Increased temperatures may safeguard the nutritional quality of crops under future elevated CO 2 concentrations

Summary Iron (Fe) and zinc (Zn) deficiencies are a global human health problem that may worsen by the growth of crops at elevated atmospheric CO 2 concentration (eCO 2). However, climate change will also involve higher temperature, but it is unclear how the combined effect of eCO 2 and higher temperature will affect the nutritional quality of food crops. To begin to address this question, we grew soybean (Glycine max) in a Temperature by Free‐Air CO 2 Enrichment (T‐FACE) experiment in 2014 and 2015 under ambient (400 μmol mol−1) and elevated (600 μmol mol−1) CO 2 concentrations, and under ambient and elevated temperatures (+2.7°C day and +3.4°C at night). In our study, eCO 2 significantly decreased Fe concentration in soybean seeds in both seasons (−8.7 and −7.7%) and Zn concentration in one season (−8.9%), while higher temperature (at ambient CO 2 concentration) had the opposite effect. The combination of eCO 2 with elevated temperature generally restored seed Fe and Zn concentrations to levels obtained under ambient CO 2 and temperature conditions, suggesting that the potential threat to human nutrition by increasing CO 2 concentration may not be realized. In general, seed Fe concentration was negatively correlated with yield, suggesting inherent limitations to increasing seed Fe. In addition, we confirm our previous report that the concentration of seed storage products and several minerals varies with node position at which the seeds developed. Overall, these results demonstrate the complexity of predicting climate change effects on food and nutritional security when various environmental parameters change in an interactive manner.

