Agronomic Performance and Flowering Behavior in Response to Photoperiod and Vernalization in Barley (Hordeum vulgare L.) Genotypes with Contrasting Drought Tolerance Behavior

In barley, owering behavior is a highly regulated and complex process where the appropriate matching of reproductive development with seasonal variation in water availability confer barley adaptation to different environments. In this study, the role of variation in owering time and drought tolerance in four selected barley genotypes was studied under eld and controlled conditions. For this purpose, eld trials were conducted for two consecutive seasons at three diverse environments where the studied genotypes were subjected to either rainfed conditions or rainfed plus supplementary irrigation under two different sowing dates. Furthermore, reproductive meristem development in two selected barley genotypes, Rum (drought tolerant) and Steptoe (drought-sensitive) was also assessed in response to both vernalization and water stress under two different photoperiod conditions.


Abstract Background
In barley, owering behavior is a highly regulated and complex process where the appropriate matching of reproductive development with seasonal variation in water availability confer barley adaptation to different environments. In this study, the role of variation in owering time and drought tolerance in four selected barley genotypes was studied under eld and controlled conditions. For this purpose, eld trials were conducted for two consecutive seasons at three diverse environments where the studied genotypes were subjected to either rainfed conditions or rainfed plus supplementary irrigation under two different sowing dates. Furthermore, reproductive meristem development in two selected barley genotypes, Rum (drought tolerant) and Steptoe (drought-sensitive) was also assessed in response to both vernalization and water stress under two different photoperiod conditions.

Results
Variation in the number of days to heading was more pronounced under rainfed conditions than under well water conditions. For agronomic performance, Rum was superior under all tested environments, which assure its general adaptability to multiple environments, while Steptoe was the poorest. The transition to reproductive meristem was faster under vernalized long-day conditions as compared to vernalized short-day conditions. The progress of shoot apical meristem development and heading under long-day conditions was signi cantly faster in Rum than that of Steptoe. A clear effect of drought stress was observed on shoot apical meristem development in Steptoe. Under short-day conditions, vernalized Rum plants subjected to water de cit showed an advanced meristem development stage a signi cant earlier HD when compared with non-stressed plants. This early owering behavior in stressed Rum plants under short-day conditions was accompanied by higher gene expression of the Vrn-H1 gene.

Conclusion
In conclusion, the integration of vernalization and photoperiod signals in drought-tolerant barley genotypes is associated with early owering behavior and higher productivity in dry environments.

