Determination of the most important characteristics in bread wheat grain yields under salinity stress

Background Salinity has adverse effects on crop production in arid and semi-arid regions but so far, less attention has been paid to this impact on wheat characteristics. A pot experiment was carried out in cage-house to determine the most important characteristics in wheat under wide range of salinity using the stepwise regression analysis. The data collected on ve groups of characteristics (i.e. seedling, phonology, spike, physiology, yield and its contributions). Results The ndings showed that the salinity levels below 3.5 dSm − 1 triggered stimulation in the growth of wheat seedling, while the salinity rates increased by more than 3.5 dSm − 1 signicantly reduced all the seedling characteristics studied. The salinity level of 3.5 dS m − 1 resulted in an earliness percentage of 4.2, 4.9, 3.8 and 2.2% respectively in the developmental stages of booting, heading, anthesis and maturity. Meanwhile, rising salinity levels up to 10.5 dSm − 1 resulted in a 21.7% decrease in emergence speed and a delay of 3.9, 10.8 and 8.5% in booting, heading and anthesis developmental stages respectively. Salinity stress decreased the ag leaf area and increased concentration of chlorophyll pigments, where the percentage increase for chlorophyll a was higher compared with chlorophyll b and total chlorophyll. The stress of salinity decreased all the studied spike characteristics with signicant effect on the number of spike − 1 and spike kernel weights. plant height from 88.27 to 53.59 cm with reduction percent reached 39.3%; root dry weight from 8.79 to 4.07 g with reduction percent reached 53.7%; 100 kernels weight from 4.44 to 3.66 g with reduction percent reached 17.6%; biological yield from 85.47 to 39.93g with reduction present reached 53.3%; grain yield from 37.3 to 15.0 g with reduction reached 59.8%.

wheat grown in arid regions still unknow. Furthermore, stepwise regression analysis could be used justify the superior characteristics to use in selection for high grain yield under normal and stress conditions but has less attention so far. The estimation of correlation and regression analysis among yield and yield components may provide effective selection criteria to improve wheat grain yield and their results observed positive phenotypic correlation for the grain yield. Therefore, the purpose of this study is to determine the most effective grain yield-related characteristics under salinity stress using the stepwise regression analysis.

Study of variance for all characteristics
For all characteristics studied, the study of variance for seedling characteristics (Table 2) showed highly signi cant variations between years, genotypes, salt concentrations and their interactions. The range of years and salinity relative to the other source of variability had the main portion of overall variation. The variation coe cient ranged from 6.1 % for shooting duration to 14.4 % for shooting fresh weight. Highly signi cant differences were reported for all sources of variation with respect to phonological characteristics, and the main portion of total variation was due to variations in years, salinity concentration and years with salt concentration. The coe cient of variation ranged from 1.3% to 5.8% for days to maturity and emergence index, respectively. Highly signi cant differences for all source of variations were observed for physiological characteristics, and the main portion of total variation was due to variance in salinity concentration. The coe cient of variation ranged from 6.3% for ag leaf area to 12.6% for chlorophyll b. Highly signi cant discrepancies were observed for spike characteristics across all sources of variations, and the main portion of total variations is due to years and variance in salinity concentration. The coe cient of variation ranged from 6.3% for spiklets spike -1 to 18.7% for spike kernel weight. Highly signi cant differences were reported for yield and its contributions for all sources of variation, except for years in grain yield. In most cases, the principal portion of overall variation was due to variations in salinity concentration. The coe cient of variation ranged from 3.9% for plant height to 21.4% for grain yield. Table 2 Seedling characteristics The impacts of salt concentrations on seedling, phonological, physiological, spike and yield characteristics and its contributions were described in Table 3. Under 3.5 dS m -1 salinity level, shoot length, shoot fresh weight and shoot dry weight were increased by 7.1, 3.4, and 4.8 %, respectively. These results indicated that the low level of salinity ≤ 3.5 dS m -1 diluted seawater causes stimulation in wheat seedling characteristics especially the shoot length and dry weight.
The salinity level of 7.0 dS m -1 caused the duration of the shoot to increase insigni cantly. Signi cant decrease in regarding shoot fresh and dry weight with reduction percent 20.2 and 9.5, respectively (Table 3). With increasing the salinity level up to 10.5 dS m -1 , the salinity effects were more severe and that re ected on reduction percent values, which reached to 8.6, 37.1 and 28.6 for shoot length, shoot fresh weight and shoot dry weight respectively. The salinity level of 3.5 dS m -1 diluted seawater usually triggered stimulation in the growth of wheat seedlings.

