Element content and distribution has limited, tolerance metric dependent, impact on salinity tolerance in cultivated sunflower (Helianthus annuus).

Disruption of ion homeostasis is a major component of salinity stress's effect on crop yield. In cultivated sunflower prior work revealed a negative relationship between vigor and salinity tolerance. Here, we determined the association of elemental content/distribution traits with salinity tolerance, both with and without taking vigor (biomass in control treatment) into account. We grew seedlings of 12 Helianthus annuus genotypes in two treatments (0, 100 mM NaCl). Plants were measured for biomass (+allocation), and element content (Na, P, K, Ca, Mg, S, Fe, B, Mn, Cu, Zn) in leaves (young and mature), stem, and roots. Genotype tolerance was assessed as both proportional decline of biomass and as expectation deviation (deviation from the observed relationship between vigor and proportional decline in biomass). Genotype rankings on these metrics were not the same. Elemental content and allocation/distribution were highly correlated both at the plant and organ level. Suggestive associations between tolerance and elemental traits were fewer and weaker than expected and differed by tolerance metric. Given the highly correlated nature of elemental content, it remains difficult to pinpoint specific traits underpinning tolerance. Results do show that taking vigor into account is important when seeking to determining traits that can be targeted to increase tolerance independent of vigor, and that the multivariate nature of associated traits should additionally be considered.

INTRODUCTION

High soil salinity is a major abiotic stress impacting crop yield worldwide. Due to natural processes or as the result of land clearing, unsustainable irrigation practices, and high evaporation high levels of salt (generally NaCl) are present in the soil (Munns & Gilliham, 2015). In these salinized environments, plants experience the twin stresses of a high soil osmotic potential, limiting water uptake, and a high concentration of Na, leading to ionic stress (Munns, 2002). This ionic stress is caused by plants taking up Na at the expense of essential elements such as K and Ca (Hawkesford et al., 2012). The accumulation of Na to toxic levels, in combination with elemental imbalances within tissues, disrupts cellular, and enzymatic functions (Chapin, 1980; Morton et al., 2018; Negrão, Schmöckel, & Tester, 2017). With 20% of the world's irrigated agricultural land affected by salinity (FAO, 2005), it is imperative that we improve crop salt tolerance to continue feeding a growing global population.

The effect of salt stress on plant elemental status has been a central focus of research. Na plays a central role in salt stress, and lower Na concentrations in shoot tissues have been shown to correlate with improved salt tolerance in crops (Flowers, 2004; Flowers & Yeo, 1995; Munns & Gilliham, 2015). Additionally, given the antagonistic effect of high Na concentrations in the soil on K uptake, maintenance of K levels is correlated with tolerance as well (Akram, Ashraf, & Akram, 2009; Demidchik, 2014; Hawkesford et al., 2012; Temme, Kerr, & Donovan, 2019). However, most of these studies are limited to leaf or shoot versus root comparisons, restricting analyses to correlations between individual elemental concentrations and salt tolerance. By taking a whole plant and multiple tissue approach, we can infer active mechanisms of elemental allocation, such as exclusion or accumulation (Abrahamson & Caswell, 1982; Romero & Maranon, 1996) and relate those mechanism to salt tolerance.

While the role of Na and K during salinity stress has been extensively studied (Munns, Passioura, Colmer, & Byrt, 2020; Shabala, 2017; Shabala & Cuin, 2008; Wu, Zhang, Giraldo, & Shabala, 2018) less focus has been placed on other elements (Broadley et al., 2012; Lewis, 2019). Over the past decade inductively coupled plasma mass spectroscopy (ICP‐MS) has opened up the door for relatively inexpensive analysis of elemental content of plant tissue (Salt, Baxter, & Lahner, 2008), leading to the identification of a host of genes involved in plant elemental content (Whitt et al., 2018). In previous work on cultivated sunflower, a moderately salt‐tolerant crop (Katerji, van Hoorn, Hamdy, & Mastrorilli, 2000), ICP‐MS analysis revealed leaf S content, besides K, to play a role in salinity tolerance (Temme et al., 2019). Given the impacts of salinity stress on plant nutrient stoichiometry, a comprehensive picture of plant elemental content has the potential to shed light on mechanisms of salt sensitivity and tolerance (Negrão et al., 2017).

Stress tolerance can be difficult to quantitate and numerous metrics for tolerance exist (Morton et al., 2018; Zhu, Shabala, Shabala, Fan, & Zhou, 2016). Salinity tolerance is often defined as having a high relative performance, that is, low proportional decline in performance between control and saline conditions, hereafter called proportional‐decline tolerance (Munns & Tester, 2008; Negrão et al., 2017). An ideal crop cultivar would combine this low proportional decline with high vigor that reflects a capacity for strong healthy growth under good conditions. However, studies on sunflower have shown there can be a negative relationship between growth capacity and stress tolerance (Koziol, Rieseberg, Kane, & Bever, 2012; Mayrose, Kane, Mayrose, Dlugosch, & Rieseberg, 2011), and indeed for cultivated sunflower there appears to be a negative relationship between vigor (defined here as biomass in control treatment) and the proportional effect of salt stress. More vigorous genotypes have a greater proportional decline in biomass under salt stress (Temme et al., 2019). Given this negative relationship, selecting for the ideal cultivar with high vigor and high tolerance may be challenging using the metric of proportional decline.

For situations where there is a negative relationship between vigor and tolerance, it can be useful to construct a, we believe novel, tolerance metric that takes the relationship of vigor to proportional decline into account. By scoring genotypes on their deviation from the expected effect of salinity stress based on vigor, defined here as their biomass under control conditions, we can assess expectation‐deviation tolerance. This metric allows us to disentangle traits related to vigor from traits related to tolerance (performing better than expected under stress). Finding traits related to this metric of tolerance could allow us to incorporate them in vigorous genotypes to select for the ideal high‐vigor and high‐tolerance genotype. However, the question remains on how expectation‐deviation tolerance metric relates to the more often used proportional‐decline tolerance metric.

