Flowering is the critical growth stage for adverse effects of salinity on the grain yield of sunflower

Crop sensitivity to root zone salinity can vary over time, which can lead to severe damage when high sensitivity coincides with high soil salinity. The variation in salinity sensitivity of sunflower during its growth cycle is unknown. Two pot experiments were conducted in sand culture with a complete nutrient solution. Solutions were flushed through pots in excess to maintain specified salt concentrations in the soil solution. In Experiment 1, salt-sensitive stages were determined by applying solutions with an electrical conductivity of < 0.7, 2, 4 or 8 dS m− 1 through the vegetative, flowering or grain filling stages. In Experiment 2, the most sensitive stage to root-zone salinity was determined by exposing plants to 10-day periods of salinity (< 0.7, 8, 16 and 24 dS m− 1) overlapping by 4 days starting from 13-leaf to grain filling. In both experiments, decreases in seed yield were associated with exposure of plants to elevated EC during the period before opening of disk flower to ~ 95% anthesis, while grain filling was the least sensitive. The decline in yield was correlated with a decrease in mature seed number. Increasing salinity from 4 to 8 dS m− 1 during flowering increased the area without seeds at the centre of the disk. In experiment 2, measurements of leaf ion concentrations and photosynthesis suggested that Na+ toxicity decreased yield due to decreases in availability of photosynthate to flowers. To maximise sunflower yield in saline soils, it is important to minimise salinity stress from before flower opening to anthesis.


Introduction
In saline agricultural soils, the intensity of salinity in the root zone can vary over time, and has its most adverse effect on crop yield if high salinity coincides with sensitive growth stages.The exact period when crop growth is most adversely affected by salinity is still subject to conjecture.Previous studies on determinate (wheat; Triticum aestivum L. and sorghum; Sorghum bicolor (L.) Moench.) and indeterminate (cowpea; Vigna unguiculata (L.) Walp.) crops reported that the vegetative stage was more sensitive to root zone salinity than the flowering stage (Maas et al. 1986;Maas and Poss 1989a, b).However, in these studies, the vegetative stage also included the period of panicle or spikelet differentiation, hence the exact developmental stage that is most sensitive to saline water irrigation was still open for debate.In these studies, crop growth was divided into intervals for saline irrigation based on three periods of equal numbers of days, and irrigation was gradually increased at the start and then gradually decreased before the end of each interval.By contrast to cowpea, wheat and sorghum, in rice, the critical stage for saline water irrigation was determined to be during the period between the 3rd -leaf (V 3 ) and panicle initiation (R 0 ) stage (Zeng et al. 2001).Other studies on determinate and indeterminate crops have suggested that exposure to salinity, water stress and high evaporation during the young microspore stage can impair pollen development and consequently, decreasing both seed set and yield (Astiz and Hernández 2013;Debaeke et al. 2017;Nguyen and Sutton 2009;Parish et al. 2012;Saini et al. 1984).
We believe that a key to understanding this apparent confusion in the period of greatest sensitivity of crops to salinity may be because of the imprecise definition of the 'vegetative' and 'flowering' stages of crop development in previous experiments although the previous studies did follow well-accepted scales for crop development (Aref and Rad 2012;Nleya et al. 2013;Rao et al. 2008;Zadoks et al. 1974) (discussed in more detail in the Discussion).We suggest that one of the keys to accurately determining the specific phase of greatest sensitivity to salinity is to have clear means of defining the stages of flower development and maturity.In particular, the vegetative stage should exclude the period of spikelet/panicle differentiation which represents an early flowering stage.Zeng et al. (2001) determined the timing of critical growth stages in rice by dividing the growth stages into five equal periods of 20 days based on phenology from unimbibed seed until the late milk stages.
In the present study, two pot experiments were conducted to compare salt-sensitivity between the vegetative, flowering or grain filling stages, and to determine the specific timing of the most critical stage with six 10-day periods of exposure to salinity from the 13-leaf stage (V13) to the grain filling (R7.6) stage.Sunflower was selected as a test species since it is a crop with determinate flowering and no tillers, and the visibility of flowering development made for clear definition of the reproductive growth stages.

