Variation in sweetcorn kernel Zn concentration a reflection of source-sink dynamics influenced by kernel number

Grain yield and mineral nutrient concentration in cereal crops are usually inversely correlated, undermining biofortification efforts. Sink size, expressed as kernel number per cob, was manipulated by controlling the time when the silks of sweetcorn (Zea mays) cv. Hybrix 5 and var. HiZeax 103146 were exposed to pollen. Twelve other varieties were manually pollinated to achieve maximum potential kernel number per cob, and kernel Zn concentration was correlated with kernel number and kernel mass. As kernel number increased, kernel Zn concentration decreased, with that decrease occurring to similar extents in both the embryo tissue and rest of the kernel. However, total kernel Zn accumulated per cob increased with increasing kernel number, as the small decreases in individual kernel Zn concentration were more than offset by increases in kernel number. When both kernel number and mass were considered, 90% of the variation in kernel Zn concentration was accounted for. Differential responses in assimilate and Zn distribution to sweetcorn cobs led to significant decreases in kernel Zn concentration with increasing kernel number. This suggests there will be challenges to achieving high kernel Zn concentrations in modern high-yielding sweetcorn varieties unless genotypes with higher Zn translocation rates into kernels can be identified.


Introduction
1 Micronutrient deficiency is a severe nutritional problem affecting approximately 30% of the world's 2 population, mainly in developing countries (Kennedy et al., 2003). One approach to addressing this 3 malnutrition issue has been through agronomic and genetic biofortification that aims to increase the target 4 micronutrient concentration in the edible fraction of staple crops. Examples of species where intensive 5 biofortification efforts have been made include rice (Oryza sativa), wheat (Triticum aestivum) and maize 6 (Zea mays) (White and Broadley, 2009). Despite these efforts, achieving the combination of high yield and 7 high micronutrient concentration has proved challenging, as crop yield is usually inversely correlated to 8 mineral nutrient concentration (Davis et al., 2004;Murphy et al., 2008). For example, a dilution in N, 9 protein and oil concentration of shoot and kernels has been observed with increasing maize shoot biomass 10 and kernel yield (Scott et al., 2006;Riedell, 2010;Abdala et al., 2018). Consequently, an assessment of 11 kernel micronutrient concentration without consideration of yield may lead to misleading genotypic 12 selections that do not provide successful biofortification outcomes. 13 An increase in grain crop yield may be due to an increase in kernel number or individual kernel mass 14 (i.e. larger kernels) or both, with each parameter influenced by genetic and environmental factors (Sadras,15 2007). For modern maize hybrids, yield improvement has mostly been achieved through large increases in 16 kernel number (Fischer and Palmer, 1984), which more than offsets small decreases in individual kernel 17 mass caused by limitations in the availability of assimilates (Echarte et al., 2000). Varietal differences in 18 kernel number and individual kernel mass between genotypes may also contribute to variation in kernel 19 micronutrient concentration (McDonald et al., 2008). However, the evidence for these effects is not 20 conclusive, with other studies finding the correlation between kernel Zn or Fe concentration and grain yield 21 to be weak or non-significant (Ortiz-Monasterio and Graham, 2000;Long et al., 2004;Chakraborti et al., 22 2009). The opportunity for micronutrient biofortification in high yielding maize varieties is therefore not 23 well characterised.