Background
Barley is considered one of the most important cereal crops in the world in terms of harvested area, trade value and human nutrition [1]. It is commonly used for human consumption, animal feed and malting and it is considered the crop of choice in marginal and dry areas [2]. In eastern parts of the Mediterranean basin, barley is generally grown in arid and semi-arid areas that are characterized by low amounts of precipitation and prolonged drought conditions [3]. As a rainfed crop, barley is frequently suffering from drought that occurs when the soil water content is not available in su cient amounts to support normal growth and development [4]. To maintain a sustainable increase in barley yield, the development of location-speci c and high-yielding varieties with improved tolerance to abiotic stresses is needed [3]. Drought stress has always played an important selective role in the evolution of plant growth, development and physiology [5]. Plant adaptation to dry conditions is considered a key factor that will determine the future of crop production systems in response to climate change and associated conditions [6]. In this perspective, shifting planting dates or switching to short growing-season varieties are considered a useful strategy to reduce the negative impact of climatic change and associated stresses such as drought and heat stress [7]. Therefore, owering time is an important trait for crop adaptation to particular environments and their associated abiotic stresses [8]. The ability to regulate owering time in cereal plants will enable them to complete their life cycle successfully under a wide range of environments [9].
Flowering is a crucial event in the life cycle of seed propagated plants and is usually affected by different environmental stimuli [8]. Timing of owering, also known as heading date in cereals, is a major trait that is linked to the adaptation to particular environments and it determines the crop performance under eld conditions [10,11]. In temperate cereals, photoperiod sensitivity and vernalization requirements are major cues that determine owering behavior [9]. Vernalization accelerates owering by promoting in orescence initiation, the rst step in the transition of the shoot apex to the reproductive phase [12].
Furthermore, vernalization is an adaptation process that prevents the exposure of sensitive oral meristems to freezing winter temperatures [9]. Winter barley requires vernalization and therefore, they are planted in the fall and require long exposures to cold temperatures during winter to accelerate owering [13]. On the other hand, spring barley is usually planted in spring and does not require vernalization to ower [10]. On the other hand, photoperiod is a major factor determining cereal plant owering and crop performance in response to day length [8]. In barley, a long-day (LD) plant, owering occurs when the day becomes longer than a critical day length [14]. Photoperiod insensitive barley enables the initiation of oral primordia without requirements for LD photoperiod, whereas sensitive genotypes need LD conditions for oral primordia initiation [15].
The manipulation of owering time in barley may strongly affect its adaptation to marginal areas and associated stresses [16]. Early genetic analysis indicates the presence of a correlation between growth habit and stress tolerance where spring barley genotypes were less freezing tolerant when compared with winter types [17]. Allelic differences and their combinations in vernalization-(VRN) and photoperiodrelated (PPD) genes seem to have a large effect on stress tolerance in temperate cereals [18,19]. For instance, the PHOTOPERIOD-H1 (Ppd-H1) gene was found to control stress-induced senescence in barley roots subjected to osmotic stress [20] and most recently it was found to control the plasticity of development in response to drought [21]. In another study, barley genotypes carrying spring VERNALIZATION-H1 (Vrn-H1) alleles showed an accelerated owering under drought stress when compared with genotypes carrying winter alleles [22]. Therefore, the interaction between environmental cues, such as day-length and low-temperatures, and the allelic variations in major owering time genes, such as Ppd-H1 and Vrn-H1, might affect the owering behavior and adaptation of barley to dry conditions [10].
The matching of reproductive development and the time of owering to the temporal variation in water availability is recognized as one of the most important traits for plant adaptation to drought [8,10]. In this study, we analyzed the owering behavior, reproductive meristem development, growth and physiological responses to drought conditions in two selected spring barley genotypes with contrasting drought tolerance behavior. For this purpose, the owering behavior and agronomic performance of droughttolerant and drought-sensitive barley genotypes were studied at different contrasting envirnoments and management practices for two growing seasons. Thereafter, the impact of drought stress on two selected barley genotypes with contrasting drought tolerance behavior genotypes was assessed under controlled conditions with different vernalization and photoperiod treatments. The expression patterns of Ppd-H1 and Vrn-H1 genes were investigated in both selected genotypes in response to drought, vernalization and photoperiod treatments. The presented results showed that drought effects varied in response to vernalization and photoperiod treatments in the tested genotypes, highlighting the importance of phenological adjustment in barley for better adaption to dry environments.

Plant material
Four barley genotypes with contrasting drought tolerance behavior were used in this study [22]: Rum cultivar (Harbinger-Arivat × Attiki; released in Jordan in 1986) was developed by CIMMYT/ICARDA and it is known for its stable productivity under arid conditions and is considered a drought-tolerant genotype. Acsad176 (CM872-189-3Y-1B-2Y-1BX1Y-OB) × (Cr.366/16/2)) developed by ACSAD center and commonly grown in semi-arid regions in Jordan where supplementary irrigation is recommended. Steptoe (Washington Selection 3564/Unitan) is derived from coast-type barley originating in North Africa and is considered a drought-sensitive genotype [22]. Morex is a spring smooth-awned spring barley cultivar derived from Manchurian barley (Cree/Bonanza). All selected genotypes are spring types (carrying dominant spring Vrn-H1 alleles) and all carry a functional Ppd-H1 allele conferring photoperiodic responses to LD conditions except for Morex, which carries a nonfunctional ppd-H1 allele. All four genotypes were used for eld trials, while Rum and Steptoe were selected for the controlled conditions experiment.