Cultivated sunflower (Helianthus annuus), a major oilseed crop, is a moderately salt‐tolerant species (Katerji et al., 2000), and previous studies have shown there is genotypic variation in response to salt stress and elemental status (Ashraf & Tufail, 1995; Shi & Sheng, 2005; Temme et al., 2019). We ask the following questions in order to determine the effect of taking a known vigor/salinity effect trade‐off into account when determining tolerance and develop a more comprehensive picture of the effects of salt stress on elemental status and tolerance. (a) Do genotypes differ in the effect of salinity on biomass and how do genotypes compare for salinity tolerance measured as either proportional‐reduction tolerance or expectation‐deviation tolerance? (b) Are there genotypic differences in the effect of salinity on whole plant (WP) and organ (leaf, stem, and root) level elemental content? (c) Across genotypes, are WP and organ level elemental content and allocation associated with either metric of salinity tolerance?

MATERIALS AND METHODS

Plant growth

Achenes from 12 genetically diverse lines of H. annuus (Table S1) were germinated at the University of Georgia greenhouses in a 3:1 sand to turface MVP® (Turface Athletics, PROFILE Products, LLC) mixture. Ten days after germination, seedlings were transplanted into 30 cm tall, 5L pots filled with the same growth substrate, supplemented with 15 ml of lime (Austinville Limestone), 15 ml of gypsum (Performance Minerals Corporation), and 39g of 15–9–12 (N‐P‐K) Osmocote Plus blend (Osmocote, The Scotts Company) per pot. Plants were arranged in a split plot design with two treatment ponds per each of four plots. Three replicate plants per genotype were placed in each pond, for a total of 288 individual plants. Shallow treatment ponds enabled the bottom 8–10 cm of the pots to be submerged in the pond solution. Open water surfaces were covered with black plastic to reduce evaporation and algal growth.

After a week of exposure to fresh water to facilitate seedling establishment after transplant, each pond was assigned a salinity treatment of either 0 or 100 mM of NaCl. All ponds were drained and refilled every day until the treatment ponds reached the target concentration (due to dilution in the soil water). Once treatment concentrations were reached after 7 days, ponds were drained and refilled every 3 days to account for evaporation and maintain a stable concentration within the ponds. Salinity concentration was monitored with a conductivity probe (HI 8733, Hanna Instruments Inc.). To homogenize the distribution of salt within the pots and to prevent salt crystallization on the surface of the soil, pots were top‐watered with a ~200 ml of the water from corresponding pond (0 or 100 mM NaCl) every time the water was replaced. After treatment initiation, plants were grown for 18 days and then harvested.

Measurements and sample collection

At harvest, plants were measured for height from the base of the stem to the apical meristem, and stem diameter at the midpoint between cotyledons and first leaf. Plants were then separated into young leaves (the most recent fully expanded leaf and younger leaves above that on the stem, and bud if present), mature leaves (leaves below the most recent fully expanded leaf, senesced leaves, and cotyledons), stem, and roots. Roots were extracted from the substrate by rinsing with fresh water. Biomass samples of these four tissue types were dried at 60°C and weighed for dry mass.

In order to have sufficient tissue for analysis of element content, genotype replicate samples within a pond were pooled for each tissue type, resulting in four biological replicates per tissue type, genotype and treatment. Biomass samples were coarsely ground using a Wiley® Mini Mill (Thomas Scientific) and finely ground with a Qiagen TissueLyser (Qiagen) for leaf tissue, and an 8000M Mixer/Mill® High‐Energy Ball Mill (SPEX) for tougher stem and root tissue. Samples were sent to Midwest Laboratories (Midwest Laboratories) for Inductively Coupled Argon Plasma Optical Emission (ICP) Analysis. Analysis provided the elemental content of B (boron), Ca (calcium), Cu (copper), Fe (iron), K (potassium), Mg (magnesium), Mn (manganese, P (phosphorus), Na (sodium), S (sulfur), and Zn (zinc) for young leaf, mature leaf, stem, and root tissues.

In order to enable comparison of genotypes at the WP level, WP elemental content and a measure of element allocation to each organ relative to a neutral distribution was calculated. The WP elemental content was calculated for each pond by multiplying the elemental content of each tissue sample by the mass of that tissue (as averaged over the plants in that pond), summing all tissue type amounts, and then dividing by the total biomass. Then, to capture differences in the proportion of an element in each organ that takes into consideration mass allocation, we devised a mass allocation based normalization (after Romero & Maranon, 1996). First, the fraction of the element budget allocated to each tissue (e.g., what percentage of total plant Na was present in the roots) was calculated by dividing the tissue type element amount by the WP element amounts. Then, the mass relative allocated amount (MRAA) was calculated as the ratio of the fraction of the element budget in that organ to the mass fraction of that organ (eq 1. A numerical example of this can be found in Supplemental datasheet 1. For MRAA, values above one indicate preferential allocation (more of the element is in that tissue then would be based on neutral distribution) and values below one indicate exclusion (less of the element is in that tissue then would be expected based on neutral distribution. To allow for comparison of ratios above and below one we log2 transformed the ratios so that a halving and doubling would have the same magnitude (−1, +1).

Thus, this MRAA value represents preferential allocation or exclusion of elements from particular tissue types independent of elemental content of the WP. We've provided an interactive data sheet that demonstrates the relationship between mass allocation, organ level elemental content, WP elemental content and MRAA (Supplemental datasheet 1).