FACE) experiment in 2014 and 2015 under ambient (400 µmol mol -1 ) and elevated (600 µmol 23 mol -1 ) CO2 concentration and under ambient and elevated temperatures (+2.7 °C day and +3.4 °C 24 at night). In our study, eCO2 significantly decreased Fe concentration in soybean seeds in both 25 seasons (-8.7% and -7.7%) and Zn concentration in one season (-8.9%) while higher temperature 26 (at ambient CO2 concentration) had the opposite effect. The combination of eCO2 with elevated 27 temperature generally restored seed Fe and Zn concentrations to levels obtained under ambient 28 CO2 and temperature conditions, suggesting that the potential threat to human nutrition by 29 increasing CO2 concentration may not be realized. In general, seed Fe concentration was 30 negatively correlated with yield suggesting inherent limitations to increasing seed Fe. In 31 addition, we confirm our previous report that the concentration of seed storage products and 32 several minerals varies with node position at which the seeds developed. Overall, these results 33 Introduction 47 Atmospheric CO2 is increasing and projected to reach between 730 and 1020 µmol mol -1 by the 48 end of the century (Collins et al., 2013). The increase in CO2 would be expected to increase leaf 49 photosynthesis and this has been documented experimentally for many C3 species (Ainsworth 50 and Long, 2005, Ainsworth andRogers, 2007) using free-air CO2 enrichment (FACE) 51 technology (Long et al., 2004, Leakey et al., 2009. As a result, biomass production and seed 52 yield are increased but often to a lesser extent than the increase in light-saturated photosynthesis 53 (Long et al., 2004, Leakey et al., 2009). In addition to increasing yield, eCO2 has the potential to 54 reduce the concentration of minerals in seeds and thereby threaten human nutrition, as 55 highlighted in two recent meta-analysis studies. Across a range of C3 plants, Loladze (Loladze,56 2014) reported that eCO2 reduced the concentrations of several minerals in foliar and edible 57 tissues by approximately 8%, with the exception of Mn that was not reduced. Overall, the 58 patterns of mineral changes were similar between foliar and edible tissue with the exception that 59 K was reduced only in edible tissues (Loladze, 2014). The second study (Myers et al., 2014) 60 focused on C3 grains and legumes and reported that concentrations of Zn and Fe were decreased 61 from ~5 to 10% at eCO2. Both studies concluded that eCO2 may result in crops that are less 62 nutritious as a result of reduced mineral concentrations, which would have enormous 63 implications for the use of crops for human food as well as animal feed (Weigel, 2014). 64 The mechanisms responsible for the effects of eCO2 on seed mineral concentrations are not 65 clear, but many biological and physical processes could contribute. For example, reduced 66 transpiration as a result of partial stomatal closure at eCO2 (Bernacchi et al., 2007, Zaman et al., 67 2013) would be expected to reduce the uptake of minerals that are acquired by mass flow, but 68 have less effect on those where movement to the root surface occurs by diffusion (McGrath and  69 Lobell, 2013). Following uptake into the plant, mineral transport continues in the xylem with 70 distribution among stems and leaves (Marschner, 1995). In grasses, nutrient distribution among 71 vegetative organs is known to be controlled independently of transpiration rate (Yamaji and Ma,72 2017), and it is likely that this occurs in dicots as well. The mineral content of seeds is thought 73 to reflect import from the phloem as a result of continued uptake by roots (followed by xylem-to-74 phloem transfer of minerals) as well as redistribution from vegetative tissues such as leaves. 75 Conceivably, many of these steps could be impacted by eCO2. Furthermore, dilution by 76 enhanced carbohydrate production at eCO2 or dilution by increased seed production could 77 contribute to the reduced mineral concentrations observed. Dilution by enhanced carbohydrate 78 production cannot explain all of the effect of eCO2 (McGrath and Lobell, 2013), but the impact 79 of changes in seed yield on the content of minerals on a plant basis (Singh et al., 2016) have not 80 been broadly considered. However, it is reasonable to assume that if a micronutrient is obtained 81 in controlled or limited amounts it would be diluted when distributed amongst a greater number 82 of seeds produced at eCO2, such that the concentration would decrease while the mineral seed 83 content per plant would not change. 84 Another environmental factor that needs to be considered along with eCO2 is elevated 85 temperature. It is generally recognized that an increase in atmospheric CO2, coupled with 86 atmospheric accumulation of other greenhouse gases, will be accompanied by an increase in 87 terrestrial surface air temperatures between 1 to 6 °C by 2050, relative to 1961-1990, depending 88 on geographic location (Rowlands et al., 2012). The impact of elevated temperature would be 89 expected to have a somewhat greater effect on reproductive development, as soybean has an 90 optimum temperature for vegetative growth of ~30 °C (Hesketh et al., 1973) whereas 91 reproductive development has an optimal temperature of 22 to 24 °C (Hatfield et al., 2011). 92 Furthermore, an increase in temperature will generally lead to an increase in vapor pressure 93 deficit (VPD) that would increase transpiration (assuming water supply is sufficient) and as a 94 result the uptake of minerals driven by mass flow. Thus, elevated temperature could impact the 95 concentration of minerals in plant tissues including seeds and in an opposite manner to eCO2. 96 Because future climate change is projected to involve both rising CO2 and temperature it is 97 important to evaluate both environmental factors together. Recently, the Temperature by Free-98 Air CO2 Enrichment (T-FACE) facility was used to explore the independent and interactive 99 effects of eCO2 and elevated temperature on soybean photosynthesis and productivity (Ruiz-100 Vera et al., 2013, Köhler et al., 2016). In general, photosynthesis, aboveground biomass, and 101 seed yields were increased by eCO2 and decreased by elevated temperature; the combination of 102 eCO2 plus elevated temperature had a somewhat variable effect (likely dependent on ambient 103 temperature during the growing season) but the generalized conclusion was that eCO2 attenuated 104 the negative impact of elevated temperature (Ruiz-Vera et al., 2013). However, in terms of seed 105 mineral concentrations, there is little information about the impact of eCO2 in combination with 106 elevated temperature. Consequently, the overall objective of the present study was to determine 107 the individual and combined effect of eCO2 and elevated temperature on soybean seed 108 composition in terms of storage products (protein and oil) and important minerals. Furthermore, 109 we examined seed produced at different positions along the main stem because seed composition 110 varies with nodal position (Huber et al., 2016). Overall, the results suggest that eCO2 in 111 combination with elevated temperature may reduce seed yield relative to eCO2 conditions but 112 will safeguard the mineral nutrition quality of soybeans.  (Table 1). 120 In both years eCO2 significantly increased total seed yield while elevated temperature 121 significantly decreased total seed yield in the bulk samples collected from the 3.   seed yield in g m -2 as derived from sampling of 3.2m rows. Ac, ambient CO2, control temperature; Ah, ambient CO2, heated + 3.5 °C; Ec, elevated CO2, control temperature; Eh, elevated CO2, heated + 3.5 °C) in 2014 and 2015. Boxplots display the median, and 25 and 75% percentiles, and whiskers extend to 1.5X interquartile range, with outliers (if any) shown. increased seed yield at the top canopy position under both ambient and eCO2 concentrations in 136 both seasons ( Figure 3b, Table 2). 137 Although there was some variation in single seed weight from the samples collected at 138 different canopy positions in the present study ( Figure S1 and Köhler et al., 2016), these 139 differences were generally small. Across all treatments, comparison of single seed weight with 140 seed yield at the corresponding canopy position (i.e., bottom, middle or top position) produced a 141 positive correlation but with a very low correlation coefficient (r = 0.28) that was not statistically 142 significant. When different environmental parameters (i.e., elevated temperature or eCO2) or 143 canopy positions were color-coded, it was apparent that the weak correlation was valid across all 144 of the parameters ( Figure S2). The implication is that number of seeds produced rather than 145 changes in seed size primarily drove changes in yield. 146

