiotic leghaemoglobins (double those present in Lotus japonicas and Glycine max), supporting the hypothesis that ancient WGD provided the genetic material to increase the complexity of rhizobial symbioses.

Another consequence of increased cell size due to polyploidy is a reduction in the surface area to volume ratio of the cell, which can influence the rate of nutrient exchange (Kondorosi et al., 2000). Reduced surface area to volume ratios of polyploid cells may result in a greater barrier to O₂ diffusion into nodules of polyploid plants, thereby providing a more efficient environment for nitrogenase function (Appleby, 1984). There are no published studies of the direct effects of polyploidy on the surface area to volume ratio of nodules and subsequent impacts on nutrient exchange, but a study on the effects of ancient WGD hypothesizes a positive effect of polyploidy on nutrient exchange in the legume–rhizobia symbiosis. In the Papilionoideae, paralogues derived from the WGD event that function in nutrient exchange (e.g. ammonium assimilation) have been retained across many papilionoid taxa (Li et al., 2013). Moreover, many gene families that function in nutrient exchange have been amplified following this WGD event, suggesting that polyploidy may have provided genes to enhance the symbiosis (Li et al., 2013).

In addition to potential changes in O₂ concentration and nutrient exchange rates, polyploid cells often have greater metabolic and transcriptional activity than diploid cells (Levin, 2002; Doyle et al., 2008; Shi X et al., 2015). Therefore, nodules that grow via endoreduplication may have an increased ability to provide energy and nutrients to rhizobia for the metabolically costly process of nitrogen fixation (Kondorosi et al., 2000; Kondorosi and Kondorosi, 2004; Mergaert et al., 2006). Since photosynthate supply and N fixation rate are positively correlated (Lawrie and Wheeler, 1973; Singleton and van Kessel, 1987; Walsh et al., 1987), if polyploid plants provide more photosynthetic resources to rhizobia within nodules, then they may acquire more fixed N via the symbiosis. Studies experimentally manipulating ploidy level have found that polyploidy directly increases photosynthetic rate and chloroplast number per cell, although these changes do not always scale to the entire plant (Warner and Edwards, 1993; Levin, 2002; Maherali et al., 2009). Although these data suggest that polyploidy can directly alter photosynthetic processes, which may result in polyploid plants having more photosynthates to allocate to nodules, empirical tests are lacking and therefore we can only speculate about potential effects of plant polyploidy on resource allocation to nodules.

Despite the numerous ways in which polyploidy may directly improve the nodule environment and increase access to fixed N, limited work is available to evaluate these hypotheses. While specific mechanisms for how polyploidy may enhance the nodule environment have not been explicitly tested, two of three studies found that polyploid plants fix N at a higher rate than diploids (Table 1 FIV). This result suggests that polyploid plants have an increased ability to fix N relative to diploids, but it is not clear whether this is due to direct modifications of the nodule environment via polyploidy. To our knowledge, no studies have explicitly tested whether polyploidy directly alters nodule structure and permeability, leghaemoglobin production and function, nutrient exchange or photosynthetic supply to nodules (Fig. 1 IV, thin boxes). Research addressing these hypotheses will be particularly insightful for understanding whether polyploidy directly improves the nodule environment allowing polyploid plants to access more fixed N.

V. QUALITY OF SYMBIONTS: IDENTITY OF RHIZOBIAL SYMBIONTS

The final pathway by which polyploidy might enhance the quality of the legume–rhizobia mutualism is via the identity of rhizobia engaged in the symbiosis relative to diploids (Fig. 1 V). Although ensuring cooperation in the legume–rhizobia mutualism is complex, legume hosts use two primary mechanisms, partner choice and host sanctions, to influence the identity and efficiency of their rhizobial partners (Sachs et al., 2004, 2010; Kiers and Denison, 2008).

Partner choice is the establishment of the symbiosis with rhizobial partners based on recognition signals (e.g. flavonoids, Nod factors, Nod factor receptors; Sachs et al., 2004; Kiers and Denison, 2008), which are genetically determined and may be altered by polyploidy (Stacey et al., 2006; Young et al., 2011; Li et al., 2013; Powell and Doyle, 2015). Polyploidy can directly increase the composition, concentration and diversity of flavonoids produced by the host plant (Levy, 1976; Levin, 2002), thereby broadening the suite of symbionts solicited for the symbiosis (i.e. host promiscuity; Li et al., 2013; Powell and Doyle, 2015).