24
Sweetcorn (Zea mays ssp. saccharata), which has a sugary endosperm conferred by a single gene 25 mutation (Creech, 1965;Simonne et al., 1999), is a close relative of maize. The micronutrients of 26 importance to human health such as Zn are found in high concentrations in the scutellum of sweetcorn and 27 maize embryos (Cheah et al., 2019b). In dry milling of maize, this micronutrient-rich embryo is typically 28 removed during maize kernel processing, whereas in sweetcorn the entire kernel is consumed. This potential 29 dietary advantage in sweetcorn is negated by the fact that most of the embryo Zn is in the form of Zn-30 phytate, which is not bioavailable to humans (Cheah et al., 2019a), although Zn accumulated in the 31 endosperm has been shown to be mostly bioavailable. Therefore, to achieve a beneficial outcome for human 32 health, genotypes that accumulate Zn in the endosperm are preferable to those that accumulate Zn in the 33 embryo. Kernel number of the commercial variety cv. Hybrix 5 was manipulated by controlling the duration 25 of silks exposure to pollen. Immediately prior to silk emergence, the sweetcorn ears of all plants were 26 covered with a plastic bag. A visual assessment of peak pollen production (approximately two weeks after 27 ear bagging) was used as a basis for treatment implementation. At this time bags were removed to expose 28 the emerged silks to pollen for 0.5, 1, 2, 4, 8, 12 or 24 hours and subsequently covered with a paper bag to 29 prevent further pollination. Ears were subsequently harvested at 18, 21, 24 and 28 days after pollination 30 (DAP). The 21 and 24 DAP harvest times were equivalent to normal commercial sweetcorn harvest, with the 31 kernels at 18 DAP being slightly immature and at 28 DAP being over-mature.
A c c e p t e d M a n u s c r i p t 7 At harvest, cobs were categorised as well pollinated (> 350 kernels), moderately pollinated (150-350 1 kernels) or poorly pollinated (< 150 kernels). Five cobs were sampled from each category and 10 kernels 2 were extracted from each cob. The kernels were dissected into embryo tissue and rest of the kernel 3 (including the endosperm, aleurone and pericarp tissues), which were measured separately for tissue dry 4 mass (DM) and Zn concentration. Kernel Zn content was then calculated as the sum of the products of each 5 tissue Zn concentration (i.e. embryo or rest of the kernel) and DM content. The rates of kernel Zn and kernel 6 dry matter accumulation were calculated using the formula: where x start and x end represent the Zn or dry matter content at the start and end of a period, n 9 represents the number of kernels, and d represents the number of days within that period.  However, cobs with less than 10 kernels were excluded from the analysis due to an insufficient kernel mass 18 for Zn determination. Total kernel Zn mass per cob was calculated using total kernel dry weight (g per cob) 19 and Zn concentration (mg kg -1 ).  (Table 3) were manually self-pollinated to achieve maximum potential 24 kernel number. These varieties were harvested at the sweetcorn eating stage of 21 DAP and kernel number 25 per cob was determined. Ten kernels were extracted from each cob and DM and mineral nutrient   The accumulation of Zn and dry matter in individual kernels in cobs of cv. Hybrix 5 with varying 10 kernel numbers at different maturity stages (18-28 DAP) are shown in Fig. 1 and Table 1. The rate of 11 accumulation of both Zn and dry matter was greatest at 21-24 DAP and was either maintained or declined to 12 different extents at 24-28 DAP ( Table 1). As kernel numbers increased, the relative changes in dry matter 13 and Zn content were different. Specifically, the average Zn accumulation rate from 18 to 28 DAP was 0.7 µg 14 Zn kernel -1 day -1 in poorly pollinated cobs (< 150 kernels), 0.5 µg kernel -1 day -1 in moderately pollinated 15 cobs (150-350 kernels) and 0.3 µg kernel -1 day -1 in well-pollinated cobs (> 350 kernels). Kernels in well-16 pollinated cobs therefore accumulated Zn at ca. 40% of the rate in kernels on poorly-pollinated cobs. In 17 contrast, the average kernel dry matter accumulation rate for the same cob classifications decreased from 18 17.1 mg kernel -1 day -1 with low kernel numbers to 14.8 and 12.2 mg kernel -1 day -1 in cobs with intermediate 19 and high kernel numbers, respectively. Kernels in well-pollinated cobs therefore accumulated assimilate at 20 ca. 70% of the rate in kernels on poorly-pollinated cobsa much smaller reduction in accumulation than 21 recorded for kernel Zn.

