Barley Yield Response to Nitrogen Application under Different Weather Conditions

Barley, one of the most important crops worldwide, will be exposed to high air temperatures as a result of global warming. Since global warming is projected to progress with annual fluctuations, weather-adaptive cultivation techniques are needed in the area of barley production. This study aimed to determine the effect of nitrogen (N) application rate at heading on the grain yield of barley grown under different weather conditions based on two years of field experiments. Grain yield increased markedly with increasing N application rate in the 2017–2018 cropping season but not in the 2016–2017 cropping season. In contrast, late-emerging tillers clearly increased with increasing N application rate in the 2016–2017 cropping season but not in the 2017–2018 cropping season. Plants grown in the 2016–2017 cropping season produced relatively few grains due to the short period of tillering as a results of high air temperatures compared with those grown in the 2017–2018 crop season. Thus, in the 2016–2017 cropping season, N application could be used for the production of late-emerging tillers as a consequence of the limited sink capacity, whereas, in the 2017–2018 cropping season, it could be used effectively to increase grain yield.

Relationships between grain yield and its components. Correlation coefficient analyses were conducted to determine the traits that were most closely associated with variation in grain yield (Fig. 2). At all sowing dates, grain number m −2 exhibited a strong positive correlation with grain yield. Spike number m −2 was positively correlated with grain yield in early and normal sowings but not in late sowing. Similarly, spike number m −2 and grain number per spike were positively related to grain yield regardless of sowing date. However, 1000-grain weight was positively related to grain yield only in normal sowing.  Table 1. Mean grain yield, its component, and protein concentration as affected by different sowing date and nitrogen application rate at heading in the 2016-2017 crop season. *Significant at P < 0.05. **Significant at P < 0.01. ***Significant at P < 0.001. † Means within a column followed by the same lowercase letter do not differ significantly (P < 0.05). ‡ Means within a column followed by the same lowercase letter do not differ significantly (P < 0.05) among N application at heading for a given sowing date. Means within a column followed by the same uppercase letter do not differ significantly (P < 0.05) among sowing dates for a given N application at heading. § Not significant at P < 0.05.
www.nature.com/scientificreports www.nature.com/scientificreports/ Late tiller emergence. Late-emerging tiller increased with increasing N application rate at heading regardless sowing date in the 2016-2017 crop seasons (Fig. 3). However, late-emerging tiller increased with increasing N application rate in only normal sowing in the 2017-2018 crop season. In addition, there were more late-emerging tillers in the 2016-2017 crop season than there were in the 2017-2018 crop season.
Relationships between growth-related traits and late tiller emergence. Correlation coefficient analyses were conducted to determine the traits that were most closely associated with variation in late tiller emergence (Figs 4 and 5). At the all sowing dates, late tiller emergence exhibited a strong negative correlation with grain number m −2 (Fig. 4) and a strong positive correlation with nonstructural carbohydrate (NSC) amount m −2 and NSC amount per grain number (Fig. 5). Since there was no significant interaction between sowing date and nitrogen application rate at heading for test weight, each data represents mean value of all sowing times.  Table 2. Mean grain yield, its component, and protein concentration as affected by different sowing date and nitrogen application rate at heading in the 2017-2018 crop season. *Significant at P < 0.05. **Significant at P < 0.01. ***Significant at P < 0.001. † Means within a column followed by the same lowercase letter do not differ significantly (P < 0.05). ‡ Not significant at P < 0.05.
www.nature.com/scientificreports www.nature.com/scientificreports/ Growth-related traits at heading in the 2016-2017 crop season. Although leaf sheath plus stem weight decreased with later sowing date, whole-plant weight was not affected by sowing time (Table 3). Leaf area index (LAI) was not affected by sowing time, while SPAD chlorophyll meter value was increased with delaying sowing date. NSC concentration, NSC amount m −2 , and NSC amount per grain number decreased in response to delaying sowing date.
Growth-related traits at heading in the 2017-2018 crop season. Whole-plant and leaf sheath plus stem weights decreased with delaying sowing date (Table 4). Dead leaf blade weight was the greatest in early sowing, followed by late and normal sowings. Spike weight increased in response to delaying sowing date. LAI was not affected by sowing date, while SPAD value was increased with delaying sowing date. NSC concentration and NSC amount m −2 were greater in early sowing than in normal and late sowings. NSC amount per grain number was the greatest in early sowing, followed by late and normal sowings.  (Table 5). Daily mean air temperatures during mid-November, late December, and early January in the 2016-2017 crop season were 2.8, 2.5, 2.2 °C higher, respectively, than those in an average season. However, daily mean air temperatures during early and mid-December, late January, and early February in the  Represent the relationships between late-emerging tiller number and grain number m −2 in early, normal, and late sowings, respectively. (D-F) Represent the relationships between late-emerging tiller number and spike number m −2 in early, normal, and late sowings, respectively. (G-I) Represent the relationships between lateemerging tiller number and grain number per spike in early, normal, and late sowings, respectively. *Significant at P < 0.05. **Significant at P < 0.01. ***Significant at P < 0.001. NS, not significant at P < 0.05.  (A-C) Represent the relationships between late-emerging tiller number and NSC concentration in early, normal, and late sowings, respectively. (D-F) Represent the relationships between late-emerging tiller number and NSC amount m −2 in early, normal, and late sowings, respectively. (G-I) Represent the relationships between late-emerging tiller number and NSC amount per grain at heading in early, normal, and late sowings, respectively. *Significant at P < 0.05. **Significant at P < 0.01. ***Significant at P < 0.001. NS, not significant at P < 0.05.  Table 3. Growth related traits at heading as affected by different sowing time in the 2016-2017 crop season. *Significant at P < 0.05. **Significant at P < 0.01. ***Significant at P < 0.001. † Means within a column followed by the same lowercase letter do not differ significantly (P < 0.05). ‡ Not significant at P < 0.05.
www.nature.com/scientificreports www.nature.com/scientificreports/   Table 4. Growth related traits at heading as affected by different sowing time in the 2017-2018 crop season. *Significant at P < 0.05. **Significant at P < 0.01. ***Significant at P < 0.001. † Means within a column followed by the same lowercase letter do not differ significantly (P < 0.05). ‡ Not significant at P < 0.05.  www.nature.com/scientificreports www.nature.com/scientificreports/ Weather conditions at crop phenological growth stages. In all sowings, daily mean air temperatures at tillering were higher in the 2016-2017 crop season but lower in the 2017-2018 crop season than in an average season ( Table 6). Daily mean air temperatures at ripening in both crop seasons were high compared with those in an average season regardless of sowing date and N application rate at heading.