Although not a legume, synthetically created autotetraploids of Phlox drummondii produced 14 novel flavonoids that were not present in their diploid progenitors, supporting a direct effect of plant polyploidy on flavonoid composition (Levy, 1976). Additionally, the flavonoid biosynthetic pathway in M. truncatula expanded considerably after WGD (Young et al., 2011) and Li et al. (2013) found at least eight enzymes in the flavonoid biosynthetic pathway were retained following the WGD event in the Papilionoideae.

Similar expansions have been observed for Nod factor receptors of M. truncatula (Young et al., 2011) as well as other Papilionoideae taxa (Li et al., 2013). Specifically, the Nod factor receptor (NFP) and transcription factor (ERN1) retained from the papilionoid WGD event exhibit nodule-enhanced expression patterns in M. truncatula, potentially reflecting sub-functionalization of ancestral genes following WGD (Young et al., 2011). Consistent with this, Li et al. (2013) found duplicated genes retained from the papilionoid WGD event that amplified the LysM receptor gene family, which are key components to Nod factor receptors. Together, these studies suggest that polyploidy can increase the abundance and diversity of flavonoids and Nod factor receptors, which can lead to enhanced and more complex signalling to rhizobial partners (Young et al., 2011; Li et al., 2013; Powell and Doyle, 2015).

Across four studies, all found that polyploids could form effective symbioses with a broader range of rhizobial symbionts than diploids (i.e. greater host promiscuity; Table 1 FV). Moreover, in synthetic autotetraploid P. aureus, diploids and tetraploids differed in nodule occupancy when co-inoculated with two rhizobial strains, suggesting an immediate effect of polyploidy on the identity of rhizobial symbionts hosted within nodules (Table 1 A3 FV; Kabi and Bhaduri, 1978).

Although limited, these data support the hypothesis that polyploid plants have the potential for increased host promiscuity and altered communities of rhizobial symbionts relative to diploids, and this has been confirmed in five cases. Additional studies testing nodulation capabilities of diploid and polyploid plants across diverse rhizobial strains will provide insight into whether polyploidy increases host niche breadth and thereby access to more beneficial symbionts. Studies evaluating whether polyploidy directly enhances the abundance, composition and/or diversity of flavonoids and Nod factor receptors will be critical for understanding the mechanistic basis of partner choice and host promiscuity apart from subsequent evolution that occurred after WGD (Fig. 1 V, thin boxes).

The second key mechanism for regulating the identity of rhizobia occupying root nodules is host sanctioning, the ability of the plant to assess nodule efficiency and invest more in efficient nodules than inefficient ones (Kiers and Denison, 2008; Sachs et al., 2010; Regus et al., 2014). The process of N fixation imposes a high metabolic cost for rhizobia; thus, ineffective rhizobia may have increased fitness relative to effective rhizobia (Kiers et al., 2003; Kiers and Denison, 2008; Sachs et al., 2010). In the context of the legume–rhizobia mutualism, where a host species often interacts with multiple symbionts, host sanctions can ensure cooperation among partners (Kiers and Denison, 2008; Sachs et al., 2010). Hosts are hypothesized to impose sanctions via several mechanisms, but primarily by limiting the supply of carbon or O₂ to inefficient nodules and allocating more photosynthetic resources to highly efficient nodules (Singleton and van Kessel, 1987; Kiers et al., 2003). If polyploidy alters O₂ permeability and concentration within nodules, as described with regards to the nodule environment, then polyploids may also exhibit differences in host sanctioning abilities relative to diploids. Moreover, if polyploid plants have increased photosynthetic resources relative to diploids, then allocating photosynthates to highly effective nodules may result in an even greater amount of fixed N acquired via the symbiosis. Although it is theoretically possible for plant polyploidy to alter host sanctions, no studies have explicitly addressed whether diploid and polyploid plants differ in host sanctioning abilities nor tested any of the underlying mechanisms proposed here (Fig. 1 V, thin boxes).