Study location, soil and weather conditions
Experiment 1 was conducted in a polythene shade house at Khulna University in Bangladesh in 2019 (see Supplementary Information Fig. S1a).PVC pots (diameter 0.25 m, height 0.75 m) contained washed sand with a field capacity (weight basis) of 16.2%, a bulk density of 1420 kg m -3 and a water holding capacity (at saturation) of 230 mL kg -1 of soil.The air temperature during the day ranged from 21.7 to 37.3 °C (mean = 30.9°C) and the temperature at night ranged from 10.0 to 28.2 °C (mean = 19.8°C) (see Supplementary Information Fig. S2a).The relative humidity varied between 22% and 100% with means during the day and night of 58% and 89% respectively (see Supplementary Information Fig. S2b).Total evaporation varied from 0.4 to 5.1 mm d -1 (mean = 2.9 mm d -1 ) (see Supplementary Information Fig. S2c).
Experiment 2 was conducted with pots in an open location 1 (see Supplementary Information Fig. S1b) at Murdoch University, Perth in 2020-21.Plastic pots (diameter 0.30 m, height 0.27 m) contained 16 kg of air-dried sand with a field capacity (weight basis) of 19.6% and a bulk density of 1600 kg m -3 .The air temperature during the day ranged from 19.0 to 42.4 °C (mean = 29.6 °C); the temperature at night ranged from 6.5 to 24.7 °C (mean = 16.2 °C) (see Supplementary Information Fig. S3a).The relative humidity varied between 9% and 97% with a mean of 32% and 76% during the day and night respectively (see Supplementary Information Fig. S3b).Total evaporation varied from 0.8 to 16.0 mm d -1 (mean = 9.7 mm d -1 ) (see Supplementary Information Fig. S3c).

Experimental design and crop management
In Experiment 1, sunflower (Helianthus annuus L., cv.Hysun 33) was grown in a completely randomized block design with three saline irrigation levels (EC values of 2, 4 and 8 dS m − 1 ) applied at three growth stages; control plants were grown with an EC < 0.7 dS m − 1 throughout (Table 1).There were seven replications.The crop was sown on 17 January 2019.Growth stages were identified according to the method of Schneiter and Miller (1981).The vegetative stage started with the appearance of the first true leaf when the cotyledons were less than 4 cm in length (VE).The flowering stage started at least 7 days before the formation of the terminal bud, when the inflorescence resembled a many-pointed star (R1) (see Supplementary Information Fig. S4b).The grain filling stage started with the completion of 80% of anthesis (R5.8) (see Supplementary Information Fig. S4f).At the vegetative stage, irrigation with saline water began on the 12th day after sowing and continued until day 47 (Table 1).At the flowering stage, irrigation with saline water was imposed from day 48 to day 84 (Table 1).At the grain filling stage, saline irrigation was imposed from day 85 until harvest on day 106 (Table 1).
In Australia, we were not able to access Hysun 33 owing to quarantine issues.We, therefore, used another sunflower cv.AussieClear 20 in Experiment 2. The experiment had a completely randomized block design with four water salinity levels (EC values < 0.7, 8, 16 and 24 dS m − 1 ) applied to plants at six growth stages (Table 2), with three replicates.Seeds were sown on 24 October 2020.To avoid an osmotic shock at 16 and 24 dS m − 1 of saline irrigation, concentrations were increased by 8 dS m − 1 per day until reaching the final concentration.The six 10-day periods of saline irrigation (< 0.7, 8, 16 and 24 dS m − 1 ) were conducted from the end of vegetative to the grain filling stage.Visual images of the stage of growth at the start of each time of salinity application can be seen in the Supplementary Information Fig. S5.The six 10-day periods of saline irrigation, overlapping by 4 days between two successive periods, covered the period from the end of vegetative growth (V13; 13 true leaves) to the end of flowering (up to R7.6).The plants were harvested at physiological maturity between days 89 and 107 when the back of the sunflower head turned brown and the floral bracts were yellow and brown.
In each experiment, three seeds were sown per pot at sowing; these were thinned to one plant per pot after establishment.The plants were irrigated with a complete nutrient solution containing 3.0 mM KNO 3 , 2.5 mM Ca(NO 3 ) 2 , 1.5 mM MgSO 4 , 0.2 mM KH 2 PO 4 , 50 µM Fe-EDTA, 23 µM H 3 BO 3 , 5.0 µM MnSO 4 , 0.40 µM ZnSO 4 , 0.2 µM CuSO 4 and 0.1 µM Na 2 MoO 4 (cf.Maas and Poss 1989b), together with the salt treatments using NaCl and CaCl 2 in a 2:1 molar ratio.Experiment 1 was irrigated three times daily while Experiment 2 was irrigated in two equal splits during the morning and afternoon.All pots were freely drained after each irrigation event, and the salinity (EC) of irrigation water and drainage water were the same.The salt treatment was terminated by irrigating the pots with a non-saline nutrient solution until the salinity of the leachate was similar to that of the non-saline nutrient solution.At harvest, seed yield and yield components (diameter of the flower head, number of mature and immature seeds and 100seed weight) were measured.