Field trials
The eld trials were conducted in three agricultural stations across Jordan for two growing seasons 31°16′37.1″ N, 35°44′26.9″ E; elevation: 920 m; average annual precipitation: 337 mm; soil type: clay) representing a semi-dry area. In each trial, the four tested genotypes were sown at two different dates: 1st of December and 1st of February. Furthermore, the genotypes were cultivated in each eld trial under either rainfed or supplementary-irrigation conditions. The weather data and the amount of supplementary irrigation applied are provided in Additional le 1.
A split-split plot design with four replications was used where each replicate contained two plots for irrigation and rainfed treatments and each plot contains two subplots for planting date treatments. Each subplot area contained the four tested genotypes randomly distributed within the plot with each genotype sown in two adjacent rows (1.5 m length and 0.3 m apart). The seeding rates were adjusted according to the results of a seed germination test (based on 90% germination) to obtain a plant density from the viable seeds of 50 plants/row (approximately 150 plants/m 2 ). In irrigated plots, average soil moisture in the rst 60 cm depth was maintained at 80% of eld capacity, which was estimated each week after sowing using soil HH2 moisture meter with Pro le Probe type PR2 (DELTA-T Devices, Cambridge, England). All eld experiments followed a fallow in the crop rotation and were established using hand broadcasting following conventional tillage performed with a chisel plow and disk harrow. In each location, experimental plots were managed following the standard agricultural practices including fertilizer application to match the crop needs, weed control (hand weeding and herbicides against broadleaf weeds) and pesticide use against main pathogens. Weather stations at the eld sites were used to record precipitation and temperature during the growing seasons.
Field trials data analysis During plant growth and before harvest, the following data were collected: days to heading (HD: days) that was determined visually by calculating the number of days from emergence to the day when 50% of plants had spikes; days to physiological maturity (MD: days) that was determined visually by calculating the number of days from emergence to the day when 90% of the plants in a plot have reached the physiological maturity stage (no green tissue remained in 90% of the plants in each plot); Grains lling period (GFP: days) that was estimated by substracting MD from HD; plant height (PH: cm) that was measured for each genotype at maturity from the ground to the spike tip (excluding awns) on three selected plants randomly distributed in the rows; peduncle length (PL: cm) that was measured as the distance from the ag leaf collar to the base of the spike on the same main stem of the plants used to measure plant height and whenever the value is negative, this indicates that the spike remained inside the sheet of the ag leaf. At the end of the experiment, plants for each treatment were harvested and the following measurements were recorded: total plant weight (TW: g/m 2 ) that was measured as the whole dry weight of the harvested plants; spikes number (SN) as the number of spikes counted per m2; spikes weight (SPW: g/m 2 ) that was measured by weighing the spikes obtained from harvested plants; Straw weight (STW: g/m 2 ) that was obtained by subtracting spikes weight from the total plant weight; for grains weight (GW: g/m 2 ), spikes were separated and threshed and the grains were cleaned from the chaff and weighed; grains per spike (G.S) was calculated by dividing grains number on the total spikes number; grains number (GN: grains/m 2 ) was obtained by counting the harvested seeds using a seed counter; thousand kernel weight (TKW: g) was calculated by dividing the number of the grains by grains weight multiplied by 1000; harvest index (HI: %) was calculated by dividing grain weight over total plant weight multiplied by 100.
For statistical analysis, the multi-environment eld trials data were analyzed as a split-split plot design using the GenStat statistical software (Release 16.1, 2013; VSN International L) and a combined analysis of variance (ANOVA) was obtained. For this purpose, each location-growing season combination was considered an environment resulting in six environments: JU2012, JU2013, JUST2012, JUST2013, RB2012 and RB2013. The ANOVA was conducted to analyze the variations due to environments, sowing dates, water regimes, genotypes and all of their interactions. The coe cients of variation (CV%) was determined for each trait and when signi cant differences were detected, a comparison of means was conducted using the least signi cant difference test (LSD, P < 0.05) to determine the performances of tested genotypes under all treatment combinations. Pearson correlation coe cients were calculated in the R software 4.0.2 using the metan package [23] using the means for all or selected combinations of environments, genotypes, water regimes and sowing dates. To analyze the speci c adaptation, superiority, and stability of the genotypes across tested environments for the GW trait, the GGE function [24] of the metan package [23] was used. For this purpose, each environment-sowing date-water regime combinations were considered as one distinctive group.

