Potential effects of a high CO2 future on leguminous species

Why this research Matters Legumes provide an important source of food and feed due to their high protein levels and many health benefits, and also impart environmental and agronomic advantages as a consequence of their ability to fix nitrogen through their symbiotic relationship with rhizobia. As a result of our growing population, the demand for products derived from legumes will likely expand considerably in coming years. Since there is little scope for increasing production area, improving the productivity of such crops in the face of climate change will be essential. While a growing number of studies have assessed the effects of climate change on legume yield, there is a paucity of information regarding the direct impact of elevated CO2 concentration (e[CO2]) itself, which is a main driver of climate change and has a substantial physiological effect on plants. In this review, we discuss current knowledge regarding the influence of e[CO2] on the photosynthetic process, as well as biomass production, seed yield, quality, and stress tolerance in legumes, and examine how these responses differ from those observed in non‐nodulating plants. Although these relationships are proving to be extremely complex, mounting evidence suggests that under limiting conditions, overall declines in many of these parameters could ensue. While further research will be required to unravel precise mechanisms underlying e[CO2] responses of legumes, it is clear that integrating such knowledge into legume breeding programs will be indispensable for achieving yield gains by harnessing the potential positive effects, and minimizing the detrimental impacts, of CO2 in the future.

atmosphere in the form of the highly potent greenhouse gas, nitrous oxide (N 2 O; Ladha, 2016;Stagnari et al., 2017). Furthermore, the production of synthetic ammonia presently consumes 1.5% of total global primary energy (US Energy Information Administration, 2015) and generates further N 2 O and carbon dioxide (CO 2 ) as by-products (Foyer et al., 2016). Therefore, decreasing our use of such fertilizers would minimize the environmental footprint of crop production considerably.
Due to their health, cropping and environmental benefits, as well as the fact that our global population is expected to reach upwards of 9.8 billion by 2050 and 11.2 billion by 2100 (United Nations, 2017), the demand for leguminous crops will almost certainly escalate substantially in coming years. Although small overall global increases in their production have been observed over the last half century, with the exception of soybean, these have been achieved mainly through expansions of cropping area (Foyer et al., 2016), which suggests that the rate of genetic yield enhancement in these crops has been relatively low. Considering that the majority of arable land is already being utilized for crop production, and that even less may be available in the future as a result of factors such as urbanization, salinization, and desertification (Alexandratos & Bruinsma, 2012), it is imperative that efforts are made to harness the substantial untapped potential present in legumes for yield improvements.
Unfortunately, the climate-related changes that are predicted to occur within this century will lend further challenges to meeting demand in the future. Rising levels of anthropogenic greenhouse gases are leading to global warming, with 1°C increases already evident compared to pre-industrial times (Allen et al., 2018) and up to 4°C increases possible by the end of the century under certain emission scenarios (Collins et al., 2013). Escalating temperatures also bring about other downstream climatic effects, such as increases in the severity and incidence of droughts, floods, and other extreme weather events (Allen et al., 2018), which all have an impact on legume production. While small increases in temperature may have positive effects in the short term, at least in certain temperate regions, increasingly high temperatures and drought will eventually reduce biomass and seed production, as well as quality (Myers et al., 2014). Indeed, negative effects on legume production arising from climate change are already apparent in many regions of the world.
For example, soybean (Glycine max) yields in the United States have declined by 2%-4% for every 1°C increase in temperature over the growing season between 1994 and 2013, which has led to losses of US$11 billion (Mourtzinis et al., 2015).
While much attention is being paid to the impact of greenhouse gas emissions on crop productivity in the context of these secondary climate change effects, less importance has been placed on the direct consequences of increased greenhouse gas levels. The rapid rise in atmospheric CO 2 concentration ([CO 2 ]) that is occurring due to a variety of anthropogenic factors, including our continued reliance of fossil fuels and deforestation, is a major driving force of climate change (Solomon et al., 2009). Currently, levels of this greenhouse gas have already increased more than 40% since pre-industrial times and now exceed 400 µmol/mol (Mauna Loa Observatory measurement as of June 5, 2019). This is the highest that CO 2 levels have been at any time in the past 650,000 years and possibly the last and could reach as high as approximately 900 µmol/mol by the end of the century (Meinshausen et al., 2011).
Since CO 2 acts as the primary substrate for photosynthesis, which is critical for crop growth, such a large alteration in its atmospheric levels will have direct implications on all agricultural systems (Tausz et al., 2013). There is also growing evidence to suggest that alterations in [CO 2 ] can have a substantial impact on various other physiological and developmental processes in plants, which could exacerbate consequences for future production. However, due to the distinct physiology of legumes, these effects may be somewhat atypical compared to other plant species. In this review, we will discuss the potential impacts of elevated [CO 2 ] (e[CO 2 ]) on legumes in the context of photosynthesis, yield, quality, and ability to withstand various types of stress. Advancing our knowledge in this area and furthering downstream research to clarify these effects will be of the utmost importance for the development of new, high-yielding, climate-smart cultivars as a contribution to food security in the future.
limitation, initial increases in the rate of carbon assimilation stemming from e[CO 2 ] are believed to eventually lead to an accumulation of non-structural carbohydrates in leaf tissues as a result of replete sinks and a lack of additional sink capacity (i.e., the ability to generate new sinks and/or expand existing sinks; Erice et al., 2011;Lewis et al., 2002). This build-up of carbohydrates to saturating levels may be responsible for the transcriptional down-regulation of several genes encoding enzymes involved in photosynthesis, such as RuBisCO (Moore et al., 1999;Sheen, 1994), resulting in a negative feedback effect on the photosynthetic process. Correspondingly, photosynthetic acclimation has been found to be more pronounced when plants are grown in small pots where root growth, and hence sink activity, is limited (Ainsworth et al., 2002).