Relationship between relative yield and irrigation water salinity (EC) or solute potential (Ψ s )
A nonlinear least-squares inversion curve (Van Genuchten 1983) was used to determine the salt sensitivity at different growth stages induced by EC or Ψ s .This curve was: where, C = EC (or -Ψ s ), C 50 = the EC (or -Ψ s ) corresponding to a 50% yield reduction, ρ = an empirical constant [Exp (sC 50 )] that specifies the steepness of the curve.The curves were fitted to the data in Genstat 18th edition (VSN International).

Leaf photosynthesis and transpiration
In Experiment 2, five of the youngest fully expanded leaves were randomly selected at the end of each timing treatment for measuring stomatal conductance, photosynthesis and transpiration using a portable photosynthesis system (LI-COR) between 11 am and 2 pm on sunny days.The live and dead leaf numbers were recorded after the completion of each timing of salinity and after 9 days of recovery.Plant height and disk diameter were recorded before and at the end of the salinity treatment, and 9 days after salinity removal.Changes in plant height and disk diameter were calculated after 10 days of salinity treatment and after 9 days of salinity recovery.In addition, images were collected of plant phenology at the time of initiation and termination of salinity (or well-defined phenology) for control and salinity-treated plants.Also, symptoms of necrosis of the leaf, stem, disk bract and tip of the disk bract after completion of salinity treatment were observed.Photographic images of abnormal seed formation and seed-set were collected for Experiments 1 and 2 respectively.
In Experiment 2, at the end of each timing treatment, two fully expanded leaves (with no necrotic tissues) from each replication were collected, washed with DI water and blotted with tissue paper.The leaf samples were oven-dried at 60 °C for 48 h or until reaching constant weight.The dried leaves were ground mechanically, and after nitric acid digestion, the Na + and K + concentrations in these leaves were determined using inductively coupled plasma-optical emission spectrometry (model: 700 Series ICP-OES).

Statistical analysis
Two-way analysis of variance (ANOVA) of seed yield and yield components was done to evaluate the effects of salinity and growth stage and their interactions using Jamovi (0.9.2.6) and R software (R Core Team 2013).A correlation matrix and regression analyses of seed yield, yield components, plant growth and crop stress parameters were conducted using Jamovi software and in Microsoft Excel.