Controlled conditions experiment
To test the effect of water de cit, vernalization and photoperiod on owering behavior and physiological responses in Rum (drought tolerant) and Steptoe (drought-sensitive), a pot experiment was conducted under controlled growth conditions. For the vernalization treatment, seeds of both genotypes were imbibed with distilled water and planted (2 seeds per pot) in foil-covered pots (30 cm length and 20 cm diameter) containing a mixture of sand and perlite (2:1) and were subjected to 45 days of vernalization at 4 ˚C under complete darkness as described previously [25]. After vernalization, the seedlings were transferred to growth chamber conditions, either under LD (16 h light/8 h dark) or short-day (SD: 8 h light/16 h dark) conditions depending on the photoperiod treatment, along with non-vernalized 7-days old seedlings (germinated under complete darkness). After one week, leaf number was counted in vernalized and non-vernalized seedling and those plants with leaf number similar to that of 7 days-old seedlings were selected for further analysis [25]. Both vernalized and non-vernalized selected seedlings were grown under either SD or LD conditions at 22 ± 1°C with a light intensity level of 56 µmol m − 2 s − 1 and the plants were allowed to grow up to 120 days.
For drought treatments, the seeds of both vernalized and non-vernalized plants were exposed to wellwatered and water de cit conditions under two different photoperiods (LD and SD) as described above.
The treated seedlings (vernalization and photoperiod) of both tested genotypes were subjected to water de cit conditions by withholding irrigation at the start of the experiment while non-stressed seedlings were kept hydrated and were used as control. Water levels were maintained at 75% of pot capacity for non-stressed plants and 15% of pot capacity for stressed plants [26]. The soil moisture levels for the pots of the well-watered and drought-stressed conditions were maintained at the targeted pot capacity by irrigating the required amounts of water after measuring soil water content using a Delta-T soil moisture probe device (Theta Probe type ML2X, Delta-T Devices, Cambridge, England).
Physiological and growth measurements of controlled conditions experiment Leaf relative water content (RWC) was measured biweekly for the rst eight weeks of the experiment as described previously [27] and the RWC was calculated according to [28]. Stomatal resistance was also measured on the most upper fully expanded leaf biweekly for the rst eight weeks of the experiment by using a Porometer device (AP4, Delta-T Devices, Cambridge, UK). For leaf chlorophyll uorescence yield, measurements were recorded directly by using a Pulse-Modulated Fluorometer (OS1-FL modulated chlorophyll uorometer, ADC Bio Scienti c Ltd., Hertford, UK). Fluorescence measurements were taken after 20 min of acclimation to darkness under leaf clips (FL-DC; Opti-Sciences). The maximum quantum e ciency (Fv/Fm) of photosystem II was calculated as Fv/Fm = (Fm-F0)/Fm, where Fm and F0 are the maximal and minimal chlorophyll uorescence measured in darkness-adapted leaves, respectively.
For meristem development, plants were sampled biweekly for the rst eight weeks of the experiment. Shoot apices were dissected under a binocular dissecting microscope (LAS EZ, Leica Microsystems, Switzerland) and digitally photographed by using a digital camera. Phenotypic meristem development was scored according to the Waddington scale [29] using four plants per experimental treatment. The HD was recorded based on the number of days between emergence and heading following Zadoks scale [30]. A comparison between means was conducted using the standard error of means to determine the performances of the two genotypes under all treatment combinations.

Gene expression analysis
For quantitative real-time PCR (qRT-PCR) analysis of treated barley plants (controlled conditions experiment), total RNA was isolated from leaf samples collected after 10 hours from the onset of light conditions irrespective of the photoperiod or vernalization conditions. The leaf samples were collected after two and four weeks for LD treated plants and after four and six weeks for SD treated plants. Speci c primers pairs for two major owering genes in barley (Vrn-H1: fwd: 5 -TG ∀ GCTCAG ∀ ATGGA ⊤ CG -3 and Rev: 5-TATGAGCGCTACTC ⊤ ATGC -3; [31]) and (Ppd-H1: fwd: 5-C ∀ ATC ∀ AGAGCGGCGATG -3 and Rev: 5-TCTGAC ⊤ GGGATGG ⊤ CACA -3; [32]) were used. For stress-inducible gene expression, speci c primers pair for the HVA22 gene [33] were used (fwd: 5-ATGGGC ∀ ATCATGGGCGCT -3 and Rev: 5-TGCAℂ ⊤ GTGATCGGCGTC -3) and HvActin (encoding actin, a house-keeping gene that was used as an internal reference control for relative gene expression analysis as described in [27]. The ampli cation of the targeted genes was carried out using the GoTaq® qPCR Master Mix Kit following the manufacturers' instructions (Promega, Madison, WI, USA), and the real-time detection of the ampli ed products was performed in a Mini-Opticon Real-Time PCR System (BioRad, Hercules, CA, USA). All cDNA samples were analyzed in triplicate, and each replicate was derived from two biological replicates. Thermal cycling conditions consisted of an initial denaturation step of 94°C for 5 minutes followed by 40 cycles of 94°C for 30 s, 55°C for 30 s and 72°C for 45 s, followed with a nal extension at 72°C for 5 min. The relative changes in gene expression were quanti ed as described in [34].