Partitioning of Zn between embryo and rest of the kernel 24
Given the marked difference in the relative accumulation of Zn and assimilate in kernels on cobs 25 with different kernel numbers, the distribution of Zn between the embryo and rest of the kernel was also 26 examined using cobs of cv. Hybrix 5 with varying kernel numbers. The Zn concentration in both embryo 27 tissue and rest of the kernel decreased by ca. 18-33% as kernels matured over the period from 18-28 DAP 28 (Table 2). Despite the decrease in tissue Zn concentration, the Zn content of both constituents increased over 29 the 18-28 DAP period as tissue DM increased (Fig. 2). The rest of the kernel constituted a much larger 30 proportion of the kernel dry matter at all stages of kernel development (viz. 95% at 18 DAP and decreasing A c c e p t e d M a n u s c r i p t 9 to 89% at 28 DAP). Hence, despite the much higher Zn concentration in embryo tissue (Table 2), the rest of 1 the kernel constituted the major proportion of the whole kernel Zn content at all maturity stages (Fig. 2). The 2 proportion of the kernel Zn content in the embryo doubled over the sampling period, from 15-21% at 18 3 DAP to 32-36% at 28 DAP, reflecting relatively greater increases in embryo mass over the period. It is 4 worth noting that kernel number had no real effect on the ratio of Zn content in the embryo and rest of the 5 kernel at any stage of kernel development. This similarity in response in both tissues indicated that in cobs 6 with poor kernel establishment, the additional Zn available to each kernel was not preferentially stored in 7 either tissue but distributed at similar proportions across both (Fig. 2). increasing kernel number appeared to be slightly greater for var. HiZeax 103146, the difference was not 15 significant (p = 0.261). There appeared to be a relatively sharp increase in kernel Zn concentrations at very 16 low kernel numbers (< ca. 50 kernels) in cv. Hybrix 5, and although there was only one cob with kernel 17 numbers that low in var. HiZeax 103146, that sample also showed a similarly large increase in kernel Zn 18 concentration. A significant varietal effect was also observed, with var. HiZeax 103146 being 8.6 ± 1.2 mg 19 kg -1 (p < 0.001) higher in kernel Zn concentration compared with cv. Hybrix 5 at any given kernel number 20 between ca. 50-550 kernels (Fig. 3a).

21
Due to the statistically significant but relatively small decrease in kernel Zn concentrations with 22 increasing kernel number for cobs with > 50 kernels, total kernel Zn mass per cob for both varieties 23 increased with increasing kernel number (Fig. 3b). The overall higher kernel Zn concentration resulted in a  The manually pollinated cobs from the 14 sweetcorn varieties produced different numbers of kernels, 2 with different kernel mass and different kernel Zn concentrations (Table 3). A visual assessment of cobs 3 showed pollination effectiveness (the proportion of total cob length hosting kernels) ranged from 70 to 4 100%. There was a strong correlation between kernel Zn concentration and kernel number (R 2 = 0.57, p = 5 0.002, Fig. 4a) across the genotypes but no correlation existed between kernel mass and Zn concentration 6 (R 2 = 0.05, p = 0.450, Fig. 4b), suggesting that kernel number was the more important factor in determining 7 genotypic differences in kernel Zn concentration. There appeared to be a subset of four genotypes where 8 higher Zn concentrations were recorded than would be expected from the kernel numbers present, with a 9 separate relationship for kernel number and kernel Zn concentration established for this subset (Fig. 4a).

10
These genotypes recorded kernel Zn concentrations that were 11.9 ± 2.9 mg Zn kg -1 higher (p = 0.002) for 11 any given kernel number than the broader population, but unfortunately the combination of low kernel 12 numbers and low kernel DM resulted in two of these genotypes (56.3-1 and 14-6) providing only low-13 moderate total kernel Zn mass per cob (Table 3).
14 The relationship between yield and kernel Zn was explored, using total kernel DM per cob as a 15 surrogate for yield (i.e. the product of kernel number and kernel DM). As expected, there was a negative 16 correlation between total kernel DM and kernel Zn concentration (p = 0.017, Fig. 5a). However, total kernel 17 DM was positively correlated to total kernel Zn mass accumulated per cob (p < 0.001, Fig. 5b), with the 18 variability in this relationship a measure of differences in the capacity of different varieties to translocate Zn 19 into kernels. Multiple linear regression analysis showed that most of the variability in the total kernel Zn 20 mass per cob (R 2 = 90.3, p < 0.001) across these genotypes could be accounted for by differences in kernel 21 number per cob and individual kernel Zn concentration. Different genotypes were therefore able to be 22 ranked in terms of their ability to partition Zn into developing kernels (Table 3).

Effects of variability in yield influenced by source-sink dynamics on kernel Zn concentration 2
The dilution of mineral nutrient concentrations with high biomass production or grain yield is a well-3 reported phenomenon in crops (Jarrell and Beverly, 1981;Davis, 2009;Riedell, 2010). However, there are 4 no studies that have explored the physiological mechanisms underpinning the negative correlation between 5 micronutrient concentrations and grain yield. In this study, we examined the relationship between kernel Zn 6 concentration and the two components of grain yield in sweetcorn, namely kernel number and kernel mass.