Data analysis

Of the 288 plants initially included in the study, 19 died after transplant (mortality did not differ by genotype or treatment) and were excluded from all analyses. For biomass, values for samples lost due to labeling errors (8%) were interpolated from data for other individuals of the same genotype in the same treatment and plot. This allowed us to construct a complete dataset based on the “average” plant per genotype per treatment pond. Total biomass and biomass fractions of tissue types were then calculated based on the pond average of the tissue types. This resulted in four replicates per treatment per genotype.

All analyses were carried out using R v3.5 (R Foundation for Statistical Computing, Vienna, Austria). Genotype means for all traits in control and stressed environments were then calculated using a mixed‐model approach (r package lme4 [Bates, Mächler, Bolker, & Walker, 2015]) with genotype and treatment as fixed factor and pond within plot as random factor. Then from this model estimated marginal means without the random factor were calculated (R package emmeans, Lenth, 2018). Significance of main effects and their interaction was determined using Wald's Chi‐square test on the mixed model (R package car, Fox & Weisberg, 2011). Using the genotype means per treatment we additionally calculated the change in trait value between control and treatment as the difference in means.

Given the likely correlations between elements we determined common axes of variation using principal component analysis. Principal components were calculated for elemental content at the WP level, within each tissue type, and between tissue types at control conditions, salt treatment, and the difference between treatments. Additionally, we calculate the same principal components for MRAA. Differences between tissue types in element content and MRAA were calculated using Hotelling's‐T test on the first two principal components.

Salt tolerance of each genotype was calculated in two ways. First, tolerance was calculated as the proportional change in biomass between both control and salt treatment (taken as the difference between the natural log of total plant biomass in control and salt treatment). Using this metric, genotypes having a smaller negative proportional change have a greater salt tolerance. Then, to follow up on the previous observation that genotypes with high biomass in control conditions have a greater proportional decrease (Temme et al., 2019), a second measure of tolerance was defined as whether a genotype performing better or worse than expected based on size in the control treatment. To quantitate this, we fitted a linear regression through the proportional change in biomass versus the natural log of total biomass at control. Using this metric, genotypes with a more positive residual from this linear fit perform better than expected and have a greater salt tolerance.

To determine the effect of trait variation on variation in tolerance we ran a series of linear regressions between tolerance (both metrics) and the first two principal components of WP and within tissue type element content and element MRAA at control, salt stressed, and the difference between them. For completeness, all individual component traits of the PC analyses were regressed against tolerance as well. Given the issues of multiple comparisons we limited our interpretation of tolerance associations to the element principal components and only show individual elements as suggestive evidence for an effect on tolerance.

All data visualizations were made using ggplot (Wickham, 2009) and ggbiplot (https://github.com/vqv/ggbiplot).

RESULTS

Genotypic differences in salinity tolerance

Genotypes differed in their absolute biomass and response to salinity (Figure 1a, Table 1), However, proportionally (when comparing log transformed values) the difference between genotypes in the effect of salinity was much more nuanced (Figure 1b). Across all genotypes we found no evidence for a differential response to salinity. Proportionally all genotypes had a comparable effect of salinity stress. This conclusion is, however, somewhat simplistic. Although there were no differences among genotypes that rose to the level of significance, there was a significant relationship between vigor (biomass at control conditions) and the proportional effect of salinity (Figure 1c). Genotypes with a higher vigor had a greater proportional decline in biomass due to salinity stress. Thus, we used two different metrics to characterize tolerance. The first metric was the proportional reduction in biomass: hereafter “proportional‐reduction tolerance.” The second metric took this vigor/salinity effect relationship into account and was the deviation (residual) from the expected effect: hereafter “expectation‐deviation tolerance.”

FIGURE 1

Effect of salinity on biomass and tolerance to salinity. (a) Genotype mean biomass at harvest in control and saline conditions (0 and 100 mM NaCl, respectively). (b) Natural log transformed biomass values at control and salt treatment. (c) Relationship between biomass under control conditions and the proportional decline in biomass under salt treatment (ln (biomass in ctr)‐ln (biomass under salt stress)). Genotypes with more biomass under control conditions have a greater proportional decline in biomass under salt treatment (p < .05). Different colors denote different genotypes

TABLE 1

Overview of plant level traits

Trait	Control (0 mM NaCl)	Salt (100 mM NaCl)	T	G	TxG
Whole plant mass (g)	5.15 (2.02–8.96)	2.94 (1.67–6.16)	***	***	***
Young leaf mass (g)	1.26 (0.58–2.59)	0.77 (0.47–1.85)	***	***	ns
Mature leaf mass (g)	0.9 (0.37–1.82)	0.58 (0.22–1.12)	***	***	*
Stem mas (g)	1.82 (0.28–3)	0.95 (0.29–1.92)	***	***	***
Root mass (g)	1 (0.29–2.27)	0.76 (0.39–1.44)	**	***	**
YLMF	0.26 (0.22–0.38)	0.27 (0.22–0.35)	ns	***	ns
MLMF	0.21 (0.17–0.27)	0.19 (0.13–0.24)	ns	ns	ns
LMF	0.46 (0.4–0.64)	0.45 (0.38–0.55)	ns	***	ns
SMF	0.34 (0.11–0.4)	0.28 (0.1–0.37)	***	***	ns
RMF	0.2 (0.12–0.27)	0.23 (0.19–0.36)	***	***	***
Height (cm)	52.4 (11.36–61.32)	36.57 (11.92–43.13)	***	***	***
Stem diameter (mm)	8.1 (6.63–10.08)	6.22 (5.3–8.88)	***	***	ns
SLA (cm2 g−1)	368.46 (327.13–410.37)	326.95 (252.61–355.63)	***	***	**
B (ppm)	7.49 (5.93–8.3)	7.76 (6.84–10.34)	***	***	*
Ca (%)	0.79 (0.65–1.09)	0.7 (0.56–0.98)	***	***	*
Cu (ppm)	2.84 (2.58–3.7)	3.4 (2.82–5.1)	***	***	**
Fe (ppm)	27.48 (18.92–45.39)	30.6 (23.61–54.35)	**	***	*
K (%)	5.93 (5.54–6.82)	4.58 (3.86–5.49)	***	***	***
Mg (%)	0.4 (0.36–0.58)	0.33 (0.26–0.5)	***	***	ns
Mn (ppm)	25.25 (20.05–37.96)	34.83 (31.62–55.88)	**	***	**
Na (%)	0.12 (0.08–0.35)	1.75 (1.21–2.59)	***	***	***
P (%)	0.4 (0.33–0.58)	0.36 (0.31–0.46)	**	***	*
S (%)	0.53 (0.39–0.67)	0.44 (0.34–0.58)	***	***	***
Zn (ppm)	15.46 (7.42–23.73)	15.11 (6.15–26.83)	ns	***	ns