Response to timing of salinity (Experiment 1)
Figure 1 shows the average effects of the imposition of salinity at different growth stages on the yield and yield components of sunflower.Exposure of plants to an EC of 2-8 dS m − 1 at flowering was more damaging to seed yield than exposure at the vegetative or grain filling stages (Fig. 1a).Compared to the control (EC < 0.7 dS m − 1 ), the treatments of 2-8 dS m − 1 at flowering, vegetative and grain filling stages decreased seed yields by 48-55%, 12-42% and 24-38%, respectively (Fig. 1a).Increasing salinity from 4 to 8 dS m − 1 during the flowering stage increased the area without seeds at the centre of the disk (see Supplementary Information Fig. S6).A similar sensitivity to salinity at the flowering stage was observed for yield components of mature seed number, 100-seed weight, disk diameter and total seed number (Fig. 1b-e).Compared to the controls, each of these yield components showed similar patterns of variation when exposed to EC values of 2-8 dS m − 1 at the vegetative, flowering and grain filling stages (Fig. 1b-e).The yield component that responded differently was the percentage of immature seed: irrespective of timing, an EC of 8 dS m − 1 decreased the percent of immature seed to 18-26% of control values (Fig. 1f).In addition, all salinity treatments (EC 2-8 dS m − 1 ) at grain filling decreased the percent of immature seed from 100% (control) to 22-47% (Fig. 1f).
The main effects of timing of different crop growth stages and the level of salinity on seed yield and yield components are summarised in the Supplementary Information (Fig. S7).In overview, the data show that exposure to salinity at flowering gave the lowest yield (9 g plant − 1 ) and at grain filling gave the highest yield (13 g plant − 1 ) (Fig. S7a).Similar timing effects were found for mature seed number, 100-seed weight, disk diameter and total seed number (Fig. S7a-c).
Based on Van Genuchten curves fitted to the data, there was a 30% decrease in seed yield (C 70 ) at the vegetative, flowering and grain filling stages at EC values of about 4.3, 2.6 and 5.2 dS m − 1 respectively, equivalent to solute potentials (Ψ s ) of -0.21, -0.13 and -0.25 MPa respectively (Fig. 2a).
Seed yield was strongly correlated (P < 0.001 and Pearson's r = 0.72-0.90)with all measured yield components except per cent immature seed (see Supplementary Information Table S2a).There were highly significant (P < 0.001) positive relationships between seed yield and both mature seed number and 100-seed weight (r 2 values between 0.49 and 0.81) (Fig. 3a  and b).Multiple linear regression showed that the maximum variation in seed yield was accounted for (r 2 = 0.99) by the combination of mature seed number and 100-seed weight (Eq.1): Response to timing of saline irrigation (Experiment 2) The data for yield and yield components in experiment 2 are presented in three forms (Figs.2b, 4 and 5).Data  (Wolf et al. 1985) presented in Fig. 4 show that seed yield decreased at 8-16 dS m − 1 with the progress of time of application up to timings 3-5 (Fig. 4a).At EC values of 8, 16 and 24 dS m − 1 , timings 4, 5 and 3 had the lowest seed yields of 57.3, 20.1 and 1.2 g plant − 1 , respectively (Fig. 4a, and the seed set image in Supplementary Information, Fig. S8d-S8f).These patterns of variation in yield were mirrored by the variation in 4 of the 5 measured yield components (100-seed weight, total seed number, mature seed number and disc diameter); combinations of salinity and time of application that decreased yield also decreased these yield components (Fig. 4b, c, d, and e).The percentage of immature seed showed the inverse effect: combinations of treatment and time that most decreased yield, most increased the percentage of immature seed (Fig. 4f).