Phonological characteristics
The index of emergence increased by 0.4 % below the amount of salinity of 3.5 dS m -1 . On the other hand, it was substantially reduced below 7.0 dS m -1 and 10.5dS m -1 salinity levels by 14.7 and 21.7 % relative to the control treatment (Table 3). Under control and 10.5 dS m -1 salinity levels, the number of days was increased by 10.2 days (from 73.2 to 83.4) for booting stage; by 9.2 days (from 83.3 to 94.5) for heading stage and by 8.0 days (from 94.4 to 102.4) for anthesis stage, but, the days to maturity did not improve. On the other hand, the degree of salinity of 3.5 dS m -1 has contributed to a decrease of around 3 days for the number of days to boot, heading, anthesis and maturity. In general, the amounts of salinity studied had differing effects on the stages of growth of bread wheat, booting, heading, anthesis and maturity.

Table3
Spike characteristics Increasing salinity levels from control to 10.5dS m -1 caused a decrease in spike length from 11.61 cm to 9.66 cm with reduction percent reached 16.8%; number of spikelet spike -1 from 19.62 to 16.37 cm with reduction percent reached 16.6%; number of kernels spike -1 from 64.07 to 49.91 with reduction percent reached 22.1%; spike kernels weight from 2.79 to 1.80 g with reduction percent reached 35.5% (Table 3). In general, salinity stress led to a reduction of all studied spike characteristics with a strong impact on the number of spike -1 kernels and spike kernel weight.

Yield and its contributing characteristics
The number of tillers pot -1 was increased from 20.3 for control to 23.0 for 10.5 dS m -1 (13.3%). Meanwhile the number of spikes pot -1 was decreased from 17.3 for control to 13.8 for 10.5dSm -1 (20.2%). Since salinity levels were elevated infertile tillers but, fertile ones were reduced. Increasing salinity levels from control to 10.5d Sm -1 caused decrease plant height from 88.27 to 53.59 cm with reduction percent reached 39.3%; root dry weight from 8.79 to 4.07 g with reduction percent reached 53.7%; 100 kernels weight from 4.44 to 3.66 g with reduction percent reached 17.6%; biological yield from 85.47 to 39.93g with reduction present reached 53.3%; grain yield from 37.3 to 15.0 g with reduction reached 59.8%.