Median and range (in parentheses) of trait values of 12 genotypes. Stars note p‐value significance of Wald’s chi‐square test on treatment (T), genotype (G), and their interaction (TxG).

<.05;

<.01;

<.001.

The extremes in the rankings on both metrics lined up. Genotypes with the highest or lowest proportional‐reduction tolerance tended to also be the genotypes with the highest or lowest expectation‐deviation tolerance. However, genotypes in the middle of the ranking differed substantially between tolerance metrics leading to no significant (rs 0.50, p = .099) relationship between tolerance metric ranks (Figure S1).

Height and stem diameter, traits strongly related to biomass showed comparable results with biomass (Table 1, Figure S2). Allocation of mass to different plant organs (young leaf, mature leaf, stem, and root) showed a mixed response with no effect of treatment on leaf mass fraction (LMF) and no interaction between genotype and treatment on LMF and stem mass fraction. Root mass fraction did differ significantly between genotypes and treatment including a significant interaction. While not a trait of focus we did measure specific leaf area (SLA) which had a significant effect of treatment, genotype and an interaction between them (Table 1, Figure S2).

Whole plant and organ level elements

Depending on element, genotypes differed in their WP elemental content and in the effect of salinity thereon (Table 1, Figure S3). Given the highly correlated nature of elemental content, we sought to capture the variation in elemental content in principal component analysis.

At the WP level principal component analysis revealed that under control and salt treatment over 70% of the variation in elemental content was captured by the first two principal components (Figure 2a,b). Results showed strong, negative, relationships between WP Na and K content and between B and Zn content. Depending on treatment there were strong positive relationships between elements though these shifted between treatments. Additionally, over 50% of the variation in the response of WP elemental content, assessed as the difference between control and salt treatment, was captured in the first two principal components. The positive and negative relationships among shifts in elemental content were similar to those of elemental content itself (Figure 2c).

FIGURE 2

Multivariate view of whole plant elemental content. First and second principal components of whole plant elemental content under (a) control treatment, (b) salt treatment, and (c) the difference between treatments. Different colors denote different genotypes

Within organs, over 60% in the variation in element content and MRAA under saline conditions was captured in the first two principal components (Figures S4 and S5). Contrasting individual plant organs, again, depending on the element, genotypes largely differed in their organ level elemental content (Table 2, Figure 3a‐c) and in the mass relative allocated amount (MRAA; Figure 3d‐f, Table 3) to their organs (Figure S6). Notably, under salt stress, Na content increased the most in stems and roots and the least in leaves. K content showed the opposite pattern (Table 2). In terms of mass relative allocated amount (MRAA), genotypes were capable of keeping the majority of Na constrained to the roots and out of the leaves under control conditions (Table 3, Figure 3e, Figure S3). However, under salt stress the MRAA in leaves became less negative indicating Na inevitably reached these sensitive tissues (Figure 3f). Due to this the level of preferential allocation to the roots also decreased. Surprisingly, MRAA in stems greatly increased, going from nearly neutral to positive showing increased allocation of Na to stems under saline conditions with K showing an opposite pattern (Table 3, Figure S3).