7
Both these factors are important for yield improvement, as changes in kernel number may be compensated 8 for by changes in kernel mass and vice versa (Sadras, 2007). 9 We demonstrated that as the sink demand for assimilates increased due to increasing numbers of 10 established kernels, the relative change in rate of accumulation of carbohydrates in kernels was relatively 11 insensitive, compared to that of kernel Zn (Fig. 1, Table 1). The suggestion that assimilates distribution to 12 developing kernels was not strongly source-limited was consistent with the accumulation behaviour of 13 assimilates in maize, where assimilate accumulation was found to be sink-limited in most growing 14 conditions and source-limited only if assimilate availability was reduced during grain filling due to poor 15 growing conditions (Echarte et al., 2000;Borrás et al., 2004). 16 The greater sensitivity of Zn accumulation to increasing sink size (greater kernel numbers) was 17 manifested in declining individual kernel Zn concentration with increasing kernel numbers, and was 18 indicative of source limitations that will constrain Zn accumulation in kernels of sweetcorn. This 19 relationship was explored in two varieties, a commercially cultivated variety cv. Hybrix 5 and an 20 experimental variety previously identified for its higher kernel Zn concentration, var. HiZeax 103146. We 21 found that the same negative correlation between kernel Zn concentration and kernel number existed in both 22 varieties (Fig. 2a), indicating that it is possible that observed differences in kernel Zn concentration between 23 varieties could be strongly influenced by the number of kernels set. Indeed, this negative correlation 24 between kernel Zn concentration and kernel number explained a significant proportion of the variation in 25 kernel Zn concentration across a broader population of 14 genotypes (Fig. 3a).

26
Whilst the relationship between kernel number (one of the main contributors to yield increases in 27 sweetcorn breeding) and kernel Zn concentration was observed to hold across a set of 14 genotypes, the 28 relationship between kernel mass (the other key yield key determinant in sweetcorn) and kernel Zn 29 concentration was weak (Fig. 4b). This suggests that the decreases in kernel Zn concentration that has been 30 observed with increasing sweetcorn yields are likely to be mainly driven by increases in kernel number 31 instead of kernel mass. Kernel mass in maize was reported to be strongly influenced only by reduction in 32 potential assimilate availability leading to reduced kernel mass, but increase in potential assimilate A c c e p t e d M a n u s c r i p t 12 availability did not result in improved kernel mass (Borrás et al., 2004). Conversely, the selection pressure 1 for higher yields in maize, achieved mainly through increased kernel numbers (Fischer and Palmer, 1984), 2 could have unintentionally selected for lower kernel Zn concentration in modern maize hybrids.

3
While individual kernel Zn concentration decreased with increasing yield, the total amount of Zn 4 being translocated into kernels increased with yield, maximising the overall total kernel Zn mass per cob 5 (Fig. 5b). The small decrease in individual kernel Zn concentration associated with increases in kernel 6 number is more than offset by the increase in kernel mass per cob, and suggests that the extent of source 7 limitations constraining kernel Zn concentration are relatively small (Fig. 3b). This implies that while both 8 individual kernel Zn concentration and total kernel yield are both valid selection parameters, the most 9 efficient way of increasing total Zn yield ha -1 in the edible product is by increasing sweetcorn yield, 10 regardless of whether that increase is due to higher kernel numbers, higher kernel DM or both. Future 11 studies could explore whether the negative correlation between kernel number and Zn concentration only 12 applies to a single cob, or to multiple cobs on a plant. If the former, there could therefore be potential to 13 increase kernel Zn concentrations while maintaining or improving yield by selecting for genotypes with 14 multiple, smaller cobs that each support fewer kernels. to markets that consume whole cobs, selection for any combination of increased kernel number and/or 20 kernel mass will increase the total kernel Zn mass per cob, and hence dietary intake. However, in markets 21 where cut cobettes or processed kernels in fixed weight packaging are preferred, the dietary intake will be 22 based on kernel concentration so higher yields will have to be achieved through larger kernel mass to avoid 23 any negative impacts of increased kernel numbers on kernel Zn concentration. Given the dominance of 24 increasing kernel number driving yield increases in modern maize hybrids (Fischer and Palmer, 1984), and 25 the apparently strong environmental impacts on stability of genotypic differences in individual kernel mass, 26 the latter strategy presents significant challenges.