The nitrogen limitation theory, on the other hand, derives from the fact that acclimation to e[CO 2 ] also tends to be associated with diminished leaf nitrogen concentrations (Leakey et al., 2009). Since nitrogen uptake is not augmented during initial biomass expansions that are typically observed under e[CO 2 ], and may even decrease following long-term exposure to e[CO 2 ] (Guo, Sun, Li, Liu, et al., 2013;Parvin et al., 2018), the amount of existing nitrogen must be distributed over a more substantial quantity of tissue. This results in a dilution of its concentration and leads to a nutrient deficiency that hinders plant growth (Tausz et al., 2013). Decreased nitrogen partitioning to leaves (Seneweera, 2011) and a smaller proportion of the available nitrogen being allocated to RuBisCO synthesis under e[CO 2 ] (Drake, Gonzalez-Meler & Long, 1997; Sage, Sharkey, F I G U R E 1 Generalized schematic diagram of photosynthetic light and dark reactions, as well as photorespiration, in C 3 plants. Light reactions are depicted in green, dark reactions are indicated in red, and the photorespiratory pathway is shown in purple. Major enzymes are indicated in black ovals. 2PG, 2-phosphoglycolate; 3PGA, 3-phosphoglycerate; ADP, adenosine diphosphate; ATP, adenosine triphosphate; ATPS, ATP synthase; CBB, Calvin-Benson-Bassham; cytb 6 f, cytochrome b 6 f complex; NADP(H), nicotinamide adenine dinucleotide phosphate; PSI, photosystem I; and PSII, photosystem II  & Seemann, 1989) likely also contribute to reductions in leaf nitrogen concentrations. While declines in RuBisCO and other photosynthetic enzymes, as well as reduced leaf nitrogen concentrations, are clearly associated with photosynthetic acclimation under e[CO 2 ], the specific mechanisms behind these phenomena remain to be resolved and it is unclear whether they are a direct cause, or simply a consequence, of this process (Gamage et al., 2018).
Due to their unique physiology, legumes often appear to be especially responsive to long-term e[CO 2 ] exposure in the form of increased photosynthetic rates and reduced susceptibility to photosynthetic acclimation, as well as associated gains in vegetative biomass, compared to non-nodulating plants (e.g., Ainsworth et al., 2002;Ainsworth et al., 2004;Jablonski, Wang, & Curtis, 2002;Lee, Barrott, & Reich, 2011;Lee et al., 2003;Scheelbeek et al., 2018; Tables 1 and 2; Table S1). This may stem, at least in part, from the presence of root nodules and their associated symbiotic relationship with N 2 -fixing bacteria, which provides an additional strong sink for carbohydrates and allows plants to be less constrained by soil nitrogen limitations, thereby ameliorating their capacity to maintain foliar nitrogen concentrations under e[CO 2 ] (Cabrerizo et al., 2001;Lam et al., 2012;Rogers et al., 2006). Correspondingly, aboveground biomass has been found to be enhanced under long-term e[CO 2 ] in nodulated Medicago truncatula, but not an N 2 -fixation-deficient mutant (Guo, Sun, Li, Liu, et al., 2013).
However, results have been conflicting (Table 1), and at least under certain circumstances, nodulated leguminous species have been found to undergo acclimation in a manner similar to non-N 2 -fixing plant species (e.g., Aranjuelo et al., 2005;Lee, Barrott, & Reich, 2011;Sanz-Sáez et al., 2010;Xu, Gifford, & Chow, 1994). Various limiting factors, including inadequate water supply (Aranjuelo et al. 2009), low intensity of solar radiation (Ainsworth et al., 2003), and insufficient phosphorus availability (Duchein, Monicel, & Betsche, 1993), as well as other environmental pressures, may contribute to such findings. However, in at least a proportion of cases, it appears that the presence of nodules alone cannot support the high sink and/or nitrogen demands required under e[CO 2 ]. Indeed, although some studies have found foliar nitrogen concentrations to be maintained or increased under e[CO 2 ] in legumes (e.g., Sanz-Sáez et al., 2010;Parvin et al., 2018), reductions in foliar nitrogen levels are just as commonly encountered (e.g., Bertrand et al., 2007;Bourgault et al., 2016;Bourgault et al., 2017) albeit often to a lesser extent than their non-leguminous C 3 counterparts (e.g., Lam et al., 2012). While there is evidence that unlike non-N 2 -fixing plants, net photosynthetic rates may not be tightly linked to foliar nitrogen concentrations in plants that fix N 2 (Adams et al., 2016), supplementation with exogenous nitrogen fertilizer has been found to alleviate photosynthetic acclimation in some legumes (Sanz-Sáez et al., 2010;Schortemeyer et al., 1999). This supports the notion that at least under certain circumstances, N 2 -fixation is inadequate to maintain growth rates under long-term e[CO 2 ].
In line with this, photosynthetic acclimation and growth responses under e[CO 2 ] in various legume species have been found to correlate with their capacity for nodule formation (Ainsworth et al., 2004;Cernusak et al., 2011;Guo, Sun, Li, Liu, et al., 2013). As such, variations in this trait could be responsible for at least a proportion of the genotypic and/or interspecific differences observed under CO 2 enrichment (West et al., 2005). Although relatively little is currently known concerning the role of nodulation in photosynthetic acclimation to e[CO 2 ], it appears that nodule numbers and mass typically increase under these conditions (e.g., Cabrerizo et al., 2001;Lam et al., 2012;Lee et al., 2003;Sreeharsha, Sekhar, & Reddy, 2015). Despite these increases, however, it has been suggested that feedback inhibition of rhizobial nitrogenase activity under e[CO 2 ] may lead to an overall down-regulation of nodule activity and thus N 2 -fixation (Hartwig et al., 1994), which could also contribute to the photosynthetic acclimation sometimes seen in legumes grown under CO 2 enrichment.