TABLE 2

Organ level elemental content

Element	Treatment	Young leaves	Mature leaves	Stem	Root	T	G	O	GxT	GxO	TxO	GxTxO
B (ppm)	Control	9.31 (7.47–14.88)a	15.53 (12.5–18.35)c	2.47 (2.17–3.12)e	3.04 (2.45–3.33)e	***	***	***	ns	***	***	ns
	Salt	10.9 (8.53–15.58)b	17.68 (14.83–20.03)d	2.8 (2.15–4.67)e	2.86 (2.32–3.1)e
Ca (%)	Control	0.88 (0.74–1.16)a	1.12 (0.76–1.79)c	0.59 (0.52–0.72)d	0.57 (0.4–0.61)e	***	***	***	***	***	***	***
	Salt	0.82 (0.61–0.99)b	1.12 (0.77–1.8)c	0.47 (0.4–0.52)e	0.51 (0.42–0.66)de
Cu (ppm)	Control	2.41 (2.05–2.6)ab	2.03 (1.78–2.52)bc	1.23 (0.65–2.77)d	7.28 (6.2–9.38)e	***	***	***	**	***	ns	*
	Salt	2.81 (2.02–3.5)a	2.72 (1.86–3.4)a	1.26 (0.95–3.17)cd	8.15 (5.92–9.89)f
Fe (ppm)	Control	11.07 (8.9–70.53)a	11.74 (9.57–32.02)a	5.31 (3.1–12.67)a	87.76 (61.95–169.98)b	ns	***	***	**	***	**	***
	Salt	10.27 (6.8–13.35)a	12.42 (7.92–35)a	7.31 (3.83–15.02)a	101.86 (80.33–149.54)b
K (%)	Control	5.32 (5–6.34)a	7.37 (6.17–8.77)b	6.94 (5.45–8.5)c	3.37 (3.02–3.99)e	***	***	***	***	***	***	***
	Salt	5.47 (4.83–6.27)a	7.32 (4.98–9.24)bc	4.62 (3.24–5.02)d	2.05 (1.49–2.51)f
Mg (%)	Control	0.38 (0.31––0.58)ab	0.4 (0.28–0.9)cd	0.53 (0.46–0.66)e	0.18 (0.14–0.2)f	***	***	***	*	***	***	**
	Salt	0.38 (0.28–0.58)ac	0.47 (0.33–0.94)d	0.33 (0.23–0.43)b	0.17 (0.13–0.19)f
Mn (ppm)	Control	29.94 (23.58–41.2)ab	37.4 (25.7–59.25)cd	19.35 (15.35–23.83)f	20.7 (13.15–30.9)f	**	***	***	***	***	***	***
	Salt	39.62 (32.5–61.07)ac	49.4 (37.1–103.07)e	26.79 (22.45–31.73)bdf	26.86 (22.9–38.47)bdf
Na (%)	Control	0.01 (0–0.19)a	0.01 (0–0.02)a	0.12 (0.05–0.79)ab	0.43 (0.28–0.93)b	***	***	***	***	***	***	***
	Salt	0.42 (0.11–1.25)b	1.01 (0.37–1.84)c	2.96 (1.71–5.75)d	2.58 (1.81–2.78)e
P (%)	Control	0.61 (0.5–0.78)a	0.34 (0.28–0.53)c	0.25 (0.2–0.45)d	0.42 (0.31–0.6)e	**	***	***	***	***	***	ns
	Salt	0.52 (0.38–0.57)b	0.29 (0.24–0.35)d	0.24 (0.2–0.38)d	0.4 (0.31–0.51)ce
S (%)	Control	0.64 (0.6–0.77)a	0.73 (0.62–0.94)c	0.26 (0.14–0.39)d	0.51 (0.41–0.69)b	***	***	***	***	***	***	ns
	Salt	0.5 (0.42–0.57)b	0.65 (0.46–0.85)a	0.22 (0.17–0.29)d	0.46 (0.36–0.62)b
Zn (ppm)	Control	5.68 (4.65–8.47)a	18.46 (12.25–33.37)bc	8.77 (2.67–36.42)b	26.16 (5.7–64.67)cd	ns	***	***	ns	***	*	***
	Salt	5.5 (4.54–7.88)a	14.1 (9.17–33.73)b	8.54 (2.04–43.06)b	35.23 (5.63–57.93)d

Letters note significant Tukey post hoc groups based on contrasting all eight organ/treatment groups for each element. Median and range (in parentheses) of trait values of 12 genotypes. Stars note p‐value significance of Wald's Chi‐square test on treatment (T), genotype (G), organ (O) and their interactions: <.1;

<.05;

<.01;

<.001.

FIGURE 3

Multivariate view of tissue level elemental content. First and second principal component of organ level elemental content under (a) control treatment, (b) salt treatment, and (c) the difference between treatments. First and second principal component of organ level mass relative allocated amount (MRAA), a measure of preferential element allocation, under (d) control treatment, (e) salt treatment, and (f) the difference between treatments. Tissues are young leaf (YL), mature leaf (ML), stem (S), and Root (R). Different letters near tissue ellipse denote Hottelings‐T significant differences between organs on the first two PC axes