Figure 5 presents a statistical summary of the main effects of the timing of salinity application and the level of salinity for experiment 2. In overview, exposure to salinity at timings 4 and 5 gave the lowest yield (46.4 g plant − 1 ) followed by timings 3 and 6 (54 g plant − 1 ); the highest yield occurred with timing 1 (73 g plant − 1 ) (Fig. 5a).Similar timing effects were found for mature seed number, 100-seed weight, disk diameter and total seed number (Fig. 5b-e).
The main points from Fig. 5 are supported by the analysis of non-linear least-square inversion curves (Fig. 2b).These showed that the seed yield responses to salinity for timings 4 and 5 were relatively similar, while the responses for timings 3 and 6 were also similar up to EC 14 dS m − 1 (Fig. 2b).The EC value associated with a 30% decrease in yield (C 70 ) for timings 4 and 5 were 6.5 dS m − 1 (or a Ψ s of -0.31 MPa).Similarly, C 70 values for timing 1, timing 2, timing 3 and timing 6 were 15.0, 9.8, 7.9 and 7.8 dS m − 1 (or Ψ s values of -0.73, -0.47, -0.38 and -0.38 MPa), respectively.Therefore,  2).It can be concluded that the period from before the opening of disk flowers (R3.5) to 95% anthesis (R5.95) was the critical part of flowering when salinity affected seed yield.
Seed yield was strongly correlated (P < 0.001) with all measured yield components (see Supplementary Information Table S2b).The most significant (P < 0.001) positive relationships were found for mature seed number and 100-seed weight with yield, with r 2 values ranging from 0.80 to 0.86 (Fig. 3c and  d).Multiple linear regression of mature seed number and 100-seed weight accounted for 97% of the variation in seed yield (Eq. 2): Leaf photosynthesis, leaf Na + , K + and Na + /K + (Experiment 2) Measurement of photosynthesis showed that this variable decreased with the increase of salinity from EC 8 to 24 dS m − 1 at each timing relative to control (Fig. 6a).In addition, at EC values of 16 and 24 dS m − 1 photosynthesis decreased progressively  with time.Measurements of ion concentrations in the leaves showed that with EC values of 16 and 24 dS m − 1 , there were major increases in Na + and Na + / K + from timing 3 onwards, but the effects at EC 8 dS m − 1 were negligible (Fig. 6b and d).
The effect on photosynthesis of leaf Na + /K + or Na + at timings 4 and 5, relative to the maximum value in control plants is shown in Fig. 7.With timing 4 there were highly significant (P < 0.001 and r 2 = 0.93) negative relationships between photosynthesis and Na + /K + or Na + in leaves (Fig. 7a, b).With timing 5, there were similar negative relationships, but of lower significance (P < 0.01 and r 2 = 0.77) (Fig. 7c and d).
Seed yield in relation to photosynthesis, leaf Na + and Na + /K + (Experiment 2) Figure 8 shows the effects of photosynthesis on relative seed yield and Na + /K + on relative seed yield at timing 4 and 5. Increasing rates of photosynthesis had positive effects on relative seed yield (r 2 values of 0.91 for timing 4 and 0.93 for timing 5).Increasing Na + /K + had negative effects on relative seed yield (r 2 values of 0.78 for timing 4 and 0.82 for timing 5).Based on the lines of best fit, relative seed yields decreased by 50% as the rate of photosynthesis decreased from ~ 20 to 10 µmol m − 2 s − 1 , and as the Na + /K + increased from ~ 0 to 0.2-0.3.Table S3 in the Supplementary Information, shows the correlation matrix between seed yield and stress parameters.