Stepwise regression analysis
Stepwise regression analyzes were performed to capture the most critical characteristics contributing to regulated yield of bread wheat grain and three concentrations of salt (Table 4). The analysis under 0.5 dS m -1 salt concentration (control) was over in six steps. Biological yield, spike kernels weight, spike length, plant height, shoot fresh weight and number of spikes pot -1 were remained in the nal model, respectively, (R 2 = 0.72). The formula of the nal model was = 14.21+ 0.34 X1 + 8.91X2 -2.20 X3 -0.26 X4 + 5.93 X5 + 0.41 X6. With respect to the positive and signi cant regression coe cient of biological yield, spike kernels weight, shoot fresh weight and number of spikes pot -1 .It could be indicated that increasing the values of this characteristics would increase the grain yield. Considering the negative and important regression coe cient of spike length and plant height, it could be inferred that the grain yield will be decreased by growing the sum of this trait.
The regression analysis under 3.5 dS m -1 was over in four steps. Biological yield, spike kernels weight, number of spikes pot -1 and plant height were remained in the nal model, respectively, (R 2 = 0.76). The formula of the nal model was = -18.51 + 0.19 X1 + 4.77 X2 + 0.52 X3 + 0.18 X4 Table 4 Regarding the 7.0 dS m -1 salt concentration, the regression analysis was over in only two steps. Biological yield and spike kernels weight were remained in the nal model, respectively, (R 2 = 0.61). The formula of the nal model was = 0.04 + 0.28 X1 + 3.49 X2. Positive and signi cant of the regression coe cient of the two characteristics indicated that increasing the values of these characteristics would increase the grain yield.
The regression analysis under 10.5 dSm -1 was over in ve steps. Biological yield, spike kernels weight, emergence index, number of spikes pot -1 and 100 kernels weight were remained in the nal model, respectively, (R2 = 0.79). The formula of the nal model was = 1.22 +0.18X1 + 2.22 X2 -0.39 x3 + 0.35 X4 + 1.37 X5. Positive and signi cant of the regression coe cient of biological yield, spike kernels weight, number of spikes pot -1 and 100 kernels weight indicated that increasing the values of these characteristics would increase the grain yield.

Discussion
Salinity problem is a major constraint to the global cereal production but breeding for tolerance and exploring the most characteristics of wheat under salinity stress has been slow (Genc et al. 2019). Although rising salinity levels more than 3.5 dS m − 1 for all seedling characteristics studied result in a substantial decrease. Similar ndings indicate that salinity The salinity level of 3.5 dS m − 1 resulted in earliness of 4.2, 4.9, 3.8 and 2.2% in the developmental stages of booting, heading, anthesis and maturity, respectively, while rising salinity levels caused a decrease of 21.7% and lateness in all the developmental stages studied which reached 13.9, 10.8 and 8.5 for booting, heading and anthesis under salinity level 10.5dSm − 1 . Similar ndings were obtained by (Al-Naggar et al. 2015). On the other hand, (Asfaw and Danno 2011) found that increased salinity levels delaying heading and Maturity of tef crop [Eragrostis tef (Zucc.) Trotter] accessions and varieties in Ethiopia. At the reproductive stage, (Allel et al. 2019) evaluated North African barley accessions against salinity and found that the extreme salinity levels of 200 mM NaCl delayed heading and maturity date for most moderately salt tolerant barley genotypes. Additionally, they indicated that longer heading and maturity periods can lead to salt tolerance, and delayed heading and maturity processes provide the opportunity for late differentiation and maturation, allowing the plant to maintain a higher number of spike − 1 kernels and consequently a high grain yield. The salinity levels decreased emergence pace delayed germination and had a minor impact on the percentage of nal germination which could explain the delay of heading and maturity in wheat.
It was also found that, salinity stress induced reduced area of the ag leaf and increased concentration of chlorophyll pigments, with a large percentage increase for chlorophyll a compared to chlorophyll b and total chlorophyll. Salinity stress decreased yield and its contributing characteristics with signi cant effect on plant height, root dry weight, biological and grain yield. This may be attributed to decreasing the biomass and total dry matter with higher salinity levels (STAVRIDOU et al. 2017). Positive and signi cant of the regression coe cient of the import characteristics indicated that increasing the values of these characteristics would increase the grain yield. In general, biological yield, spike kernels weight and number of spike pot − 1 characteristics were import by the stepwise regression under both control and salinity treatments, so these characteristics had an important role for selection criteria of the high grain yield under both normal and salinity stress. These results were in agreement with those found by (Fouad 2018) who used stepwise multiple linear regression analysis and they revealed that number of kernels spike − 1 , number of spike plant − 1 and 100 kernels weight were the most affected characteristics on grain yield under both normal and stress conditions (water regime). (Abd El-Mohsen and Abd El-Sha 2014) reported stepwise multiple linear regression analysis revealed that four traits, i.e., the number of tillers plant − 1 , the number of grains spike − 1 and the 1000 grain weight with R 2 = 97.29%, had justi ed the best grain yield prediction model under normal condition.