TABLE 3

Element mass relative allocated amount (MRAA) per organ

Element	Treatment	Young leaves	Mature leaves	Stem	Root	T	G	O	GxT	GxO	TxO	GxTxO
B	Control	0.36 (0.06 to 0.74)a	1.18 (0.74 to 1.26)b	−1.5 (−1.79 to 1.27)c	−1.28 (−1.49 to 0.98)d	ns	*	***	**	***	***	ns
	Salt	0.48 (−0.02 to 0.67)a	1.23 (0.72 to 1.37)b	−1.53 (−1.88 to 1.18)c	−1.48 (−1.75 to 1.27)c
Ca	Control	0.2 (0.01 to 0.43)a	0.52 (0.18 to 0.71)b	−0.36 (−1.01 to 0.09)d	−0.51 (−1.05 to 0.23)e	ns	***	***	ns	***	***	***
	Salt	0.2 (−0.08 to 0.42)a	0.72 (0.41 to 0.88)c	−0.53 (−1.08 to 0.37)e	−0.45 (−0.83 to 0.18)d
Cu	Control	−0.28 (−0.76 to 0.04)a	−0.58 (−0.95 to 0.13)b	−1.24 (−2 to 0.44)c	1.27 (0.98 to 1.51)d	***	***	***	**	***	***	*
	Salt	−0.41 (−1 to 0.06)ab	−0.37 (−0.98 to 0.02)ab	−1.46 (−1.82 to 0.7)c	1.13 (0.82 to 1.32)e
Fe	Control	−1.11 (−1.98 to 0.14)a	−1.03 (−1.87 to 0.5)a	−2.41 (−3.36 to 1.54)c	1.7 (0.82 to 2.11)d	**	***	***	***	***	***	***
	Salt	−1.81 (−2.13 to 1.17)b	−1.11 (−2.22 to 0.79)a	−2.21 (−3.04 to 1.17)c	1.66 (1.18 to 1.85)d
K	Control	−0.15 (−0.31 to 0.09)a	0.31 (0.11 to 0.5)b	0.24 (−0.08 to 0.47)b	−0.81 (−0.96 to 0.64)d	*	***	***		***	***	***
	Salt	0.23 (0.06 to 0.49)b	0.62 (0.29 to 0.76)c	−0.11 (−0.45 to 0.19)a	−1.17 (−1.64 to 0.83)e
Mg	Control	−0.08 (−0.35 to 0.09)a	0.07 (−0.39 to 0.64)c	0.38 (0.18 to 0.75)d	−1.17 (−1.67 to 0.9)e	***	***	***	ns	***	***	***
	Salt	0.22 (0.02 to 0.37)b	0.47 (0.02 to 0.91)d	−0.02 (−0.66 to 0.48)ac	−1.04 (−1.55 to 0.68)f
Mn	Control	0.26 (−0.2 to 0.56)a	0.53 (0.27 to 0.66)b	−0.44 (−0.94 to 0.13)c	−0.38 (−0.99 to 0.05)c	ns	***	***	ns	***	ns	***
	Salt	0.21 (−0.12 to 0.45)a	0.56 (0.07 to 0.87)b	−0.48 (−0.81 to 0.21)c	−0.37 (−0.84 to 0.05)c
Na	Control	−4.09 (−5.56 to 2.61)a	−4.17 (−5.87 to 2.7)a	−0.18 (−1.26 to 0.39)d	1.58 (1.13 to 2.04)f	***	***	***	*	***	***	ns
	Salt	−2.38 (−3.8 to 0.74)b	−1.11 (−2.47 to 0.03)c	0.72 (−0.51 to 1.71)e	0.5 (−0.48 to 1.16)e
P	Control	0.57 (0.42 to 0.77)a	−0.16 (−0.44 to 0.18)c	−0.63 (−0.98 to 0.41)e	0.05 (−0.3 to 0.33)f	ns	*	***	ns	***	***	**
	Salt	0.43 (0.14 to 0.72)b	−0.33 (−0.58 to 0.12)d	−0.63 (−0.74 to 0.19)e	0.15 (−0.09 to 0.31)f
S	Control	0.33 (−0.1 to 0.66)a	0.55 (0.26 to 0.78)c	−1.04 (−1.53 to 0.52)d	−0.06 (−0.19 to 0.37)e	ns	***	***	ns	***	***	*
	Salt	0.16 (−0.17 to 0.54)b	0.55 (0.32 to 0.83)c	−0.99 (−1.28 to 0.57)d	0.01 (−0.15 to 0.31)e
Zn	Control	−1.45 (−1.81 to 0.28)a	0.08 (−0.44 to 1.67)bc	−0.7 (−1.54 to 0.61)d	0.27 (−0.61 to 1.65)bc	ns	ns	***	ns	***	*	***
	Salt	−1.36 (−2.12 to 0.1)a	−0.17 (−1.23 to 1.66)b	−0.75 (−1.38 to 0.66)d	0.97 (−0.91 to 1.63)c

Letters note significant Tukey post hoc groups based on contrasting all eight organ/treatment groups for each element. Median and range (in parentheses) of trait values of 12 genotypes. Stars note p‐value significance of Wald's chi‐square test on treatment (T), genotype (G), organ (O) and their interactions: <0.1,

<.05;

<.01;

<.001.

While genotypes differed in the elemental content and MRAA in a given organ, the difference among organs was far greater than the difference among genotypes. Over 70% of the variation in elemental content between organs could be captured in the first two principal components with organs having a significantly different suite of elemental content. Under benign conditions, genotypes had a greater Na, copper and iron content in roots and more Mn, B, and Ca in leaves (both young and mature; Figure 3a). Under saline conditions, this changed such that stems and then roots had a greater increase in Na content and that leaves had a lower decrease in K content (Figure 3b,c). Separation of tissue types on element MRAA was comparable to element content (Figure 3d–f).

Relationships between mass allocation, elements, and salinity tolerance

We performed a series of linear regressions to determine the association between elemental traits and tolerance metrics. Given the large number of comparisons this brings up the issue of false discoveries. Thus, while we present these results, care should be taken in interpreting causative mechanisms from weakly significant results. To partially combat this here we focus on the relationship between the principal components of elemental content and allocation with tolerance instead of each individual element. It must be noted however that even these relationships do not hold up to stringent multiple comparisons correction and should only be interpreted as indicative of potential associations.

At the WP level, elemental content was not strongly associated with tolerance (Table S2). Only the first principal component of elemental content in control conditions (Figure 2a) was associated with proportional‐reduction tolerance. The elements most strongly tied to this axis were Ca, S, and copper (Table 4). This result likely recapitulates the vigor/effect of salinity relationship since this same PC axis is suggestively correlated with vigor under control conditions (p < .1).

TABLE 4

Element loadings on PCA axes linked with tolerance

Element	Proportional‐reduction‐tolerance			Expectation‐deviation‐tolerance
	WP content PC1 (control)	Stem content PC1 (salt)	M leaf MRAA PC1 (delta)	Root content PC2 (salt)	Root content PC2 (delta)
B	0.19 (9)	0.30 (7)	0.61 (1)	0.29 (3)	0.02 (9)
Ca	0.88 (1)	>0.01 (10)	0.53 (3)	0.01 (10)	0.06 (8)
Cu	0.76 (3)	0.60 (4)	0.43 (5)	0.05 (6)	0.06 (7)
Fe	0.34 (7)	0.01 (9)	>0.01 (11)	0.06 (5)	0.36 (4)
K	0.09 (11)	0.50 (6)	0.02 (9)	0.75 (1)	0.44 (2)
Mg	0.75 (4)	>0.01 (11)	0.17 (8)	0.23 (4)	0.12 (6)
Mn	0.75 (5)	0.78 (1)	0.53 (4)	0.05 (7)	0.02 (10)
Na	0.18 (10)	0.53 (5)	0.40 (6)	0.42 (2)	0.60 (1)
P	0.69 (6)	0.73 (2)	0.55 (2)	0.02 (9)	>0.01 (11)
S	0.81 (2)	0.63 (3)	0.01 (10)	0.05 (8)	0.38 (3)
Zn	0.25 (8)	0.14 (8)	0.19 (7)	>0.01 (11)	0.17 (5)

Fraction of variation in element content or MRAA that loads onto the PCA axes that suggestively (non‐significant after multiple comparison correction) associate with variation in tolerance (Table S5). Top three traits per PC axis highlighted in bold. Rank of fraction explained in parentheses. M leaf, mature leaf; WP, whole plant.