VI. ACCESS TO FIXED NITROGEN VIA ENHANCED SYMBIOSES

Direct alterations in the quantity and quality of rhizobial symbionts due to polyploidy could allow polyploid plants greater access to fixed N (Fig. 1 VI), thereby increasing their biomass and reproductive success relative to diploids (Parker, 1995; Heath and Tiffin, 2007; Munoz et al., 2016). Host benefit from the mutualism can be tested by comparing N content of plant tissue or overall plant size (Munoz et al., 2016); although fitness estimates (e.g. seed production) would be best, it is difficult to assay nodule traits and plant reproductive success simultaneously (Regus et al., 2015).

Across all studies, most (11 of 15) found that polyploid plants had greater N content, size and/or biomass than diploids (Table 1 FVI). Most tested the effects of polyploidy on plant biomass when plants were inoculated with single rhizobial strains. Importantly, only two cases directly compared plant biomass to uninoculated controls, the most salient metric of host benefit from the mutualism, within ploidy level, both of which found polyploids produced more biomass when inoculated with rhizobia than diploids (Evans and Jones, 1966; Leps et al., 1980). However, the work of Weir (1961b) can shed additional light on this issue, as our post hoc comparison between inoculated and uninoculated plants within ploidy level in his study revealed that polyploids had greater increases in plant biomass when inoculated with rhizobia than diploids.

Taken together, these data suggest that polyploid plants benefit more from the mutualism than diploids, but it is important to note that many do not include uninoculated controls and/or rely on indirect measures of host benefit. Comparisons of growth of inoculated to uninoculated plants within ploidy levels (i.e. host growth response) for diploids and polyploids are essential to separate the effects of the mutualism from the effects of ploidy alone. If possible for the plant taxa, assaying nodule traits and plant reproductive success (e.g. seed set) within an experiment will aid in determining the effects of the mutualism on host fitness. In addition, experiments measuring N content derived from the mutualism rather than the environment among diploid and polyploid plants can be conducted using ¹⁵N methodologies, acetylene reduction assays or other techniques (Anglade et al., 2015; Chalk et al., 2016). Such studies are imperative to determine whether polyploid plants have increased access to N relative to diploids and whether polyploid plants benefit more when engaged in the symbiosis as opposed to obtaining N solely from the environment.

RECOMMENDATIONS FOR FUTURE WORK

Our synthetic framework makes clear the myriad ways in which plant polyploidy can directly and immediately affect how legumes interact with their rhizobial symbionts as well as the magnitude of the benefits they derive from the interaction. However, the literature review also illustrates that we do not yet have a consensus on whether polyploidy directly enhances host benefits from the mutualism and we have very limited understanding of the mechanisms that underlie the variation in results achieved thus far (Fig. 1, thin and medium boxes). Previous work using Natural Comparisons, Phylogenetically Informed Comparisons and Experimental Manipulation approaches as well as studies evaluating the effects of ancient WGD on the mutualism provide strong support for the role of polyploidy in enhancing key aspects of the symbiosis, yet rigorous experimental tests are lacking. Here, three key areas in need of attention are highlighted to clarify the direct effects of plant polyploidy on the legume–rhizobia mutualism: improved experimental tests, untested mechanistic hypotheses and studies in natural environments.

Several limitations in the work conducted thus far are the small number of studies that have experimentally tested the effects of plant polyploidy on the mutualism and the lack of variation in plant taxa, rhizobial taxa, nodulation type and polyploid type within these studies. Only three studies have used the Experimental Manipulation approach and only one study has used the Phylogenetically Informed Comparisons approach to compare isogenic diploid and polyploid plants (Table 1 A2,3). Additional studies using the Experimental Manipulation approach (Table 1 A3) by synthesizing auto- and allopolyploids will be particularly informative for assessing the immediate and direct effects of polyploidy on the mutualism, and allow separation of these from subsequent adaptation of the host or rhizobia following WGD (Segraves, 2017). Another limitation from the previous work is lack of taxonomic diversity, as most studies used Trifolium species (Table 1 B). Studies across a broader range of legume hosts are needed to test aspects of this framework and gain a generality. Moreover, these studies exhibit limited diversity in nodulation type as 12 of the 17 used legume taxa that produce indeterminate nodules. It would be interesting to determine how hypotheses within this framework vary for legumes with different nodulation types (e.g. desmodioid, aeschynomenoid, lupinoid) as well as additional mechanisms that can be included in the framework relevant to these different nodulation types (e.g. hormones, flavonoids that affect auxin fluxes; Grunewald et al., 2009; Ferguson et al., 2010; Sprent et al., 2013). Lastly, the majority of legume taxa used in these studies are autopolyploids (Table 1 C); thus, additional tests of allopolyploid legumes and their known diploid progenitors are essential for evaluating how the effects of polyploidy as well as hybridization impact the mutualism.