The strain of rhizobial inoculant can also have a substantial impact on a legume's response to e[CO 2 ] (Bertrand et al., 2007;Sanz-Sáez, Erice, Aguirreolea, Muñoz, et al., 2012), possibly through an increased efficiency of N 2 -fixation in particular symbiont/legume genotype combinations (Provorov & Tikhonovich, 2003 (Sanz-Sáez et al., 2015). This lack of stimulation was suggested to result, at least in part, from competition with native rhizobia in the soil, which could be one cause of discrepancies observed between experiments carried out in pots (e.g., Bertrand et al., 2007) versus field conditions. However, it is also possible that a reduced investment in nodule biomass resulting from the use of a rhizobial strain exhibiting a particularly high N 2 -fixation efficiency may lead to a reduction in nodule sink strength, thus limiting the plant's ability to take advantage of e[CO 2 ] conditions (Sanz-Sáez et al., 2015).
In addition to their relationship with rhizobia, legumes can also establish symbioses with arbuscular mycorrhizal fungi, which promote nutrient uptake (particularly phosphorus) and act as an additional carbon sink (Mortimer, Pérez-Fernández, & Valentine, 2008).
This implies that their presence might also prevent photosynthetic acclimation and enhance biomass production under e[CO 2 ] in leguminous species. However, results have been highly variable to date (e.g., Baslam, Erice, & Goicoechea, 2012;Gavito et al., 2000;Goicoechea et al., 2014;Jakobsen et al., 2016;Olesniewicz & Thomas, 1999), which may derive from differences in experimental conditions, including genotype/species assessed, level of CO 2 to which the plants were subjected, and rooting volume.
The importance of the source/sink ratio on the propensity for photosynthetic acclimation under e[CO 2 ] in legumes has also been demonstrated by the fact that the cutting or grazing of a legume crop, which is the typical manner in which forage legumes such as alfalfa are managed, can improve photosynthetic response (Erice et al., 2006a) and enhance biomass gains under long-term e[CO 2 ] (Erice et al., 2006b). This occurs because the removal of shoots, which normally act as source organs that provide carbohydrates to sink tissues such as roots and young developing shoots, initiates the mobilization of carbon and nitrogen reserves from roots to shoots (Avice et al., 2003). This essentially results in an inversion of source and sink organs, leading to the generation of a new strong sink. Defoliation also triggers the degeneration of existing nodules in legumes (Vance et al., 1979), and their regeneration gives rise to yet another sink (Erice et al., 2006a).

| EFFEC T OF ELE VATED [CO 2 ] ON LEG UME S EED YIELD
Although increases in net photosynthetic rates and biomass production are often observed under long-term e[CO 2 ], such responses have not always been found to correlate well with reproductive traits across plant species in general (Ackerly & Bazzaz, 1995;Farnsworth & Bassaz, 1995). Therefore, seed yield responses under e[CO 2 ] are not necessarily easy to predict. In the majority of legumes assessed thus far, seed yields tend to increase under long-term e[CO 2 ] compared to ambient [CO 2 ] ( Figure 2, Table 2), mainly as a result of greater pod numbers rather than the number of seeds per pod or individual seed weight (Ainsworth et al., 2002;Bourgault et al., 2016;Kumagai et al., 2015;Li et al., 2019). However, in many instances, these enhancements are often less substantial than those observed in vegetative biomass, and even less so than increases in photosynthesis (e.g., Ainsworth et al., 2002;Bishop et al., 2015;Dijkstra, Schapendonk, & Groenwold, 1993;Prasad et al., 2002). These findings likely arise due to the reductions in harvest index (proportion of total biomass that is partitioned into the seeds) that are frequently observed under CO 2 enrichment (Ainsworth et al., 2002;Bishop et al., 2015;Kumagai et al., 2015;Morgan et al., 2005).
To further complicate matters, environmental factors appear to play a substantial role in seed yield effects under e[CO 2 ], which is evidenced by the finding that smaller stimulations in yield are typically observed in FACE experiments compared to those conducted in controlled environments (Ainsworth et al., 2008). Furthermore, a large amount of intraspecific variability is evident in legume species with respect to the magnitude of yield gains under e[CO 2 ] (Bishop F I G U R E 2 Typical growth and quality effects of long-term e[CO 2 ] exposure in legume species grown under field conditions. Fe, iron; IVDMD, in vitro dry matter digestibility; N, nitrogen; TSS, total soluble sugars; WUE, water-use efficiency; and Zn, zinc TA B L E 1 Examples of effects of long-term a e[CO 2 ] on photosynthetic rate and biomass production in a selection of forage and other non-grain legumes  (Kumagai et al., 2015). However, this did not appear to be a dominant factor when comparisons were made across a broad range of soybean cultivars (Kumagai et al., 2015) and differences in the magnitude of seed yield stimulation under e[CO 2 ] are also observed among cultivars displaying the same growth habit (e.g., Lam et al., 2012).