Visual symptoms of salinity (Experiment 2)
The application of saline irrigation water did not affect plant phenology but adversely affected plant growth and the colour and size of the bud and disk.Stunted plant growth (i.e.decreases in plant height, leaf number and leaf area) was a common response to salinity and was mostly observed from timing 1 to 3. Increased necrosis in older leaves, stems and the tips of the bracts were mostly observed from timing 2 to 6.The intensity of these adverse effects was most evident at the highest salinity (see Supplementary Information, Fig. S9).

Discussion
The two experiments conducted here are part of a logical sequence.Using a system that repeatedly flushed sand with excess irrigation to maintain Fig. 6 Changes in photosynthesis, Na + , K + , and Na + /K + due to the interaction effect of timing and irrigation salinity level.a photosynthesis; b Na + ; c K + ; d Na + /K + in Experiment 2. Photosynthesis, Na + and K + are measured at the end of the period of salinization.Values plotted are the mean of 72 replicates with P-values indicated.Here, the crop growth stages at the six timings are defined as Timing 1 = 13-leaf (V13) to small disk (Before R2); Timing 2 = Before R2 to > 2 cm nodal distance (R3), Timing 3 = 0.5-2.0cm nodal distance (R2) to 50% disk opening (R4.5);Timing 4 = Before opening of disk flower (R3.5) to 95% anthesis (R5.95);Timing 5 = 70% anthesis (R5.7) to small yellow colour at the back of the disk (R7); Timing 6 = Wilting of ray flowers (R6) to large yellow colour at the back of the disk (R7.6) stable soil solution EC values, we found that in sunflower, flowering was the most sensitive stage to salinity stress with EC values ranging from 2 to 8 dS m − 1 in the soil solution.The decline in sunflower yield was most closely correlated with a decrease in mature seed number.The second experiment pin-pointed this time of greatest salt sensitivity more accurately to the relatively short period between R3.5 (before the opening of disk flower) and R5.95 (95% anthesis).These findings have particular implications for sunflower growth and irrigation management in environments with variable soil salinity during the growing season.

Critical stages of salinity effect on seed yield
In Experiment 1, the fitting of non-linear least-square inversion curves showed that exposure to an EC of 2.6 dS m − 1 (solute potential Ψ s , -0.13 MPa) at the flowering stage caused a 30% decrease in yield relative to the  (Maas et al. 1986;Maas and Poss 1989a, b) reported the vegetative stage to be the most sensitive to salinity of the irrigation water in the determinate-flowering sorghum and wheat, as well as in the indeterminate-flowering cowpea.These differences in the growth stage associated with the greatest salt sensitivity between the present study and previous studies might relate to the definition of the start and end of the flowering stage.In the present study, sunflower growth stages were identified following Schneiter and Miller (1981) in which the start of flower initiation occurs 7 days before any visible appearance of the flower.Critically, it is during this period that early pollen development occurs.In the study on wheat, the flowering stage was defined as being from the extension of the flag leaf sheath up to anthesis, and in sorghum, the flowering stage was defined as being from the visible flag leaf in a whorl up to 50% anthesis (Maas et al. 1986;Maas and Poss 1989b;Rao et al. 2008;Zadoks et al. 1974).Hence, the vegetative stage in the previous studies with wheat and sorghum would have included the period of spikelet/panicle differentiation (Rao et al. 2008;Zadoks et al. 1974).In the cowpea experiment, the 'flowering stage' was defined as starting with the appearance of the first floral bud, continuing up to pod development (Maas and Poss 1989a;Nleya et al. 2013); in this study also, the early stages of pollen development would have fallen into the 'vegetative' rather than the 'flowering' period.These studies in which critical events associated with flowering were included in the vegetative phase can be compared with a study in rice.Zeng et al. (2001) avoided the above confusion by classifying the vegetative stage as starting from 3-leaf to panicle initiation (R 0 ) while the flowering stage covered from R 0 to early booting (R 2 ).In rice, the critical stage for sensitivity to salinity was determined to be during the period between 3-leaf (V 3 ) and panicle initiation (R 0 ) (Zeng et al. 2001).
Several other studies with rice also point to the sensitivity of the reproductive stages to salinity (Aref and Rad 2012;Castillo et al. 2003;Singh et al. 2021).In an experiment conducted in a phytotron, a 15-day exposure to an elevated salinity of the soil solution (EC 18 dS m − 1 ) at the panicle initiation and flowering stages reduced seed yield by 64 and 32%, respectively, compared to a non-salinized control (Castillo et al. 2003).In a glasshouse trial, in which counts were made of empty and filled grains per panicle, for plants salinized at 6 dS m − 1 , adverse effects were greatest when salt was applied at panicle initiation.Plants salinized at panicle initiation had 42% filled grains per panicle, whereas plants salinized at other stages of growth (tillering, panicle emergence and ripening) had 67-79% filled grains per panicle, and plants grown without salt had 81% filled grains per panicle (Aref and Rad 2012).In a recent review, Singh et al. (2021) concluded that the reproductive stage in rice is most sensitive to Do other studies point to exposure to environmental stress during early flowering as being critical to the adverse effects on yield?In cereals (wheat, rice, barley and sorghum), drought stress during spikelet differentiation affected spikelet number and initiation of flower formation on the spike, and further stress (during the young microspore stage) caused pollen sterility (Dolferus et al. 2011;Ji et al. 2010).The young microspore stage occurs well before anthesis.In the present experiment, we did not conduct histological examinations of sunflower to define when the pollen meiosis stage occurred.However, we assume (based on Schneiter and Miller 1981) that this occurs before the opening of the disk flower; microsporogenesis is very quick and generally takes ~ 1-2 days.There are indications that abiotic stress (e.g.drought and temperature extremes) can affect the natural development and function of the tapetum, the inner layer of cells in the anther closest to the microspore mother cells (Parish et al. 2012).A similar sensitivity of sunflower tapetal development and pollen growth, might be responsible for the decrease in seed number per disk and yield in our study.
In Experiment 2, 10 days of saline irrigation from R3.5 to R5.95 during the flowering stage had more damaging effects on seed set and seed yield than saline irrigation from the end of the vegetative (13-leaf) to early flowering (before R2) stages or between late flowering (100% of anthesis, R6) and when a large portion of the back of the disk turned yellow (R7.6).This suggests that development within the disk from the R3.5 stage to 95% anthesis is highly sensitive to salinity, but the period after anthesis is more tolerant to salinity.Marc and Palmer (1981) outline five developmental stages of sunflower flowers (FS6-FS10) that correspond with the stages of R3.5-R4 given by Schneiter and Miller (1981).From FS6 to FS10, the development of the floret with bracts, the 5-lobed corolla and hairy floret bracts at the centre of the receptacle radius occurs and can be visible after cutting the disk.However, at R5.1 to R5.95, the development of anthers and fertilization occurs.