Field trials
Analyzing weather data during the two growing seasons (2012 and 2013) in the three locations indicates that January and February were the coldest months (Additional le 1). During this time, all eld trials were exposed to temperatures less than 5 ℃ indicating that the tested genotypes were exposed to vernalization conditions. For rainfall amounts, RB station received ~ 10% (33.5 mm) less precipitation than the long-term average of rainfall during the rst growing season, while JU and JUST stations received 11.5% (58 mm) and 40% (86.5 mm) more than the long term average of rainfall, respectively (Additional le 1). During the second growing season, all locations received more rainfall than the longterm average where RB received 3% (11 mm), JU 6% (28 mm) and JUST 10% (20.6 mm) above long-term average of rainfall. The rainy season in 2012 was terminated in April in all stations and was accompanied by high temperatures at the end of the season (Additional le 1). In 2013, the rainy season was terminated earlier than expected in Rabbah starting March, while only 7.8 mm were received in JUST that resulted in severe terminal drought conditions. The amounts of water provided by supplementary irrigation for irrigated plots are given in Additional le 1.
The combined ANOVA for the studied traits showed highly signi cant differences (P < 0.01) between genotypes, locations, water regimes, sowing dates, and their interactions (Additional le 2). The mean values for the studied traits across the six environments and their combinations with the sowing date and water regime are given in Additional le 3. For correlation analysis, positive signi cant correlations (P < 0.01) were found between all traits where GY showed a strong correlation with yield component-related traits and to less extent with HD, MD and GFP traits (Additional le 4).
The mean values of GY of tested genotypes (main effect of genotype) ranged from 585 g/m 2 in Rum (signi cantly the highest genotype) to the signi cantly lowest mean value of 367 g/m 2 in Steptoe. The main effect of sowing date treatment showed that December produced signi cantly the highest GY mean value (605 g/m 2 ) compared with February (341 g/m 2 ). The main effect of water regime treatment showed a signi cantly higher GY mean value for irrigated treatment (650 g/m 2 ) when compared with rainfed (296 g/m 2 ). The combined effect of the sowing date, water regime and genotype on GY showed high signi cancy level (P < 0.01) (Additional le 2). Rum produced signi cantly the highest mean value of To analyze genotype-speci c adaptation to speci c environment for the GY trait, the GGE biplot was used. For this purpose, the environments were reclassi ed based on the combination of each location, year, sowing date and water regime treatments to produce 24 distinct environments. The relationships between the GY of tested genotypes and the 24 tested environments and the degree by which each environment is represented are shown in Fig. 2a. The two principal components (PC1 and PC2) together captured 98.32% of the interaction effects and the variations due to GGE. Three irrigated environments from the 2013 season fell in the sector in which Acsad176 cultivar was the vertex genotype, which means that Acsad176 was the best genotype in these environments (Fig. 2a). On the other hand, Rum was the winner cultivar in the rest of the environments (21 out of 24) and speci cally in all rainfed environments irrespective of location, sowing date or season. By contrast, Morex and Steptoe did not win in any testing environments and the two genotypes were considered losers and they were the poorest genotypes across all tested environments (Fig. 2a). Interestingly, both genotypes were placed on different sectors indicating different responses to tested environments. For stability as measured by projection to the Average-Tester Axis y-axis, Rum was the most stable followed by Acsad176, while Steptoe and Morex were considered the least stable genotypes (Fig. 2b).
Based on the results above and to analyze the relationship between heading date and agronomic performance, a new correlation analysis was perfumed using the eld data of Rum and Steptoe after reclassifying the environments into four different groups based on the combination of sowing dates and water regime treatments (December-irrigated, December-rainfed, February-irrigated and February-rainfed). As shown in Additional le 5, negative signi cant correlations (P < 0.05) were found across the four groups between HD with GW and SPW. For December-irrigated conditions (Additional le 5), the negative signi cant correlations with HD included HI and TKW, while for December-rainfed conditions the negative signi cant correlations with HD included TW and STW (Additional le 5). For the February sowing date and irrespective of water regime treatment, negative signi cant correlations were detected between HD with TW, SN, GN, G.S. TKW and HI (Additional le 5).