27
On a positive note, var. HiZeax 103146 exhibited higher kernel Zn concentration than cv. Hybrix 5 28 at any given kernel number (Fig. 3a). This suggested that there are genotypes which are more efficient at 29 translocating Zn into cobs for distribution across the developing kernels, a trait independent of kernel 30 number variation, which was also observable in a subset of the 14 genotypes from Experiment 3 (Fig. 4a).

31
Although the actual mechanisms for efficient uptake and accumulation of Zn in these varieties are unclear 32 and warrant further investigation, these traits offer valuable resources to be exploited for genetic A c c e p t e d M a n u s c r i p t 13 biofortification of Zn in sweetcorn. Any genetic advances will need to be supported by appropriate 1 agronomic biofortification strategies to achieve the desired enhancement of Zn partitioning and the 2 concentration of Zn in individual kernels.

4
Accumulation of Zn and assimilates in embryo and rest of the kernel 5 Given the importance of speciation to the bioavailability of Zn for absorption after consumption, the 6 allocation of Zn between embryo and endosperm tissues will be an important factor in securing desired 7 biofortification outcomes. This is important because of the contrasting Zn concentrations ( Table 2) and 8 bioavailability between these kernel constituents, with Zn in embryo tissues stored predominantly as Zn-9 phytate which has low human bioavailability, whereas Zn stored in the endosperm is complexed with N-or 10 S-containing ligands of higher bioavailability (Cheah et al., 2019a). In addition, the stability of that Zn 11 allocation between kernel constituents in response to increasing source limitations caused by increasing 12 kernel number will be particularly important for breeding programs seeking to improve both yield and 13 bioavailable Zn content in sweetcorn.
14 In this study, the trends in Zn concentration and Zn content of both embryo tissue and rest of the 15 kernel as the kernels mature on a well-pollinated cob of sweetcorn were similar to those observed in earlier 16 studies (manuscript in review). We showed that the ratio of embryo Zn content to Zn content in rest of the 17 kernel was constant irrespective of established kernel number, suggesting there was no preferential storage 18 of Zn in either tissue with increasingly constrained Zn supplies. This implies that in order to further improve 19 sweetcorn as a source of dietary Zn, there may be scope change the relative proportions of Zn in different 20 kernel constituents by reducing the mass and/or volume ratio of the embryo relative to the endosperm or rest 21 of the kernel (Zhang et al., 2012;Nagasawa et al., 2013;Chen et al., 2014;Golan et al., 2015;Suzuki et al., 22 2015).

23
A c c e p t e d M a n u s c r i p t 14 Conclusion 1 As the number of established kernels on a sweetcorn cob increased, plants were able to better 2 maintain the rate of assimilate supply to developing kernels than they were the rate of Zn supply. This 3 resulted in a decrease in kernel Zn concentration as kernel number increased. However, increases in kernel 4 mass per cob (the product of kernel number and dry matter content) due to increased kernel number more 5 than offset these small decreases in individual kernel Zn concentrations, such that the total kernel Zn mass 6 per cob was maximised when grain yield was maximised. This result suggests that in the absence of sink 7 size limitations, increasing the supply of Zn into the kernels, either through enhanced genotypic efficiency in 8 translocating Zn or ensuing adequate Zn availability via agronomic means, would be needed to maintain 9 high kernel Zn concentrations. Targeted crosses of high-yielding varieties with efficient kernel Zn 10 accumulation could potentially achieve concomitant improvements in both Zn concentration and yield in 11 modern sweetcorn cultivars.   1: Accumulation of (a) Zn content and (b) dry matter content in individual kernels of sweetcorn (Zea mays) variety cv. Hybrix 5. Kernels were extracted from cobs that supported < 150 kernels (triangles), 150-350 kernels (squares) and > 350 kernels (circles). Samples were harvested at 18, 21, 24 and 28 days after pollination (DAP).      Table 1: Accumulation rate of kernel Zn content and dry matter content in sweetcorn (Zea mays) variety cv. Hybrix 5 between 18-28 days after pollination (DAP) when cobs were well, moderately, or poorly pollinated.