Putative mechanism of yield decreased at flowering
Our work has demonstrated that the adverse effects of salinity on grain yield are associated with decreased photosynthesis just before anthesis (Fig. 8a).In previous work, water stress-induced osmotic shock affected male gametophyte development in wheat by reducing sugar delivery to reproductive tissue due to a decrease in photosynthesis (Saini 1997).In rice, 3-days of water deficit at Ψ s ≤ -0.50 MPa during early stages of anther development (pollen mother cell to vacuolated microspore stage) reduced seed yield due to reduced pollen fertility (Nguyen and Sutton 2009).A strong correlation between grain set and the viability of young microspores indicated that a 50% reduction in viable pollen reduced grain set by ~ 59% (Nguyen and Sutton 2009).In wheat, the effect of heat stress (30 °C for 3 days), and water stress (leaf water potential -2.54 MPa) at the meiosis stage caused pollen abortion or total sterility due to failure of tapetal development and non-completion of the first mitosis of microsporogenesis (Saini et al. 1984).In our study, there were higher Na + concentrations in the leaf tissue at high salinities (24 dS m − 1 and from timing 3 onwards at 16 dS m − 1 ) and photosynthesis decreased with the increase in salinity (Fig. 7a and  b).Our results are consistent with the view that yield decreased at high salinities due to the changes in the availability of photosynthates to flowers correlated with Na + toxicity.In other words, the sterility effect in salinity-affected sunflower that depressed seed number was caused by a lack of carbohydrate supply during pollen development.Support for this hypothesis could be gained by an experiment that supplies sugar to florets during exposure to root zone salinity whilst also measuring ion, sugar and carbohydrate concentrations in the affected florets.