Controlled conditions experiment
In this study, the transition to reproductive stage under LD was signi cantly faster in Rum genotype when compared with Steptoe irrespective of photoperiod, vernalization and water regime combinations ( The effects of water stress on RWC, stomatal resistance maximum quantum e ciency (Fv/Fm) of photosystem II have been analyzed in treated plants as an index for drought tolerance of the two tested genotypes. Signi cant differences between well-watered and stressed plants were observed for the tested parameters in the two tested genotypes under different photoperiod and vernalization conditions (Additional le 8). For instance, the mean values of RWC of the tested genotypes were lower in stressed plants irrespective of photoperiod or vernalization treatments and the effects were more pronounced after four weeks of incubation (Additional le 6). Starting the 4th week, the mean values of RWC of nonvernalized Steptoe plants grown under SD and water de cit conditions were signi cantly the lowest. Clear effects of water de cit stress on stomatal resistance and maximum quantum e ciency (Fv/Fm) of photosystem II were also observed in both genotypes irrespective of day length conditions or vernalization treatments (Additional le 8).
The expression patterns of Vrn-H1, Ppd-H1 and HVA22 genes were investigated in both tested genotypes in response to water de cit, vernalization and photoperiod treatments at selected time points. After two weeks of LD conditions, the relative expression levels of Vrn-H1 were higher in vernalized plants when compared with non-vernalized for both tested genotypes (Fig. 4). Irrespective of vernalization treatment, Rum always showed higher Vrn-H1 expression after two weeks of LD incubation than that observed in Steptoe. Irrespective of water regime treatment, Vrn-H1 expression levels after two weeks of LD incubation were signi cantly low in non-vernalized Steptoe plants compared with vernalized plants (Fig. 4). After four weeks of LD incubation, the same trend of high expression of Vrn-H1 was observed in vernalized plants when compared with non-vernalized plants, however, the expression levels were signi cantly lower in stressed Steptoe plants when compared with stressed Rum plants (Fig. 4). Under SD conditions, Rum showed higher Vrn-H1 expression levels when compared with Steptoe (Fig. 4). Interestingly, the expression of Vrn-H1 was signi cantly the highest in stressed and vernalized Rum plants when compared with Rum plants subjected to other combinations.
After two weeks of LD incubation and irrespective of water regime treatment, the Ppd-H1 expression in vernalized Rum plants showed the highest levels when compared with non-vernalized Rum (Fig. 4). At the same time, no signi cant differences were detected between Ppd-H1 levels in Steptoe plants subjected to different combinations of water regime and vernalization. After four weeks of LD incubation and irrespective of water regime treatment, vernalized Rum plants showed lower Ppd-H1 expression levels when compared with non-vernalized Steptoe and non-vernalized Rum plants (Fig. 4). For SD conditions, Ppd-H1 levels were slightly higher in Rum plants when compared with Steptoe with a slight increase under well-watered conditions (Fig. 4). For HVA22 gene expression, no major changes were observed between the different growth conditions under SD and LD for both Rum and Steptoe genotypes where both showed an increment in expression level under stress conditions (Fig. 4).