Conclusions
The salinity level of 3.5 dS m − 1 resulted in earliness of 4.2, 4.9, 3.8 and 2.2% in the booting, heading, anthesis and maturity developmental stages, respectively. Although rising salinity levels > 3.5 dSm-1 resulted in a 21.7% decrease in the rate of emergence and a delay in all the developmental stages studied which reached 13.9, 10.8 and 8.5 for booting, heading and anthesis below the salinity level of 10.5dSm − 1 .. Salinity stress decreased ag leaf area and increased chlorophyll pigments concentration, with high increase percent for chlorophyll a compared with chlorophyll b and the total chlorophyll. The stress of salinity decreased all of the studied spike characteristics with signi cant impact on the number of spike − 1 and spike kernel weights. Salinity stress reduced yield and its contributing characteristics with a strong impact on plant height, dry root weight, biological yield, and yield of grains. The stepwise regression veri ed that under both normal and salinity stress, biological yield, spike kernel weight and number of spikes pot − 1 characteristics had a signi cant role in selection criteria for high grain yield.  (Table 1). Table 1 The pots irrigated every ve days with amounts of 2 liters pot -1 of irrigation solution corresponding to each salinity level, taking into consideration the leaching requirements to avoid salt accumulation. The salt stress treatments were applied from the sowing irrigation. The NPK multi-nutrients fertilizer 20:10:20 were used by 0.5 g pot -1 week -1 dissolved in irrigation solution. Chelating microelements FULV-E (0.6%Zn.0.2% Cu,5,0%Mg, 2.0%B, 5%N,4.0%K 2 o,4%Fe, 1.2 Mn, 8% fulvic acid and 6% citric acid) was sprayed every week with the rate of 3cm L -1 . The plants were protected against fungi diseases using the fungicide CABRIO TM TOP 60% wg with rate 1g/L and against insect damage using the insecticide NASR LATHION / CHEMINOVA 57% with the rat of 5cm L -1 . The eighteen bread wheat genotypes were arranged in randomize complete block design with three replications. Each salt treatment was considered as an independent experiment. After four days from sowing, the emerged seedlings were counted daily and the speed of seedling emergence was estimated by the formula described by the Association of o cial seed analysis (A.O.S.A. 1983) with some modi cations.
The grate value of EI is the high speed of emergence. After twenty days from sowing, the plants were thinned and only ve seedlings were carefully left in each pot to grow until maturity. From the thinned seedlings, ve seedlings were used to measure shoot length (ShL, cm), shoot fresh weigh (ShFW, g) and Shoot dry weigh (ShDW, g). The studied characteristics at adult stage were number of days to booting (DB), number of days to heading (DH), number of days to anthesis (DA), number of days to maturity (DM), plant height (PH, cm), number of tillers pot -1 (TP -1 ), number of spikes pot -1 (SP -1 ), spike length (SL, cm), number of spikelets spike -1 (SS -1 ), spike kernels weight (SKW), number of kernels spike -1 (KS -1 ), hundred kernels weight (100KW), biological yield pot -1 (BY, g) and grain yield pot -1 (GY, g). After heading, ag leaf area was determined following the formula of Carleton and Foote (1965)  Where A xxx.x refers to the absorbance value at the speci c wavelength (xxx.x in nm).
Using ne tap water on 2.5 mm plastic mesh, the root system was extracted from each pot after harvesting and dried at 65 ° C to estimate root dry weight (RDW, g).MSTATC microcomputer software statistically analyzed the collected data, integrating one factor model over the years, and salt treatments. MSTATC microcomputer software statistically analyzed the collected data, integrating one factor model over the years, and salt treatments. Stepwise regression was used to automatically classify the grain yield affecting characteristics under normal and salinity stresses. Multiregression analysis approach between grain yield as dependent variable and the remaining studied features as independent variables made using IMB® SPSS version 25 (2017).