Further considerations
Our experiments were conducted at different places and with different varieties but the results were consistent.The present experiments were done in open conditions with contrasting day and night environments and further variation during treatment periods (see Supplementary Information Fig S1 and S2,and Table S1a and S1b) but the other studies discussed above were all done in controlled environments.In our work, there may have been other effects like high temperature in conjunction with salinity that contributed to yield reduction.It might therefore be valuable as a confirmation of the present conclusions to grow sunflower and contrast it with cereal in the Vol:.( 1234567890) same experimental setup, treating them at welldefined stages of reproductive development, with the collection of pollen viability data, to establish whether the critical timing of pollen development for root-zone salinity stress is consistent between species.
In the present study, we did not measure the activity of plant hormones in Experiment 2. It is possible that the effects of salinity on photosynthesis described here could have been mediated by such responses.Further investigation of these effects could be worthwhile.
In future studies, it might also be valuable to directly confirm the effects of the timing of root-zone salinity on pollen fertility/infertility.Future research involving the transfer of pollen from salt-affected to salt-unaffected disc florets (and vice versa) might also help in more closely defining the extent to which pollen development as opposed to corolla fertility is directly impaired by salinity.

Conclusions
Continuous root-zone salinity during the flowering stage caused the greatest decrease in sunflower seed yield relative to exposure to salinity at the vegetative or grain filling stages.An EC of 2-8 dS m − 1 at the vegetative, flowering and grain filling stages gave an average of 32, 52 and 30% reductions in seed yield, respectively, compared to the control.Relative seed yield decreased by 30% (C 70 ) with EC of 4.3, 2.6 and 5.2 dS m − 1 during root-zone salinization at the vegetative, flowering and grain filling stages.Within the reproductive growth stage, saline irrigation (8 and 16 dS m − 1 ) from the R3.5 to R5.95 growth stages was most harmful based on sunflower yield decrease and lower EC values associated with 30% decrease in relative yield.These results suggest that saline irrigation from R3.5 to R5.95 can decrease sunflower yield more severely than the same quality of irrigation water at preceding and following growth stages.In situations where there is insufficient fresh irrigation water to complete the lifecycle of sunflower, it is suggested that non-saline water could be used at the flowering stage from R3.5 to R5.95 while slightly to moderately saline water could be used at other crop growth stages with less serious adverse impacts on seed yield.

Fig. 1
Fig. 1 Effects of salinity (2, 4 and 8 dS m -1 ) at the different growth stages (vegetative, flowering, grain filling) on yield and yield components in Experiment 1. a seed yield; b mature seed Fig. 2 Relative seed yields of sunflower as a function of EC. a crop growth stages in Experiment 1; b six timings in Experiment 2. The fitted non-linear least-square inversion curves show the EC values (extrapolated from the measured value) associated with a decrease in relative yield.The dotted lines

Fig. 3
Fig. 3 Relationship between sunflower seed yield and yield attributes for Experiment 1 (a & b) and 2 (c & d). a & c mature seed number; b & d 100-seed weight.The points have been fitted to linear lines of best fit with P-and r 2 values indicated Seed yield = −37.5 + 0.044 × mature seed number + 9.65 × 100 − seed weight

Fig. 4 Fig. 5
Fig. 4 Changes in yield and yield attributes due to the interaction effect of timing and irrigation salinity level.a seed yield; b mature seed number; c 100-seed weight; d disk diameter; e total seed number; f % immature seed in Experiment 2. Values plotted are the mean of 72 replicates with P-values indicated.Here, the crop growth stages at the six timings are defined as Timing 1 = 13-leaf (V13) to small disk (Before R2); Timing 2

Fig. 7
Fig. 7 Relationship between percent photosynthesis relative to maximum and leaf Na + /K + or Na + concentration for the individual effect of timing 4 (a & b) and timing 5 (c & d).See Fig. 6 for a definition of Timing 4 and 5. Values plotted are for

Fig. 8
Fig. 8 Relationship between sunflower relative seed yield (Y r ) and leaf photosynthesis (a) or leaf Na + /K + (b) with timing 4 (white filled symbols fitted in broken line) and timing 5 (black filled symbols fitted in solid line).See Fig. 6 for a definition of Timing 4 and 5. Leaf photosynthesis, Na + and K + concentra-

Table 2
Schneiter and Miller (1981)tion treatments in Experiment 2 based on phenology according toSchneiter and Miller (1981)a a Thermal time was calculated in the same way as in Table1