Discussion
The occurrence of abiotic stresses in marginal Mediterranean environments, primarily heat and drought, are strongly variable over space and time [35]. In these environments, terminal drought under rainfed conditions can lead to premature termination of plant growth before grain maturity and consequently can affect grain yield severely [36]. Therefore, the adoption of stable and drought tolerant cultivars is considered a useful approach to improve yield in such areas [3]. In this study, late sowing accompanied by drought stress resulted in lower GY when compared with early sowing treatment (Fig. 1a). This is expected knowing that late sowing will subject barley plants to different abiotic stresses, such as drought and heat, which are expected to negatively affect the growth and yield [37]. Furthermore, variations in the response of barley genotypes to different environments and treatments were obvious, con rming that GY was greatly affected by genotypic effects where the adapted local cultivars Rum and Acsad176, outperformed the non-adaptive cultivars, Steptoe and Morex (Fig. 2). In this study, Rum was superior and the most stable genotype under all tested environments, which indicated its general adaptability to multiple environments. Rum is the oldest released cultivar in Jordan that was bred for dry environments by CIMMYT/ICARDA and it is still preferred by many farmers in dry areas receiving more than 250 mm of rainfall. In many studies, Rum outperformed several Jordanian landraces and cultivars and was considered the most stable genotype under mild drought conditions [37,38]. Nevertheless, Rum performance was relatively moderate when compared with 150 Jordanian landraces under severe drought conditions and therefore, it is more suited to areas with mild drought conditions 346]. Compared to Rum, Steptoe was the poorest performer and the least stable genotypes across all tested environments and in particular under rainfed conditions (Fig. 2). Similar results were reported recently where Rum outperformed Steptoe in eld trials under drought conditions [16] and it was the most drought-affected genotype under controlled conditions [22]. In this study, the GY of tested genotypes was affected by the environment and treatment combinations, which in turn re ect different durations and intensities of drought stress and developmental stage at which the stress occurred. For instance, February-rainfed treatments resulted in a severe reduction in GY when compared with December-rainfed treatments, which could be attributed to severe terminal drought conditions accompanied by early owering phenotypes associated with the onset of LD conditions. Similar results were found by [37], who found that terminal drought conditions for late sown plants had adverse effects on barley yield.
In this study, drought seemingly accelerated HD by three days under eld conditions when compared with irrigated conditions. This is in general agreement with [39], who observed accelerated heading in barley plants under stress conditions. Steptoe was the last genotype to ower across different environments and was the most affected genotype by drought stress. Steptoe was described previously as a late owering genotype that has a reduced peduncle and few tillers per plant [22]. Such late-owering behavior was correlated with a severe reduction in GY when compared with early owering behavior observed in Rum (Additional le 5). This behavior was reported previously under controlled greenhouse conditions, where HD was negatively correlated with GY irrespective of water stress treatment [22]. In the same study, Rum owered earlier and outperformed Steptoe for GY and other yield-related traits under controlled drought conditions. In another study, Rum owered earlier under eld conditions when compared to Steptoe with a strong positive correlation between early owering and GY under drought conditions [16]. This is supported further by the fact that in dry environments, delay in owering time is commonly correlated with yield reductions [40].
Flowering time in temperate cereals, including barley, is a tightly regulated process where ower initiation and development are controlled by environmental signals, primarily by the exposure to low temperatures and day-length ( [14,25,41]. The responses to photoperiod and vernalization classify barley varieties according to their adaptation pattern into winter, spring and facultative types [9]. Such responses are governed by genetic networks and allelic variations in key regulatory genes that ne-tune barley owering in response to environmental stimuli [10,42]. During the fall, when barley seeds start to germinate, Vrn-H3 is repressed by Vrn-H2 [9,43]. During winter, vernalization up-regulates Vrn-H1, which results in the repression of Vrn-H2 in the leaves and, consequently, the activation of Vrn-H3 transcription in the spring [43,44,45]. Vrn-H2 functional allele was detected in Rum, while it was absent in Steptoe [22]. The presence of Vrn-H2 is required to prevent premature owering under LD conditions particularly after fall sowing [46]. However, no clear repression effect of Vrn-H2 on owering under LD in Rum was observed, although vernalization treatment seems to accelerate oral meristem development in Rum ( Fig. 3 and Additional le 6). Low expression levels of Vrn1 were associated with a delay in reproductive meristem initiation in wheat plant [47], while the high expression of Vrn-H1 accelerated owering in barley [12]. Rum carries a spring allele of Vrn-H1, which was up-regulated in response to vernalization and LD conditions in this study (Fig. 4), and this might alleviate the repressive effect of Vrn-H2 under LD conditions [31,43]. On the other hand, Ppd-H1 expression was also increased earlier in Rum under LD conditions in response to vernalization, which might indicate that the repression effect of Vrn-H2 can be alleviated in vernalized spring barley type as observed previously for vernalized winter types [43]. Interestingly, Steptoe lacks functional Vrn-H2 alleles and like Rum carries spring allele of Vrn-H1 and a functional Ppd-H1 but still, Steptoe behaved differently from Rum under LD conditions, which might indicate that the genetic network governing heading under LD conditions is different from that existing in Rum.
In this study, Rum was more responsive to vernalization treatments and owered earlier when compared with Steptoe (Additional le 7). In general, the overall gene expression levels of Vrn-H1 in response to vernalization treatment were higher in Rum than that in Steptoe (Fig. 4). This could explain the faster transition to the reproductive stage that occurred in Rum compared to Steptoe under LD and SD conditions and particularly in response to drought stress under SD conditions. Such differences in Vrn-H1 gene expression between genotypes were reported previously where higher expression levels were detected in S42-IL107 compared with Scarlett under LD conditions irrespective of drought treatment [21].
In this study, both tested genotypes carried different Vrn-H1 alleles where Rum has an insertion (Vrn-H1-7) and Steptoe has a deletion (Vrn-H1-4) at Vrn-H1 rst intron [22], which might explain the differential gene expression in response to different treatments. The high Vrn-H1 expression levels in vernalized Rum plants under SD and drought stress conditions were associated with a clear advanced developmental stage, while Steptoe, under the same conditions, showed a clear slower reproductive meristem development ( Fig. 3b; Additional le 6; Fig. 4). This also indicates that even though both genotypes are vernalization independent, there are still responses to vernalization in combination with other growth conditions and in response to stress, which might have implications for eld-grown plants [19].
In barley, allelic differences in Ppd-H1 and Ppd-H2 genes are known to be associated with natural variations in the responses to day length [48]. A functional LD-responsive Ppd-H1 allele was identi ed in both Rum and Steptoe, which explain their accelerated owering responses under LDs conditions when compared with SD conditions [42]. A positive role of the LD-responsive Ppd-H1 allele on GY was observed recently that was associated with shorter growing seasons under drought conditions [16]. Under these conditions, the functional Ppd-H1 allele was associated with increased GFP and increased TKW in barley plants. Still, Rum was found to be hyper-responsive to LD conditions compared to Steptoe indicating the existence of different genetic networks to perceive photoperiod signals in barley plant [49,50]. This is further supported by Ppd-H1 expression data under LD condition, where both genotypes behaved differently in response to vernalization treatment (Fig. 4). In Rum genotype, the expression level of Ppd-H1 after two weeks of LD incubation was higher under vernalization when compared with non-vernalized conditions, while the opposite was observed after four weeks of LD incubation. Notably, this affected the owering behavior under LD conditions where vernalized Rum plants owered earlier than non-vernalized plants (Additional le 7). Furthermore, no clear effect of drought on meristem development in vernalized Rum was observed under LD conditions. A similar trend was reported by [51] where the developmental stage of the apical meristem was similar in introgression lines carrying Ppd-H1 under control and stress conditions. In a recent study, [21] found that the expression of Ppd-H1 was not affected in response to drought under LD conditions in two tested genotypes. Still, the same study concluded that Ppd-H1 integrates photoperiod and drought stress signals to ne-tune reproductive development in barley. In Steptoe, there was no major difference in Ppd-H1 expression under LD conditions after two weeks of incubation, but after four weeks, the expression levels under vernalized conditions were higher when compared with non-vernalized plants. In this study, no clear association was found between Ppd-H1 and Vrn-H1 expression patterns under LD conditions with the owering behavior in Steptoe under drought conditions (Fig. 4).
Ppd-H2 is another gene that has been long acknowledged as responsible for the acceleration of owering in facultative barley types in response to the SD photoperiod and under mild winter conditions [52]. Ppd-H2 was identi ed previously in Rum and not in Steptoe [22], which might explain the failure of Steptoe to ower under SDs conditions. It was proposed that Ppd-H2 promotes the early owering of winter cultivars irrespective of photoperiod conditions and the existence of a dominant allele can promote owering in plants that have not satis ed their vernalization requirement [52]. Barley genotypes carrying the dominant allele of Ppd-H2 showed higher expression levels of Vrn-H1 and this might explain the higher levels of Vrn-H1 in Rum plants under SD conditions. It is worth noting that the expression level of Vrn-H1 under SD and stress conditions was higher in vernalized Rum plants (Fig. 4). An interplay between Ppd-H1 and Ppd-H2 to ne-tune plant development under SD and LD conditions in response to vernalization and stress conditions was recently suggested [16].