As is the case for photosynthetic and biomass amplification under e[CO 2 ], seed yield gains also appear to be closely related to plant nitrogen concentrations. Indeed, it has been found that the content of fixed nitrogen in soybean seeds, and hence symbiotic compared to non-leguminous species (Ainsworth et al., 2002;Ainsworth et al., 2004), the stimulation of soybean seed yield does not differ substantially from that of other non-N 2 -fixing C 3 crops such as wheat (Triticum aestivum) and rice (Oryza sativa) (Bishop, Leakey, & Ainsworth, 2014). This supports the notion that N limitation may still contribute to the inability of some legumes to maximize yield responses under e[CO 2 ], and suggests that disparities in N 2fixation capacity, or the ability to maintain nitrogen levels specifically in seeds, may also contribute to cultivar-and species-specific differences in yield gains under these conditions.
Taken together, these findings indicate that legume seed yield responses to e[CO 2 ] may be influenced by a variety of factors, including sink strength, N 2 -fixation efficiency, and the translocation of resources to the seeds. In soybean at least, it has been suggested that historically, breeders may have unintentionally selected for genotypes that are less stimulated by e[CO 2 ] in terms of seed yields. This is evidenced by the fact that at least certain older genotypes have been found to be more responsive to e[CO 2 ] than modern cultivars (Leakey & Lau, 2012;Ziska, Bunce, & Caulfield, 2001).
Although the specific traits that best predict maximum yield responses to e[CO 2 ] in legume species (or other crop species) remain unclear (Bishop et al., 2015;Leakey & Lau, 2012;Ziska et al., 2012), the failure of seed yield increases to parallel the potential provided by the stimulation of canopy photosynthetic rates under e[CO 2 ] implies that there is substantial potential to improve the capacity of legumes to better take advantage of future atmospheric conditions. TA B L E 2 (Continued) elevated levels of CO 2 have also been found to lead to other consequences that could have serious deleterious outcomes in terms of crop quality. These include reductions in protein concentrations (Fernando et al., 2015), as well as declines in the levels of certain vitamins and other macro-and micronutrients (Högy & Fangmeier, 2008;Myers et al., 2014; Figure 2). While all of these effects would be disadvantageous to legume production overall, whether a crop is harvested as a forage (high protein vegetative tissue), oilseed (high oil seed) or pulse (high protein seed), will determine the precise components that are of greatest concern in each case.

| Forage quality
In the case of leguminous forage species, two important parameters for shoot quality include the concentrations of non-structural carbohydrates, which provide a source of rapidly fermentable energy, and crude protein, which supplies the nitrogen building blocks necessary for the production of other proteins by microbial populations within the rumen from which livestock-derived products such as meat and  Table 3).
Unsurprisingly, reductions in leaf nitrogen concentrations are typically associated with a decline in crude protein levels, which has the potential to decrease forage quality and digestibility (Milchunas et al., 2005). Ruminants require at least 70 g protein/kg dry matter (DM) for maintenance, 100-140 g protein/kg DM for growth, and 150 g protein/kg DM for lactation (Izaurralde et al., 2011). Since leguminous forages tend to possess relatively high crude protein values ranging between approximately 170 and 310 g protein/kg DM (Fulkerson et al., 2007;Lee, 2018), their quality may be less impacted by small drops in crude protein than their non-leguminous counterparts. However, if crude protein levels declined to a great enough extent, feed derived from even leguminous forages would need to be supplemented with additional nitrogen to offset these losses under future atmospheric conditions, which would increase the cost of production substantially (Izaurralde et al., 2011).
While high levels of crude protein are beneficial in terms of forage quality, proteins from leguminous forages tend to be digested at a much higher rate within the rumen than the plant cell wall components (cellulose and hemicellulose) that make up the main source of a forage's energy. This asynchrony leads to energy limitation for the ruminal microbial population due to a surplus of forage proteolysis products and a shortage in cell wall breakdown products (reviewed by Singer, Weselake, & Acharya, 2018). A substantial proportion of this superfluous nitrogen is then excreted onto pasture land in urine, resulting in nitrogen losses of up to 70% (Kingston-Smith, Marshall, & Moorby, 2013). As such, it has been suggested that increased concentrations of rapidly digestible soluble carbohydrates in forages could theoretically improve quality by augmenting the availability of a rapidly digestible energy source to balance forage proteolysis and fuel livestock protein biosynthesis (Brito et al., 2008). Under e[CO 2 ] conditions, elevations in non-structural carbohydrates and C:N ratios are typically seen in foliage (e.g., Aranjuelo et al., 2005;Sanz-Sáez et al., 2010), which could improve quality and lead to increased nitrogen use efficiency (Bertrand et al., 2007). Nitrogen excreted by livestock can also lead to the leaching of nitrates into groundwater and/or act as substrate for the generation of the potent greenhouse gas, N 2 O, which are both environmentally deleterious (Oenema et al., 1997;Wachendorf et al., 2008). Therefore, the boost in readily digestible carbohydrates typically observed under e[CO 2 ] could also serve to minimize such negative environmental impacts of livestock production.
The presence of relatively low concentrations (between 3% and 4% DM) of polyphenolic condensed tannins in the vegetative tissues of forages is also highly beneficial to ruminant production due to their ability to complex with plant proteins, thus slowing down their degradation within the rumen and improving nitrogen use efficiency (Aerts, Barry, & McNabb, 1999). As is the case with increased non-soluble carbohydrate concentrations, this leads to enhancements in the production of meat and milk products (McMahon et al., 2000), as well as reductions in greenhouse gas emissions resulting from ammonium excreted in urine (Smith et al., 2008). In addition, their presence has also been found to reduce the incidence of pasture bloat (Tanner et al., 1995), and condensed tannin-containing legumes such as sainfoin, sulla (Hedysarium coronarium), and birdsfoot trefoil are therefore gaining popularity as forages (Acharya et al., 2013). Since condensed tannins are carbon-based and do not contain nitrogen, it is not surprising that their concentration has been found to increase under e[CO 2 ] due to increased carbon availability (Carter, Theodorou, & Morris, 1999;Stiling & Cornelissen, 2007).