Seed Protein and Oil Concentrations 147
Canopy position significantly affected seed composition of major storage products, with higher 148 oil concentrations (Figure 4a Conversely, the elevated temperature treatment significantly increased oil concentration at all percentage of total seed under the four treatments. Ac, ambient CO2, control temperature; Ah, ambient CO2, heated + 3.5 °C; Ec, elevated CO2, control temperature; Eh, elevated CO2, heated + 3.5 °C. Boxplots display the median, and 25 and 75% percentiles, and whiskers extend to 1.5X interquartile range, with outliers (if any) shown.         Magnesium concentration [Mg] was not significantly affected by eCO2 (Figure 5b). Seed 189 [Mg] [Mg]. Seed produced at the bottom, middle and top third of the main stem were analyzed from plants grown at eCO2 and elevated temperature. Boxplots display the median, and 25 and 75% percentiles, and whiskers extend to 1.5X interquartile range, with outliers (if any) shown. Ac, ambient CO2, control temperature; Ah, ambient CO2, heated + 3.5 °C; Ec, elevated CO2, control temperature; Eh, elevated CO2, heated + 3.5 °C.

Correlations between seed mineral concentrations and seed yield 234
In order to determine whether changes in seed yield influenced seed mineral concentrations, we 235 compared seed mineral concentrations with seed yields from the corresponding canopy position 236 and results are summarized in Table 6. With a few exceptions where there was a positive 237 correlation (i.e., one year of study each for P and S), in the majority of cases the correlations 238 were not strong and were not statistically significant (p < 0.10). However, two exceptions 239 emerge with seed [B] and [Fe], where significant negative correlations were observed in both 240 years of the study (Table 6 and Figure 9). These results suggest that B, Fe, and perhaps to a 241 lesser extent Mg and Ca, are in limited or fixed supply to the developing seeds, and that changes 242 in concentrations will reflect, in part at least, changes in number of seeds produced in a given 243 environment. 244