Conclusions
In this study, the eld performance of four selected spring barley genotypes was assessed across different environments and treatments. Rum and Acsad176 were well-adapted and stable across different environments and management practices when compared with Morex and Steptoe. The delayed heading in Steptoe was correlated negatively with GY under drought conditions, while it was associated with better performance in Rum. Analyzing the responses of both genotypes under controlled conditions identi ed a link between shoot apical meristem development under SD and drought stress conditions in vernalized Rum plants that were associated with higher gene expression of the Vrn-H1 gene. Furthermore, accelerated owering in Rum plants was observed in response to vernalization under LD conditions. This was also re ected in the early expression of Vrn-H1 and Ppd-H1 in Rum plants under these conditions. Under rainfed conditions, the early sowing of Rum subjected the plants to SD and vernalization followed by the onset of LD conditions at the end of the season. Under such conditions, a 30% reduction in GY was observed in Rum compared with a 60% reduction in Steptoe. This highlights the importance of selecting genotypes that can integrate vernalization and photoperiod responses to secure stable yield production in dry environments. Further studies are needed to dissect the genetic basis of oral meristem development and to identify QTL associated with yield stability under drought conditions. Finally, matching the developmental behavior of barley genotypes using different management procedures and dissecting the genetic information of these behaviors will help in developing new genotypes with stable performance under dry conditions.

Funding
This research was funded partially by the Deanship of Scienti c Research, The University of Jordan (Amman, Jordan) and the Arab Fund for Economic and Social Development (AFESD; Kuwait) project through ICARDA.

Con ict of Interest Statement
The authors have declared that no con ict of interest exists.