Unfortunately, at levels above an approximately 5% DM threshold, condensed tannins become anti-nutritional due to associated reductions in voluntary intake (McSweeney et al., 2001) and impairment of rumen function (Kumar & Singh, 1984). Concentrations rarely reach these levels in condensed tannin-containing leguminous forages under current environmental conditions; however, growth under e[CO 2 ] has been found to increase condensed tannin levels on average 22% in plants (Robinson, Ryan, & Newman, 2012). Therefore, it is possible that future atmospheric conditions could tip quantities over the limit, and further research in this area, along with careful monitoring in the future, is warranted.
Inefficiencies in the fermentation of forages within the rumen also contribute to the livestock-derived emission of methane, which is another greenhouse gas that drives climate change (IPCC, 2007). This occurs through the digestion of cellulosic forage components by microorganisms within the rumen, which produce hydrogen as a by-product that inhibits further fermentation. As a means of preventing reductions in fermentation, the hydrogen (along with CO 2 ) is converted to methane by rumen methanogens. This process utilizes between 2% and 15% of the energy consumed by a ruminant, and the methane is then excreted into the atmosphere (Van Nevel & Demeyer, 1996) process, there is some evidence that CO 2 enrichment may lead to increases in the potentiality of a forage for eliciting methane emissions (Baslam et al., 2013), which would be very detrimental in the context of climate change.
Fiber concentration, which is often estimated as neutral detergent fiber (NDF; comprising cellulose, hemicellulose, and lignin), acid detergent fiber (ADF; comprising cellulose and lignin), and acid detergent lignin (ADL), can also impact forage digestibility.
Lignin in particular is known to be largely indigestible to ruminants, and decreases in this compound are therefore associated with improved forage quality

2012).
In addition to C:N ratio and digestibility, micronutrient levels can also play an important role in forage quality. While particular levels of micronutrients are required for plant growth itself, the presence of these elements in forage vegetative tissues is also of importance for animals that feed upon the plants, which means that sufficiency levels for both plants and animals need to be con-  Pal et al., 2004). However, it appears that patterns of change for micronutrients tend to be similar between foliar and seed tissues in plants overall (Loladze, 2014), which suggests that findings for leguminous seeds in this context (see next section) may also hold true in forage crops and that responses may vary depending on the particular mineral.  Table 4), compared to the 10%-15% reduction in seed protein levels seen in other field-grown C 3 crops (Jablonski, Wang, & Curtis, 2002;Taub, Miller, & Allen, 2008). This is not unexpected seeing as legume seeds acquire the majority of their nitrogen through remobilization from leaf tissues during the early stages of seed filling (Ortez et al., 2019;Schiltz et al., 2005). However, this smaller impact on seed nitrogen levels is not always the case, and in chickpea, for example, greater reductions in seed protein concentration (between 8.4% and 10.2%) have been observed . In this particular instance, such a finding may be attributable to the fact that chickpea has a low capacity for N 2 -fixation compared to other legumes (López-Bellido et al., 2011). Furthermore, a high leaf nitrogen concentration at the flowering stage has not always been found to be a good indicator of a legume's capacity to maintain seed protein levels under e[CO 2 ] overall (Bourgault et al., 2016), and e[CO 2 ] has been found to decrease the partitioning of nitrogen from vegetative tissues to seeds in at least certain cultivars of soybean (Li et al., 2017). In line with this, species-and cultivarspecific differences involving not only the capacity for N 2 -fixation, but also the ability to maintain nitrogen concentration specifically in seeds under e[CO 2 ], have been suggested to play a role in such differential responses (Bourgault et al., 2016;Lam et al., 2012;Li et al., 2017).

Legume
While seed protein concentration is certainly an important consideration in terms of nutritional quality, as is the case in forages, essential nutrients also comprise a vital component. Indeed, it has been estimated that upwards of 2 billion people globally are currently deficient in one or more nutrients, with approximately 1.5 billion being deficient in Zn (Smith & Myers, 2018) and approximately 2 billion lacking in Fe (Viteri, 1998 Hao et al., 2016;Li et al., 2018; 2019; Table 4).

The mechanisms responsible for changes in mineral concentra-
tions under e[CO 2 ] remain to be elucidated; however, it has been speculated that reductions in stomatal conductance seen under these conditions may decrease the uptake of nutrients that depend upon transpiration-driven mass flow (Houshmandfar et al., 2018;McGrath & Lobell, 2013). Alternatively, it has also been proposed that the stimulation of seed yield often seen under e[CO 2 ] may lead to a concomitant "dilution effect" with respect to seed nutrient concentration (Poorter et al., 1997), although this does not appear to be the case for all minerals (Parvin et al., 2019). Whatever the mechanism, it appears that e[CO 2 ] can have both negative and, in certain instances, positive effects on micronutrient levels in legume seeds, and further research will be required to better understand such responses.
In addition to the value of legumes as an important source of protein, certain species such as soybean are utilized as an oilseed crop and are also economically vital to the agricultural sector , it is likely that cultivar-and species-specific differences, as well as growth and environmental conditions, may play a role in these inconsistencies .