Another major gap is that many mechanisms in this framework have never been tested (Fig. 1, thin boxes). Although several studies have quantified ploidy effects on mutualism traits that alter the quantity of rhizobial symbionts (e.g. nodule number, nodule size and biomass), there remains a lack of understanding of how plant polyploidy influences symbiont quantity inside the nodule. Studies comparing bacteroid quantity and function as well as the process of terminal bacteroid differentiation in isogenic diploids and polyploids will provide critical insight into whether and how polyploidy affects the quantity and quality of bacteroids. Moreover, there is limited understanding of the mechanisms that may underlie variation among diploid and polyploid plants in terms of symbiont quantity. Studies testing whether and how plant polyploidy alters autoregulation of nodulation and endoreduplication of nodule tissue may reveal pathways by which polyploidy has directly enhanced mutualisms with rhizobia.

The current data also highlight key gaps in our understanding of how plant polyploidy affects the quality of rhizobial symbionts hosted by legumes (Fig. 1 III–V, thin boxes). To our knowledge, no studies have tested whether polyploidy alters the process of terminal bacteroid differentiation, thereby increasing the quality of bacteroids hosted by polyploid plants. Studies assessing how polyploidy directly alters the internal structure of nodules of diploids and polyploids are urgently needed to clarify how differences in cell size can impact nodule cortex structure, O₂ concentration, nutrient exchange and N fixation capacity. Tests of photosynthetic rates and resource allocation to nodules will also contribute to a mechanistic understanding of whether and how plant polyploidy enhances the nodule environment, potentially increasing host access to fixed N. Lastly, studies testing whether isogenic diploids and polyploids differ in signalling to rhizobia (e.g. flavonoids, Nod factor receptors) and subsequent effects on the identity of rhizobial symbionts will inform how polyploidy might directly alter host promiscuity and sanctioning mechanisms.

Finally, all aspects of the framework should be evaluated in ecologically relevant contexts (Segraves, 2017). Because all experimental studies were conducted in either a glasshouse or growth chamber (Supplementary Data Table S1) it remains unclear how plant polyploidy affects the mutualism in natural environments. Moreover, mutualism traits and host benefit are often context-dependent and influenced by environmental conditions (Heath and Tiffin, 2007; Kiers et al., 2010). For instance, variable parameters such as light limitation and nutrient availability in the soil are likely to affect the relative importance of specific mechanisms (e.g. efficiency of N fixation, allocation of O₂ and photosynthates), and subsequent differences between diploid and polyploid plants. Studies conducted in natural environments and evaluating how specific environmental parameters alter mutualism traits are essential for addressing the ecological and evolutionary consequences of plant polyploidy on the symbiosis. Interestingly, polyploidy and the bacterial mutualism are hypothesized to enhance plant invasion success (Daehler, 1998; te Beest et al., 2012; Pandit et al., 2014), and it would be particularly informative to evaluate the proposed mechanisms among diploid and polyploid taxa in native and non-native habitats.

CONCLUSIONS

The conceptual framework reveals tantalizing support for the role of plant polyploidy in directly enhancing the legume–rhizobia mutualism and provides novel mechanistic hypotheses that may underlie this pattern, but it also highlights many unexplored avenues that warrant further investigation. Thus, it makes clear where future work on the effects of plant polyploidy on the legume–rhizobia mutualism will be most beneficial. Such work is even more pressing in light of current global concerns such as food security and climate change, yet we cannot address these challenges without a thorough understanding of the direct effects of polyploidy on the mutualism as well as the underlying mechanisms.

SUPPLEMENTARY INFORMATION

Supplementary data are available online at https://academic.oup.com/aob and consist of the following. Table S1: In-depth details about each study in the review, including experimental methods, plant taxa information, origin of polyploid plants and nodulation traits. Table S2: Results from all studies in the review organized by trait within the hierarchy.

FUNDING STATEMENT

This work was supported by the NSF Graduate Research Fellowship [1247842] to NJF and the NSF DEB [1241006] to TLA.