As a contributor to variation in MRAA we related changes in biomass allocation to tolerance as well. Biomass allocation to plant organs was differentially associated with both metrics of tolerance (Table S3). Proportional‐reduction tolerance was negatively correlated with leaf mass fraction at control conditions. Although this simply recapitulates the correlation found in Figure 1c due to the fact that genotypes with more biomass had a lower leaf mass fraction. Expectation‐deviation tolerance was correlated with the change in stem mass fraction (genotypes with a lower reduction in stem mass fraction had a lower than expected decline).

When we related within organ principal components (Figures S4 and S5) to both metrics of tolerance (Table S4) a contrasting picture emerged. Proportional‐reduction tolerance was associated with the first principal component of stem elemental content under saline conditions (Figure 4a), with highest contributions of Mn, P, and S content (Table 4). Proportional‐reduction tolerance was also associated with the second principal component of the change in MRAA of elements to mature leaves (Figure 4b), with highest contributions of B, P, and Ca (Table 4). Expectation‐deviation tolerance was associated with the second principal component of root elemental content under saline conditions (Figure 4c), with highest contributions of K, Na, and B (Table 4). Expectation‐deviation tolerance was associated with the second principal component of the change in root elemental content (Figure 4d), with the highest contributions of Na, K, and S (Table 4).

FIGURE 4

Relationship between “significant” organ level elemental content principal components and tolerance. Proportional‐reduction tolerance and (a) stem element content under salt treatment PC1 and (b) the change in mature leaf element MRAA PC2. Expectation‐deviation tolerance and (c) root element content under salt treatment PC2 and (d) the change in root element content PC2. Panel insets show the PCA graph corresponding to the panel. Note that while all slopes are p < .05, these are non‐significant after multiple comparisons correction (Table S4)

Surprisingly, the elements that contributed the most to the principal components associated with either tolerance metric weren't necessarily the most significantly associated with tolerance when looked at all elements individually. We found putatively significant relationships between element content, MRAA and both metrics of tolerance but the sheer number of multiple comparisons (>100) lead to these being weakly suggestive at best (Figure S7, Table S5).

DISCUSSION

Salinity is a major abiotic stress decreasing crop yield by limiting water uptake and disrupting cell metabolism and ion homeostasis. Mitigating some of the adverse effects requires some combination of energy intensive Na exclusion or sequestration in vacuoles and/or non‐photosynthesizing tissues (Munns et al., 2020; Tyerman et al., 2019). Indeed, in cultivated sunflower prior work has shown that genotypes more vigorous in benign conditions are proportionally more affected by salinity (Temme et al., 2019). High tolerance could come at the cost of vigor under benign conditions (Munns & Gilliham, 2015). However, here we show that tolerance can further be examined by scoring genotypes on their deviation from the proportional decline expected on the basis of vigor. This expectation‐deviation tolerance metric allows us to identify traits associated with genotypes performing better or worse than expected.

Across these 12 cultivated sunflower genotypes, we confirmed the negative relationship of vigor (defined as biomass under benign conditions) and proportional declines in response to salinity as previously found in a partially overlapping set of sunflower genotypes (Temme et al., 2019). It should be noted that while this relationship is significant and explains a moderate portion of the variance in proportional decline (Figure 1c, p < .05, R2 adj 0.35) the magnitude of variation in proportional decline between genotypes is such that it doesn't rise to the level of significance when simply contrasting genotypes (Figure 1b). Moreover, it should be noted that despite this negative relationship between vigor and proportional decline the most vigorous genotypes in control conditions tended to remain the more vigorous under salt treatment. When genotypes were ranked for proportional‐decline tolerance and expectation‐deviation tolerance, the rankings differed and were uncorrelated (Figure S7). Thus, we further explored the link between variation in genotypes elemental traits and the connection with both tolerance metrics.

We found strong correlations among elemental contents at the WP level (Figure 2a,b). Under both benign and stressed conditions, the strong negative correlations between Na and K likely reflect variation in uptake selectivity and the antagonistic nature of Na and K uptake (Blumwald, 2000; Mäser, Gierth, & Schroeder, 2002). Additionally, we found a negative relationship between B and Zn, which has also been noted for maize (Hosseini, Maftoun, Karimian, Ronaghi, & Emam, 2007) and wheat (Singh, Dahiya, & Narwal, 1990). Out of 11 elements, only WP Zn content was unaffected by salinity (Table 1, Figure S1). These results show the strong effects of high soil salinity on plant elemental content beyond those typically measured for Na and K (Mäser et al., 2002; Wu et al., 2018). Prior research has shown distribution of these elements to be element specific and impacted by stress (Conn & Gilliham, 2010). We found clear differentiation among organs and those differences were far greater than differences among genotypes (Figure 3a,b), as found in willow (Agren & Weih, 2012).