Soybeans also contain large amounts of carbon-based isoflavones, which are polyphenolic secondary metabolites considered to have many health benefits, including the inhibition of ovarian and colon cancer cell growth (Chang et al., 2007;MacDonald et al., 2005) and the reduction of serum low-density lipoprotein cholesterol levels (Taku et al., 2007). Although

| EFFEC T OF ELE VATED [CO 2 ] ON S TRE SS RE S P ON S E
The production of reactive oxygen species (ROS), including superox- of aerobic metabolism (Mittler et al., 2011). These molecules are produced constitutively in plants at relatively low levels through photosynthesis, photorespiration, β-oxidation of fatty acids, and the mitochondrial electron transport chain, for example (Apel & Hirt, 2004;Sharma et al., 2012). However, when exposed to abiotic and biotic stress conditions, ROS levels tend to increase rapidly, whereby they function as signal transduction molecules (Baxter, Mittler, & Suzuki, 2014). This leads to alterations in the redox state of regulatory proteins, as well as changes in gene expression and translation, that allow adaptation to stress conditions (Choudhury et al., 2016;Xia et al., 2015). While providing benefits in terms of tolerance to various forms of stress, above certain thresholds ROS can harm plant cells through their oxidation of membrane lipids, as well as the damage they incur to proteins, chlorophyll, and nucleic acids (Foyer & Harbinson, 1994). To diminish these deleterious effects, plants possess mechanisms to scavenge and de-toxify excess ROS through an integrated system of enzymatic and non-enzymatic antioxidants, which also tend to be up-regulated under stress conditions (Das & Roychoudhury, 2014). oxidative damage, which could explain the reduction in total antioxidant capacity that is often evident in legumes under these conditions (Aranjuelo et al., 2008;Gillespie, Rogers, & Ainsworth, 2011;Pritchard et al., 2000). However, it is also possible that a decrease in ROS, which also act as a signal to augment antioxidant defense systems as a means of priming plants to respond to additional stress exposure (Tausz et al., 2013), could also feasibly lead to reduced stress tolerance.

An overall "CO
Alterations in the levels of phytohormones, which are used universally by plants to effect appropriate responses to both abiotic and biotic stress (Xia et al., 2015), are also commonly observed under long-term exposure to e[CO 2 ]. In particular, reductions in basal jasmonic acid (JA) and ethylene (ET) levels, along with the expression of related transcripts, have been encountered in some, but not all, soybean cultivars under e[CO 2 ] Guo et al., 2014;Zavala et al., 2008). Conversely, concomitant increases in the antagonistic salicylic acid (SA)-dependent pathway have been observed across multiple cultivars of both soybean and bean exposed to CO 2 enrichment (Casteel et al., 2012;Mhamdi & Noctor, 2016). Since phytohormones exert their function in part through the activation of ROS production (Xia et al., 2015), it may be that an interplay between phytohormone and ROS levels is responsible, at least in part, for generalized alterations in stress response seen under e[CO 2 ]. response (Kazan, 2015; Figure 3). However, studies carried out thus far suggest that the relationships between CO 2 levels and abiotic stress in terms of plant response and resilience are very complex (Table 5), and large gaps remain in our understanding of these interactions.

| Abiotic stress resilience
In all plant species, including legumes, e[CO 2 ] tends to reduce stomatal conductance, which leads to lower levels of leaf transpiration and hence increased water-use efficiency (WUE; Leakey et al., 2009). Such an improvement in WUE has the potential to lead to soil water conservation that could be utilized later in the growing season (Niklaus, Spinnler, & Körner, 1998 (Clifford et al., 1993) and open-top field chambers with soybean (Rogers, Cure, & Smith, 1986).
However, increased soil water conservation under e[CO 2 ] does not appear to always be the case in practice (e.g., Gray et al., 2016;Parvin et al., 2018;Saha et al., 2011), and early increases in root and shoot biomass seen under e[CO 2 ] may outweigh reductions in transpiration on a leaf-level basis. If this is indeed the case, growth under e[CO 2 ] would instead lead to increased water usage at the whole plant level and thus an earlier onset of drought. This could cause more intense water deficits during developmental stages, which are particularly harmful, diminishing any potential gains that could be incurred through CO 2 enrichment (Bourgault et al., 2017). Furthermore, since N 2 -fixation is particularly sensitive to water stress (Serraj, Sinclair, & Allen, 1998), nitrogen dilution in response to e[CO 2 ] may also occur more readily under drought than well-watered conditions, leading to a higher propensity for photosynthetic acclimation, less substantial yield gains, and reductions in quality (Ainsworth et al., 2002).
Correspondingly, in the majority of studies, photosynthetic and/or yield gains in response to e[CO 2 ] have been found to be reduced or lacking in legumes without an adequate water supply compared to those that were well-watered (e.g., Aranjuelo et al., 2009;Gray et al., 2016;Parvin et al., 2018). In addition, small but significant decreases in legume seed protein concentration have been observed in field pea (Bourgault et al., 2016), lentil (Parvin et al., 2018) and soybean ) under e[CO 2 ] in arid conditions, while no changes or even increases were observed in high rainfall environments in soybean and lentil, respectively Parvin et al., 2018).
With respect to rising temperatures, effects on legume production are likely to depend upon the geographical region, with temperature increases in more temperate climates enhancing e[CO 2 ]-derived yield gains and those in already warm areas leading to a decline in the stimulatory effects of CO 2 (Kromdijk & Long, 2017). Indeed, when temperatures are above an optimal level, legumes tend to be more susceptible to photosynthetic acclimation under e[CO 2 ], even when provided with exogenous nutrients (Sanz-Sáez, Erice, Aguirreolea, Ziska & Bunce, 1994). Such a phenomenon tends to be associated with an absence or reduction in yield enhancements Heinemann et al., 2006;Prasad et al., 2002;Ziska & Bunce, 1994). As is the case with drought stress under e[CO 2 ], these findings likely derive, at least in part, from the decreases in N 2 -fixation often seen under high temperatures (e.g., Aranjuelo, Irigoyen, & Sánchez-Díaz, 2007;Hungria & Kaschuk, 2014), although effects may also be dependent upon the strain of symbiont utilized (Sanz-Sáez, . Reproductive development in plants, including the production of flowers, pollination, and seed filling, tends to be more sensitive to elevated temperatures than vegetative growth or photosynthesis. In line with this, temperatures that are beneficial for biomass production and photosynthesis often prove harmful for reproductive growth, leading to reductions in yield (Prasad et al., 2002).
