Crop growth and yield responses to either phosphorus (P) starvation or rising atmospheric carbon dioxide (CO₂) concentration are well documented. However, responses of the nutritional aspects of crop yield particularly seed or grain to these factors or their interactions have rarely been explored. P deficiency reduces growth, yield, and seed quality of soybean (Ranjan et al., 1962; Israel and Rufty, 1988; Israel, 1991; Singh et al., 2014). However, elevated CO₂ (eCO₂) stimulates crop growth primarily through increased photosynthesis and leaf area (Cure et al., 1988; Prior et al., 1998; Ainsworth et al., 2002). Despite the opposite effects of these two factors on crops, studies that combine these factors have shown that eCO₂ compensates, at least partially, for the adverse effects of P deficiency on plant growth and yield (Cure et al., 1988; Prior et al., 1998; Singh et al., 2013, 2014). Previous studies indicated alterations in the various seed components including minerals and metabolic profiles such as carbohydrates, organic acids, amino acids, protein, or oil, when plants were exposed to stress situations including P deficiency (Walker et al., 1985; Israel, 1991; Gifford et al., 2000; Prior et al., 2008; Rotundo and Westgate, 2009; Singh et al., 2014). Moreover, eCO₂ had been shown to lessen the concentrations of minerals such as nitrogen, phosphorus, iron, and zinc, and protein in the seeds or edible parts of major crops (Högy et al., 2013; Loladze, 2014; Myers et al., 2014). The unwanted changes in the mineral elements or metabolic profiles due to environmental factors in seeds of the main food crops can have profound impacts on the human and animal nutrition.

Deficiency of a specific nutrient can also alter the dynamics of other elements in the plant organs by affecting the uptake and assimilation (Singh and Reddy, 2014; Singh et al., 2014; Gao and Ma, 2015). Previous reports showed that P deficiency increased both N concentrations and the N/P ratio of plant tissues (Singh et al., 2013, 2014). The tissue N/P ratio is a potential indicator of P limitation in crops (Koerselman and Meuleman, 1996; Singh et al., 2014). The soil nutrient status strongly affects the uptake and tissue concentration of mineral elements and metabolic profile (Prior et al., 2008). In general, at the early stage, plant leaves act as a sink for mineral nutrients and carbohydrates, while later they become nutrient sources for the new growth and seed development (Himelblau and Amasino, 2001). Hanway and Weber (1971) reported that approximately half the nutrients from vegetative tissues were mobilized to seeds during reproductive growth stages. However, the transport and metabolism of various seed constituents were affected by the tissue concentration of nutrients and external environmental factors. Despite the commonly observed lower tissue concentrations of mineral nutrients under CO₂ enriched environments, evidence suggests greater nutrient demand by crop plants due to an overall increase in biomass accumulation (Conroy, 1992; Rogers et al., 1993; Singh et al., 2014). Thus, macronutrient deficits, such as low P concentrations, might exert substantial limitations on crop responses to eCO₂. Factors such as dilution of tissue nutrients due to increased photosynthesis and plant size (biomass and yield), thicker leaves, and restricted uptake due to lower transpiration may contribute to the lower tissue nutrient concentrations at elevated CO₂ (Gifford et al., 2000; Prior et al., 2008; Taub and Wang, 2008; Singh et al., 2014). However, the overall yield of nutrients on a per plant or area basis (also called as contents) are often greater under eCO₂ because the total nutrient yield is a function of tissue concentration and total production of biomass or yield (Prior et al., 2008; Singh et al., 2015). Therefore, alterations of the plant growth, nutrient uptake and utilization under P deficiency and eCO₂ are likely to affect the concentration and the total yield of seed components due to the changes in plant metabolic processes. Thus, it is important to recognize the effect of P deficiency on the constituents of grains in major row crops grown under ambient CO₂ (aCO₂) and eCO₂. The response in the nutritional attributes of seeds to P deficiency will be useful to understand its nutritional impact and might offer an opportunity for quality improvement under P-stress.