245
In the present study, we analyzed the impact of eCO2 and elevated temperature, singly and in 246 combination, on soybean seed yield and seed protein, oil and mineral concentrations. The major 247 conclusions are that eCO2 increased seed yield and reduced the concentrations of a few minerals  those with diffusion to the root as the main mechanism (e.g., K, P, Mn, Zn, Cu and Fe) (Oliveira 296 et al., 2010, Oliver andBarber, 1966). 297 To accumulate in seeds, minerals need to be available in the soil water; presented to the root 298 surface by mass flow or diffusion; taken up into the root; distributed within the plant; and finally 299 partitioned to the developing seeds. Limitations at any of these steps could result in changes in 300 seed mineral concentrations when the number of seeds produced is altered and seed size does not 301 change (as was observed in this study). For example, a mineral present in limiting or fixed 302 amounts may be reduced in concentration in seeds when the number of seeds produced is 303 increased (e.g., at eCO2) and conversely increased in concentration when the number of seeds 304 produced is decreased (e.g., at high temperature). In such a case, seed mineral concentration 305 would be negatively correlated with seed yield, reflecting enrichment when yield was reduced 306 and dilution when growth was increased. Indeed, this was observed for seed [B] and [Fe] in both 307 years of the study, and in one year of study for seed [Ca] and [Mg]. Exactly which step in the 308 process of mineral accumulation in the developing seed is limiting or highly regulated may vary 309 with the mineral, and will be an important focus for future research. Fundamentally, this study 310 suggests that the effects of climate change will be variable on nutrient and caloric levels, and our 311 current knowledge is insufficient to predict the results. 312 The unusual response of B to eCO2 and elevated temperature 313 Boron is an essential micronutrient and while the functions of this micronutrient are not fully 314 understood it is recognized that B may impact various aspects of plant metabolism and transport 315 (Pilbeam and Kirkby, 1983) and cell wall composition (O'Neill et al., 2004). In terms of 316 transport, B is present at neutral pH primarily as boric acid, which can be transported along with 317 the borate anion (Woods, 1996). However, B can be more readily transported as a B-polyol 318 complex (Brown and Shelp, 1997). It is thought that species that produce polyols (e.g., sugar 319 alcohols such as sorbitol and mannitol) may display greater B mobility within the plant, in 320 particular in the phloem, and may explain the large variation in B mobility among species that 321 has been noted. One intriguing possibility is that at elevated temperature, soybean leaves 322 produce more sugar alcohols compared to ambient temperature, which could result in the 323 increased seed [B] observed. This speculative mechanism would also be consistent with 324 increased seed [B] at the top rather than bottom of the canopy (Figure 8). This is clearly an area 325 for further study. 326 confirmed in the present study, and conversely, the increase in seed mineral concentrations by 330 growth at elevated temperature (at ambient CO2) was shown. In general, growth at eCO2 plus 331 elevated temperature restored seed yield and mineral concentrations to nearly those observed at 332 current ambient conditions, calling into question the predictive value of studies looking only at 333 eCO2 to identify threats to human nutrition. The general inverse relationship between mineral 334 concentration and seed yield per plant as affected by eCO2 and elevated temperature, suggests 335 that dilution by increased growth (at eCO2) or enrichment by reduced growth (at elevated 336 temperature) may be general factors impacting seed mineral concentrations. It remains to be 337 determined how broadly applicable our results are outside of the single cultivar of soybeans used 338 with outliers (if any) shown. Ac, ambient CO2, control temperature; Ah, ambient CO2, heated + 3.5 °C; Ec, elevated CO2, control temperature; Eh, elevated CO2, heated + 3.5 °C.

Concluding Remarks
in this experiment. Other cultivars and species may respond differently to either or both stresses, 339 with unknown effects on the nutritional components. We are also unable to predict whether 340 breeding programs conducted in these predicted conditions will be able to ameliorate the loss of 341 nutritional content while maintaining the yield gains. 342 Another noteworthy point is that seed composition of protein, oil and some minerals varies 343 with node position at which the seeds developed, and confirms our earlier study (Huber et al., 344 2016). Moreover, environmental factors that affect composition tend to impact seed produced at 345 all positions within the canopy, even though the effects on yield are not necessarily uniform 346 across the canopy. The largest exception to this generalization was seed [B], which was 347 strikingly increased in seeds at the top of the canopy when plants were grown at elevated 348 temperature. As discussed above, the potential for altered metabolism at elevated temperature 349 resulting in enhanced B mobility within the plant is an attractive hypothesis for further testing. . Using a PID feedback control 368 system we warmed the crop canopy to a target elevation of +3.5 °C above that of the canopy 369 temperature in the unheated sub-plot. The target temperature increase was based on the low-370 response model predictions for surface temperature in the Midwest in 2050 (Rowlands et al., 371 2012). During the day (6 am to 6 pm) and with rainy days excluded, mean temperature 372 differences between the sub-plots were between 0.5 and 1.0 °C lower than the target set point 373 (Table 1), resulting in an average temperature increase of +2.7 °C. During the night the average 374 temperature difference was +3.4 °C. The heated sub-plot diameter was 3.5 m resulting in an 375 effective heated sub-plot area of 9.6 m 2 . Canopy temperature in each sub-plot was measured by 376 infra-red radiometers (SI-111; Apogee Instruments) connected to data-loggers CR1000 377 Micrologger, Campbell Scientific, Logan, UT, USA). Canopy temperature measurements were 378 collected every 5 seconds to control the heater output, averaged every 10 min, and these values 379 were stored. For more information, see Köhler  Höganäs, Sweden), which is a true Near Infrared Transmission instrument that generates a 391 spectrum from 850 to 1050 nm via the monochrome light source and mobile grating system. A 392 50-ml seed sample was used that allowed for 10 subsample readings reported on a 13 % moisture 393 basis. 394 395

Mineral Analysis 396
Six seeds from each sample were selected at random from the seed yield samples for mineral 397 analysis as described in (Ziegler et al., 2013). In brief, single seeds from each main stem canopy 398 position were weighed using a custom-built seed weighing robot and then digested in 399