Therefore, as with drought stress, the timing of heat stress can have a substantial impact on its interactive effect with e[CO 2 ], with high temperatures and/or heat waves incurred during reproductive growth tending to have the most deleterious outcomes (Jifon & Wolfe, 2005;Thomey et al., 2019). In addition, the threshold temperature for seed set has been found to be reduced under e[CO 2 ], which indicates that yield losses due to high temperatures will likely increase with rising CO 2 (Prasad et al., 2002), at least in sub-tropical and tropical regions where average temperatures are already at the optimum for seed production in legumes such as soybean, cowpea (Vigna unguiculata), and peanut (Prasad, Allen, & Boote, 2010).
While the mechanism(s) driving such a phenomenon has yet to be elucidated, a lower tolerance for high temperatures may be related to the finding that long-term exposure to e[CO 2 ] increases leaf temperatures in soybean (Valle et al., 1985) and kidney bean (Prasad et al., 2002) highly multifaceted (e.g., AbdElgawad et al., 2015;Bishop, Leakey, & Ainsworth, 2014;Delahunty et al., 2018). Since optimum temperatures and water availability for photosynthesis, vegetative growth, and reproductive development can differ quite drastically among species (Prasad, Allen, & Boote, 2010), precise interactions will likely be species-dependent. In addition, the severity and timing of stress conditions, genotype, nutritional factors, soil type, and geographical region will almost certainly also play a substantial role in the particular response elicited.  (Casteel et al., 2012;Kazan, 2018;Mhamdi & Noctor, 2016;Zhang et al., 2015;Zhou et al., 2019; Figure 3), with SA playing a key role in local and systemic responses to biotrophic and hemibiotrophic pathogens and JA/ET providing a central function in response to necrotrophic pathogens (Glazebrook, 2005). The constitutive up-regulation of the SA pathway seen in legumes grown under e[CO 2 ] could therefore theoretically lead to increased resistance to biotrophs under these growth conditions. Correspondingly, the severity of downy mildew, caused by the biotrophic pathogen Peronospora manshurica, was found to be consistently reduced in soybean under e[CO 2 ] in three years of a FACE study, despite annual differences in temperature and rainfall (Eastburn et al., 2010).
However, the up-regulation of the SA pathway under CO 2 enrichment has not always been found to be the case in non-legumes such as barley (Hordeum vulgare) (Mhamdi & Noctor, 2016) and tobacco (Nicotiana tabacum) (Matros et al., 2006), suggesting that this may be a species-specific response. Further research will be required to determine whether the response of the SA pathway under e[CO 2 ] also varies among legumes.
Conversely, declines in the JA/ET pathway under e[CO 2 ] Casteel et al., 2012) could conceivably lead to decreased resistance to necrotrophic pathogens. In line with this, while e[CO 2 ] treatment had no significant effect on the incidence of Septoria brown spot of soybean caused by the necrotrophic pathogen Septoria glycines, there was a significant increase in disease severity in all 3 years of the study (Eastburn et al., 2010 (Braga et al., 2006;Kretzschmar et al., 2009;Mhamdi & Noctor, 2016), and augmented foliar carbohydrate concentrations could also potentially alter the growth of at least certain biotrophic pathogens (reviewed by Manning & Tiedemann, 1995). Improvements in resistance could also feasibly be elicited through the production of elevated quantities and altered compositions of leaf surface waxes, as well as increases in the number of epidermal cell layers (reviewed by Kazan, 2018). In addition, the reductions in stomatal conductance observed under e[CO 2 ] have been implicated in increased resistance to stomatal invading pathogens such as bacteria and many biotrophs, as these pathogens would encounter fewer infection sites and lower humidity on leaf surfaces due to stomatal closure (Eastburn, McElrone, & Bilgin, 2011;Li et al., 2015). will also require additional study to develop epidemiological models under climate change for diseases of legumes.

| Pest tolerance
Crop damage sustained from insect herbivores is already a major challenge for the production of many crops, and it is recognized that future atmospheric and climatic conditions could worsen such issues (Gregory et al., 2009 itself has often been shown to decrease plant tolerance through alterations in particular quality traits, defense molecules, and/or antioxidant enzyme activities (e.g., Martin & Johnson, 2011;Stiling & Cornelissen, 2007;Zavala et al., 2008 Figure 3). Such a response has been suggested to derive, at least in part, from the increase in foliar C:N ratios commonly encountered under these growth conditions in many plant species. Since nitrogen is the most limiting resource for phytophagous insects (Mattson, 1980) both constitutively and under induction by herbivory Casteel et al., 2012;Zavala et al., 2008). This could function to suppress the capacity to mount an effective defense against herbivorous insect pests, thus increasing susceptibility. Furthermore, alterations in JA-and ET-signaling pathways under e[CO 2 ] have also been found to correlate with a reduction in the production of cysteine proteinase inhibitors, which act as specific deterrents to coleopteran herbivores . In line with this, gut cysteine proteinase activity was found to be higher in Japanese beetles and Western corn rootworm (Diabrotica virgifera virgifera) that consumed the foliage of soybean grown under CO 2 enrichment than those fed foliage grown under ambient [CO 2 ], which led to increases in their growth and development under e[CO 2 ] .