In addition to elemental content and distribution, we investigated the effect of salinity on the preferential allocation of elements to tissue types as the mass relative allocated amount (MRAA), that is, the deviation from a neutral distribution across all organs (after Romero & Maranon, 1996). In benign conditions, plants kept Na out of the leaves and kept it in the roots (Table 3, Figure S3). Under salt treatment this changes with less exclusion of Na from the leaves and relatively less preferential allocation to the roots. Interestingly, MRAA of Na to the stem increases (Table 3, Figure S3). These results suggest that mechanisms pulling/keeping Na out of the transpiration stream at the root level (liken to HKT1 transporters Hasegawa, 2013; Munns & Tester, 2008) are overwhelmed but that mechanisms at the stem or leaf level kick into gear. Resulting in Na build up in the stem. Possible mechanisms could include high activity of tonoplast transporters sequestering Na into the vacuole in stem tissue (Bassil, Zhang, Gong, Tajima, & Blumwald, 2019), transporters pulling out Na in the petiole (Jeschke & Pate, 1991), or active phloem loading, moving Na out of the leaf (Zhong, Kaiser, Köhler, Bauer‐Ruckdeschel, & Komor, 1998).

K distribution broadly mirrors that of Na with increased allocation of K to leaves and lower allocation to roots under benign conditions. Under saline conditions, this pattern is magnified with even greater K allocation to leaves and less to roots (Table 3, Figure S6). As at the WP level K content decreases, this heightened differentiation between plant organs showcases the key role of K in leaf processes (Hawkesford et al., 2012). In the stems, the greater accumulation and allocation of Na is not matched by an increase in K MRAA. Since K and Na have antagonistic effects in the cytosol, maintaining a proper balance between these ions is vital (Munns & Tester, 2008). The discrepancy between stem Na and K content and allocation suggests that sequestration in the vacuole to be a possible mechanism of Na accumulation here to maintain cytosolic Na:K ratio's while elevating Na sequestration in the stem tissue (Bassil et al., 2019; Blumwald, 2000; Zeng, Li, Wang, Zhang, & Du, 2018).

Despite the effects of salinity stress on WP elemental content, the mass relative allocated amount (MRAA) was not dramatically affected for elements other than Na or K. Interestingly, Cu, Fe, Na and Zn have a positive MRAA in roots (both under control and stressed conditions), whereas for all other elements it's neutral or negative. This is opposite findings in wheat (Garnett & Graham, 2005) and carrot (Inal, Gunes, Pilbeam, Kadioglu, & Eraslan, 2009). It could be that this reflects ions bound to the root apoplast from the soil solution (Negrão et al., 2017) but given our thorough washing and the lower MRAA of other elements in roots, that seems not to be a major factor. Given sunflower's status as a moderate heavy metal accumulator (Adesodun et al., 2010), this points to active exclusion of these elements from the shoot or a functional role in the root for these elements.

Linking variation in elemental traits to variation in tolerance has some inherent difficulties in our experimental design. Given the large number of traits versus a comparatively small number of genotypes, multiple comparisons corrections are a large dampener on the significance of our findings. Care must be taken in interpreting our findings, as chance playing a role in our findings cannot be ruled out. For future work testing hypotheses generated by our results we strongly recommend a larger number of genotypes to reach greater power to detect links between a continuum of tolerance values and a host of elemental traits.

Bearing in mind multiple comparisons issues on significance, we found that the associations between tolerance and elemental traits differed by tolerance metric. Proportional‐decline tolerance was associated with shifts in mature leaf element allocation and stem elemental content under salt stress. Expectation‐deviation tolerance was associated with root elemental content and it's shifts under salt stress (Figure 4). Top elements loading onto these multivariate axes were Mn and B for proportional‐reduction tolerance and K and Na for expectation‐deviation tolerance (Table 4). Strikingly, these top loading elements were not necessarily the elements that were associated with tolerance when looked at individually. Though fraught with multiple comparisons correction issues (>100 comparisons), when looking at each individual element in each plant part in turn, results support the role of root K content for expectation‐deviation tolerance but also suggest a role for root and plant S as well as leaf and plant Mn content. Leaf S content was previously found to play a role in sunflower salt tolerance (Temme et al., 2019), possibly related to root sulfolipid content (Erdei, Stuiver, & Kuiper, 1980). Maintenance of Mn content is linked to salt tolerance in Brassica (Chakraborty, Sairam, & Bhaduri, 2016), possibly as a component of manganese superoxide dismutase which functions as a reactive oxygen species (ROS) scavenger and the oxidative component of salt stress (Hernández, del Río, & Sevilla, 1994; Wang et al., 2010).

Given the large number of traits increasing the number of genotypes assessed for these elemental traits will shed more light on significant association with salt tolerance. Moreover, given the short duration of this experiment a next step will be to track element status across developmental stage and it is impact on salinity tolerance beyond the seedling stage. Developmental status has a large impact on element distribution in wheat (Weih, Pourazari, & Vico, 2016), barley (Birsin, Adak, Inal, Aksu, & Gunes, 2010). With a negative relationship between plant vigor and proportional‐reduction tolerance, and a limited overlap between traits connected with both metrics of tolerance, these results suggest the possibility for decoupling vigor related traits and tolerance. By decoupling vigor from tolerance, we could select for high yielding genotypes that are able to maintain their growth under saline conditions. A key open question then is whether these tolerance and vigor traits share a common genetic basis or if tolerance related genes could be pyramided into high vigor varieties (Morton et al., 2018).

Here, we found that increased Na sequestration in stems and roots and increased K allocation to leaves appears to be a common response due to salinity. However, given the highly correlated nature of organ elemental content, it is difficult to pinpoint whether variation in these specific traits underpins variation in tolerance. What is clear is that the multivariate nature of tolerance associated traits needs to be taken into account especially when considering elemental content.

AUTHOR CONTRIBUTIONS

VAB, AAT, and LAD designed the experiment. VAB carried out the experiment and performed initial data analyses and wrote the first manuscript draft. AAT carried out further data analyses. AAT and LAD made subsequent manuscript revisions with all authors approving the manuscript for publication.

Supplementary Material

Click here for additional data file.

Supplementary Material

Click here for additional data file.

Supplementary Material

Click here for additional data file.

Supplementary Material

Click here for additional data file.

Supplementary Material

Click here for additional data file.