Reports on the effect of P deficiency on the seed quality of grain crops, such as soybean, are extremely limited. Soybean [Glycine max (L.) Merr.] is one of the most important sources of protein and vegetable oil for both human and animal nutrition. Soybean seeds contain high levels of protein (≈ 42%) and oil (≈ 23%), making it one of the most versatile crops in the world (Dornbos and Mullen, 1992; Medic et al., 2014). Soybean seeds and soy products are also an important source of carbohydrates, amino acids, and mineral elements, which contribute to determining the overall nutritional value (Medic et al., 2014). In addition to genetic factors, environmental factors can considerably alter seed composition (Dornbos and Mullen, 1992; Rotundo and Westgate, 2009; Medic et al., 2014; Bellaloui et al., 2015). Moreover, the crop growth environments might have a varying impact between the concentration and total yield of a given seed constituents (Farhoomand and Peterson, 1968; Bauer et al., 1997). The concentration (e.g., mg g^-1 seed) is an intensive property and reflect the percentage of a given constituent among other, whereas the total yield of a given constituent is an extensive property reflecting the total content (e.g., g plant^-1). The concentration of a seed component is independent of the size of a sample, whereas the yield or content depends on the sample size or total production (Farhoomand and Peterson, 1968). Evidence suggests that the analysis of both the concentration and content of plant tissue components in a given species in response to an environment is important to fully characterize the nutritional quality and nutrient turnover (Bauer et al., 1997). Studies evaluating varying features between concentration and total yield of seed components of row crops in a given environment are extremely limited (Bruns, 2016).

Elemental P is an integral part of various metabolic processes, especially in energy transfer, that might affect the biosynthesis of seed constituents (Chiera et al., 2005). The majority of cropland is poor in P and rapidly depleting its resources, which is of major global concern (Cordell et al., 2009). Moreover, atmospheric CO₂ concentration is expected to double from the current level of approximately 400 μmol mol^-1 by the end of this century (IPCC, 2013). The seed filling processes of soybean depend on the remobilization of resources such as carbohydrates, minerals, and amino acids from plant organs, and plant photosynthesis capacity. All of these processes are influenced by P nutrition and eCO₂, which in combination might affect seed quality traits in soybean (Israel, 1991; Conroy, 1992; Högy et al., 2013; Loladze, 2014; Myers et al., 2014; Gao and Ma, 2015). In the previous reports as a part of this study, we found alteration in the tissue N concentration and lower photosynthesis of soybean under P deficiency (Singh et al., 2014; Singh and Reddy, 2016), however, eCO₂ stimulated growth, and seed yield that might have been influenced by the increased carbohydrate supply during the seed filling processes. Most of the prior studies on seed components often focused on the specific constituents, such as mineral elements or carbohydrates while other on the protein and oil or fatty acids. Also, the majority of these prior studies were conducted under natural environments and were complicated by the occurrence of heat and drought stresses (Ranjan et al., 1962; Prior et al., 2008; Högy et al., 2013; Krueger et al., 2013; Myers et al., 2014; Bellaloui et al., 2015; Gao and Ma, 2015). A comprehensive analysis of soybean seed quality under P deficiency and eCO₂ is lacking. The objectives of this study were to determine the combined effect of P nutrition and growth CO₂ on the concentrations and the yield of soybean seed mineral elements and metabolites. We hypothesize that (a) P deficiency will alter the concentrations but decrease the yield of seed components and (b) eCO₂ will reduce the concentration of seed components but increase the yield, which might depend on the P nutritional-status of soybean.

This study revealed the combined effect of P starvation and eCO₂ on the concentration and yields of soybean seed components under well-watered conditions. The concentration of seed components varied in their sensitivity to P nutrition and growing environments (Ranjan et al., 1962; Israel, 1991; Rotundo and Westgate, 2009; Högy et al., 2013; Loladze, 2014; Medic et al., 2014; Bellaloui et al., 2015). The concentration of amino acids, in particular, showed P × CO₂ interactions, which might be attributable to the lack of their response to eCO₂ under control P treatment. Whereas, yields for majority of components exhibited P × CO₂ interaction due to their minimal response at eCO₂ under the lowest P treatment. The decline in seed and leaf tissues P concentration under P deficiency was reported previously by Singh et al. (2014) and was comparable to other observation in the field (Walker et al., 1985). Remarkably, several mineral elements declined due to P starvation, even though they were not limiting in the rooting media. However, the exceptions to this rule were N, which increased, and Fe and Mn, which were unaffected by P deficiency. Similarly, P deficiency did not affect canola seed Mn concentration when averaged across N-fertilization treatments (Gao and Ma, 2015). The observed high N concentration in P-stress seeds was in agreement with the previous report which also suggested that a marked increase in the fraction of root biomass might have assisted an increased N uptake (Singh et al., 2014). When a nutrient such as P is limiting, excess N in the tissue is often stored as the nitrogenous compounds such as amino acids or protein as observed in this study and elsewhere (Israel and Rufty, 1988; Staswick et al., 1991). In fact, there was a strong relationship (r² = 0.88, P = 0.0005) between the measured protein concentrations and that estimated from seed N across CO₂ (Figure 5G). Lower C/N or higher N/P ratios in response to P-deficiency have been observed previously and might be attributed to lower tissue P and carbon-rich components, such as oil, without a commensurate decrease in the N or N-rich compounds (Vasilas et al., 1984; Singh et al., 2014).