Phloem-feeding aphids, which comprise one of the most detrimental pests to crop production globally, may also become more problematic under future atmospheric conditions ( Figure 3). Indeed, certain aphid species themselves display enhanced survival, fecundity, and abundance under CO 2 enrichment (Robinson, Ryan & Newman, 2012), which could prove problematic. In legumes, the pea aphid (Acyrthosiphon pisum (Harris)) can be a particular challenge as it exhibits a very wide geographical distribution and feeds on an extensive range of species (Blackman & Eastop, 2000). As is the case with chewing insects, down-regulation of the JA/ET-signaling pathway has been found to occur in M. truncatula during pea aphid infestation under e[CO 2 ] compared to ambient [CO 2 ] (Guo et al., 2014), which may contribute to the increases in susceptibility that are typically observed in this species at elevated CO 2 levels (Guo et al., 2014). Aphid-infested M. truncatula plants have also been found to exhibit lower activities of various antioxidant and secondary metabolism enzymes under e[CO 2 ] compared to ambient [CO 2 ], as well as elevated leaf temperatures and reduced stomatal apertures, which could also play a role in increased susceptibility under these conditions (Guo et al., 2014;O'Neill et al., 2011;Sun, Guo, & Ge, 2016).
Increases in the availability of non-essential amino acids have also been observed in M. truncatula under e[CO 2 ], which may enhance their nutritional quality and thus promote the growth of pea aphid populations (Guo, Sun, Li, Tong, et al., 2013).
Although relatively few studies have been carried out thus far regarding the effect of CO 2 levels on aphid-legume interactions, it is clear that responses are highly genotype-specific. It has been suggested that quantitative and qualitative alterations in foliar amino acids under e[CO 2 ] may play a role in the susceptibility or resistance of particular genotypes under these conditions, although direct evidence of causality is still lacking (Guo, Sun, Li, Tong, et al., 2013;Guo et al., 2014;Johnson, Ryals, & Karley, 2014). Differential genotype-specific responses under e[CO 2 ] have also been attributed to the presence or lack of resistance (R) genes targeting particular aphid biotypes, along with associated effector-triggered immunity (ETI) responses. For example, M. truncatula genotypes possessing an R gene have been found to increase JA signaling and exhibit increased resistance to pea aphid under e[CO 2 ], while genotypes lacking an R gene displayed reductions in JA signaling and decreased resistance to infestations (Sun et al., 2018). However, this does not always appear to be the case, since R gene-dependent resistance to the European large raspberry aphid (Amphorophora idaei) has been found to decrease in red raspberry (Rubus idaeus) grown under e[CO 2 ] (Martin & Johnson, 2011). These findings imply that different cultivars/species may exhibit distinct alterations in their susceptibility to aphid infestations under a future of rising CO 2 levels, which will be an important consideration for downstream breeding endeavors.
Taken together, it seems that overall, plant responses to e[CO 2 ] will likely lead to altered interactions between plants and insects, resulting in more acute and frequent outbreaks of invasive insect pests. Genotypic differences in such responses to particular insect pests suggest that the genetic potential exists in at least certain legumes for improvement of these outcomes in the future. However, further research in this area will be imperative to better our understanding of both the responses of legumes to insect pests under e[CO 2 ] and the precise mechanisms driving these responses in order for such improvements to be realized.

| CON CLUS IONS
Given the importance of leguminous species as sources of high quality food and feed, increasing both vegetative biomass and seed yields, while maintaining quality, will be crucial to food security as our global population soars in coming years. Since substantial expansions in cropping area will not likely be an option, enhancing yield gains using both conventional and biotechnological breeding approaches will be critical to achieving this goal. The development and commercialization of new crop cultivars can take decades; therefore, these new varieties will need to be optimized now for the vastly different atmospheric conditions that will be evident in coming years.
While much research is dedicated to the effects of climate change on crop production within a weather-related framework (i.e., increasing temperatures, drought, flooding), the direct roles of greenhouse gases such as CO 2 are often overlooked. To date, findings regarding the effects of e[CO 2 ] on the yield and quality of leguminous forages and pulse/soybean seed have been inconsistent for the most part. This likely results from the complexity of the responses themselves, as well as the vast number of interacting elements, including genotype/species-specificity, rhizobial strain, CO 2 treatment level, rooting volume, water/nutrient availability, soil type, irradiation level, and stress incidence. Therefore, it currently remains uncertain how future atmospheric conditions will actually impact legume yield and quality. While it seems that legumes may exhibit superior yield benefits and reduced protein decreases under e[CO 2 ] compared to non-N 2 -fixing plants, in many instances they still do not maximize yield gains or completely prevent nitrogen depletion when grown at elevated CO 2 levels.
Indeed, evidence is mounting to indicate that under limiting conditions, which may well be encountered rather frequently in field conditions in a future of climate change, it is probable that overall declines in many of these parameters will instead be the reality. Such a consequence would have a considerable negative effect on food security, as well as human and animal health. While teasing apart the intricate process of e[CO 2 ] response in legumes will certainly not be simple, furthering our knowledge concerning e[CO 2 ]-related outcomes in legumes with respect to yields, quality, and ability to withstand other forms of abiotic and biotic stress will be a requisite for the breeding of new high-yielding legume varieties for the future.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.
[Correction added on 24 May 2021, after first online publication: Conflict of Interest statement added to provide full transparency.]

ACK N OWLED G EM ENTS
The authors are grateful for the support provided by Agriculture and Agri-Food Canada and the Beef Cattle Research Council.

AUTH O R CO NTR I B UTI O N S
SDS conceived the topic of the article. SDS, SC, RYS, US, GC, and SNA all contributed to the writing of the manuscript. All authors read and approved the final version of the manuscript.