The eCO₂-mediated decreases in leaf N concentration have been reported across P nutrition in soybean and cotton (Singh et al., 2013, 2014). In this study, eCO₂ decreased seed N concentration, primarily under P-stress. The increase in C/N ratio at eCO₂ in soybean seed was minor (1.3–3.2%) and consistent with the observation made in wheat grain (Högy et al., 2013). The seed N/P ratio also increased at eCO₂, when averaged across P treatments, whereas it was not observed in soybean leaves (Singh et al., 2014).

The decrease in K concentration under P-deficiency was in agreement with a previous report and reflected an association between P nutrition status during plant growth and seed K concentration (Krueger et al., 2013). The current study indicated that soybean seed experiences lower Mg, S, Cu, B, and Zn concentrations in P-deficient soils, thus affecting its nutritional quality. The Zn is an essential mineral nutrient in the human diet and necessary for the proper function of immune system and to avoid stunted growth in infants and children (cf. Myers et al., 2015). The observed high Fe concentration under eCO₂ contrasted the earlier reports of either lack of CO₂ effect (Prior et al., 2008) or increased iron concentration at eCO₂ in soybean (Myers et al., 2014). The nutrient status of the rooting medium during the course of experiment significantly affected the mineral uptake and their tissue concentrations (Conroy, 1992; Prior et al., 2008). A continuous supply of Fe and larger root growth at eCO₂ might have facilitated enhanced Fe acquisition in this study resulting into the increased seed iron concentration (Wiersma, 2005).

Sucrose was the most abundant carbohydrate across P nutrition indicating its principal role in energy and carbon supply for the metabolic processes (Fait et al., 2006). The decrease in sucrose concentration under P-stress implied its utilization in the biosynthesis of other seed components. Sucrose might serve as an efficient medium for carbohydrate transport and as a precursor for biosynthetic pathways after hydrolysis into glucose and fructose. Diminished sucrose concentrations under P deficiency were accompanied by enhanced levels of glucose, ribose, and the sugar alcohols, myo-inositol, mannitol, and pinitol, that previously have been shown to accumulate in response to abiotic stresses and might serve as osmolytes to protect cellular structures during stress and the seed desiccation process (Fait et al., 2006; Barnaby et al., 2015). The eCO₂ tended to decrease seeds carbohydrates as also reported previously in wheat grains (Högy et al., 2013).

The concentrations of organic acids including TCA intermediates have been studied in plant leaves, but responses of organic acids to P deficiency and eCO₂ in seeds of major crops are limited (Fait et al., 2006; Barnaby et al., 2015; Sicher, 2015). Organic acids, particularly those that are TCA cycle intermediates, play a vital role in metabolic processes including biosynthesis of secondary metabolites, fatty acids and amino acids (Nelson and Cox, 2008). One of the plausible causes of the organic acid accumulation under P-stress might be its reduced utilization into fatty acid biosynthesis as shown by lower oil concentrations. However, a concurrent accumulation of succinate and amino acids were also reported previously in maturing seeds (Fait et al., 2006). Glycerate is involved in fatty acid metabolism, and its phosphate derivatives are key biochemical intermediates in the glycolytic pathway that converts glucose into pyruvate (Nelson and Cox, 2008). Decreased concentrations of many TCA cycle intermediate under eCO₂ was in agreement with the previous study in soybean leaves (Sicher, 2015).

The consistent increase of free amino acids in P-deficient seeds implied their reduced utilization into the protein lattice as reported previously (Ranjan et al., 1962; Fait et al., 2006). Greater amino acid levels in P-deficient seeds might be attributed to depressed soybean growth rates and varying desiccation processes during the seed maturation. Fait et al. (2006) found an association between increased amino acid levels and mRNA encoding enzymes of amino acid biosynthesis indicating their accumulation during seed desiccation. Among measured amino acids, Gln was the most abundant amino acid in seeds, which was in agreement with the previous finding for other legumes (Yemane and Skjelvåg, 2003). The accumulation of Pro and Putrescine (a polyamine which might resutl form the breakdown of amino acids) in plant tissues in response to abiotic stress likely functions as an osmotic adjustment and as a defense against the reactive oxygen species (Bouchereau et al., 1999; Barnaby et al., 2015). Putrescine also plays a role in seed germination, and its external application has been reported to enhance pod retention and seed yield in Chickpea (Cicer arietinum L.) under cold stress (Nayyar, 2005). A general decrease in seed amino acids at eCO₂ was in agreement with other C₃ species (Högy et al., 2013). Interestingly, concentration of several amino acids showed P × CO₂ interaction, which was attributable to their substantial reduction at eCO₂ under P starvation, but not under control.

Regardless of CO₂ treatment, the P starvation shifted the balance between protein and oil concentrations in soybean seeds. Similar changes exhibiting an inverse relationship between the seed protein and oil concentrations have also been reported previously under various environmental conditions (Dornbos and Mullen, 1992; Fenner, 1992; Rotundo and Westgate, 2009; Rotundo et al., 2009; Singh et al., 2012). The P deficiency appeared to show a trade-off between these two seed components by increasing (≈ 4.7%) the protein while decreasing (≈ 6.4%) the oil concentration by a comparable amount. During the seed filling processes, the accumulation of protein depends more on the remobilized resources from leaves as compared to the oil which mostly rely on the current photosynthesis (Rotundo and Westgate, 2009, 2010). The deceased photosynthesis under P deficiency in soybean has previously been reported as a part of this study (Singh et al., 2014; Singh and Reddy, 2016) and by others (Cure et al., 1988; Israel and Rufty, 1988) which might have led to the lower seed oil concentration. However, the eCO₂ enhanced seed oil concentration (1.7–2.5%) due to a direct stimulatory effect on the photosynthetic processes that was observed in the previous reports (Singh et al., 2014; Singh and Reddy, 2016). Similar to the observation made in this study, Myers et al. (2014) did not find a significant effect of CO₂ on the protein concentration in soybean seeds. However, total seed protein and oil yields were greater at eCO₂ due to increase in seed productivity in this study.

Lower concentration of mineral elements, soluble sugars, and N-rich compounds such as protein and amino acids in plant tissues including seeds under eCO₂ have been frequently reported and found to vary among cultivars or species (Prior et al., 1998, 2003; Gifford et al., 2000; Högy et al., 2013; Myers et al., 2014). Increased plant size and accumulation of photosynthates in plant tissues might partly explain the reduction of mineral concentrations and other components at eCO₂ due to ‘carbohydrate dilution’ (Gifford et al., 2000; Prior et al., 2008). However, concentrations of not all constituents were decreased in response to eCO₂ suggesting possibilities of other mechanisms such as lower transpiration-driven mass flow of nutrients, alteration in the uptake and assimilation, and an increased leaf thickness might also contribute to the lower mineral concentration at eCO₂ in plant tissues (Taub and Wang, 2008; Myers et al., 2014 and references therein). In contrast to the concentration, the total yield (i.e., content) of a majority of the seed components was increased in response to eCO₂, which was closely associated with the increases in the seed production. For example, the degree of eCO₂-mediated enhancement in the yield of various seed components systematically declined as the P supply was reduced, which resulted in a minimal response of eCO₂ at the lowest P treatment. This also resulted in to the observed P × CO₂ interaction for yields of several seed components. The minimal response was attributable to the lack of stimulatory effects of eCO₂ on seed production and various growth and physiological traits under severe P deficiency as found in previous reports (Singh et al., 2014, 2015). Since components yield is a function of the concentration of a given component and total seed production, factors affecting the plant growth and production will also influence the yields of seed components on the per plant or area basis.

In summary, this study largely supported the first hypothesis that P starvation will alter the concentration, but decrease the yield of seed components. P treatment significantly altered the concentration of 36 out of 38 seed components, of which 25 showed increasing while 11 showed decreasing patterns under P deficiency. Thus, P deficiency will alter the concentration seed components likely by reducing oil and mineral elements, but enhancing majority of other components including protein and the stress-responsive metabolites such as sugar alcohols, proline, and glycerate in soybean seeds across CO₂ levels. The P treatment also affected the yield of 36 seed components and on an average majority of them declined except amino acids under P-stress. Remarkably, a number of amino acids compensated, at least partially, the drop in their total yield, especially due to several-fold enhancements in their concentrations under P-stress. Results also partly supported the second hypothesis that eCO₂ will decrease the concentration but increase the yields of seed components which will depend on the P nutritional-status of soybean. The CO₂ treatment significantly affected the concentrations of 28 out of 38 seed components, of which 23 (e.g., P, K, B, sucrose, glucose, citrate, aconitate, and several amino acids) showed decreasing while other (C, iron, Mn, Glycerate, and oil) components showed increasing patterns at eCO₂. Furthermore, CO₂ significantly affected the yield of 27 seed components and majority of them increased, especially at control and 0.10 mM P nutrition. In addition, the irresponsiveness of several amino acids concentration and yields of majority of components to eCO₂ under control and the lowest P treatment, respectively, resulted into P × CO₂ interaction. Thus, eCO₂ appears to compensate, at least partially, for the overall yield of several seed components due to increased seed production, particularly near sufficient P nutrition.

SS and VR conceived and designed the experiment. SS conducted the experiment, collected and analyzed the data, and wrote the manuscript. JB and RS contributed in the seed metabolites analysis and assisted with the manuscript preparation. All author approved the final manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.