Month
In early sowings, daily mean solar radiations at stem elongation were higher in the 2016-2017 crop season than in an average season (Table 6). In normal and late sowings, daily mean solar radiations at stem elongation in the 2017-2018 crop season were high relative to those in an average season. In addition, in early sowings, daily mean solar radiations at ripening were higher in the 2017-2018 crop season than those in an average season.
In normal and late sowings, daily precipitations at stem elongation in the 2016-2017 crop season were lower than those in an average season. In early sowings, daily precipitations at stem elongation in the 2017-2018 crop season were higher than those in an average season. In addition, in all sowings, daily precipitations at ripening in the 2016-2017 crop season were lower than those in an average season.  (Table 7). Heading date in the 2016-2017 crop season was 3 to 12 days earlier and the duration of stem elongation was 3 to 10 days longer than those in the 2017-2018 crop season.
In both years, maturity was delayed a few days by the application of additional N at heading regardless of sowing date (Table 7).

Discussion
As a result of global warming, barley crops will be exposed to higher air temperatures and their grain yields are estimated to decrease significantly 17 , which will exacerbate the problems associated with the continuing increase in the world population further 3 . To feed a large number of people, improved cultivation techniques to increase the grain yield of barley crops, one of the most important crops worldwide, are necessary. Here, we reveal the effect of N application rates at heading on grain yield of barley grown under different weather conditions before heading and propose the possible role of the weather-adaptive N application techniques to improve the grain yield of barley.
It has been indicated that barley grain yield and grain weight decreased when plants ripened under high temperatures [9][10][11] . A previous study has reported that N application at heading resulted in increasing grain yield and grain weight 12 , whereas another study has reported no such effect 15 . In the present study, grain yield and test weight increased markedly with increasing N application at heading in the 2017-2018 cropping season but not in the 2016-2017 cropping season regardless of sowing date (Tables 1 and 2). In contrast, late-emerging tillers clearly increased with increasing N application rate in the 2016-2017 cropping season but not in the 2017-2018 cropping season (Fig. 3). Below, we discuss the reasons why N application was effective at increasing yield in the 2017-2018 cropping season but not effective in the 2016-2017 cropping season.
At all sowing dates, grain number m −2 , spike number m −2 , and grain number per spike in the 2016-2017 cropping season were very low compared with those in the 2017-2018 cropping season (Tables 1 and 2). It has earlier been reported that these three yield components of barley decreased when plants were grown at high air temperatures at tillering as a consequence of a shortening of the duration of tillering 18,19 . Daily mean air temperatures in mid-November, late December, and early January in the 2016-2017 crop season were higher than those in average season, while daily mean air temperatures in early and mid-December and late January, and early February in the 2017-2018 crop season were lower than those in average season (Table 5). Consequently, at all sowing dates, daily mean air temperatures at tillering in the 2016-2017 crop season were higher than those in the 2017-2018 crop season (Table 6), a difference associated with a shorter tillering period in the 2016-2017 crop season, relative to that in the 2017-2018 crop season (Table 7). Meanwhile, it was reported that barley grain number per m −2 , spike number, and grain number per spike decreased when plants grew under low solar radiation at stem elongation partially through a lack of photosynthetic products 20 . In normal and late sowings, daily mean solar radiations at stem elongation in the 2016-2017 crop season were lower than those in the 2017-2018 crop season (Table 6) (Tables 1 and 2). In addition, a previous study indicated that barley yield was decreased by high precipitation during growth but not high temperature based on the results of multiple regression analysis among yield and weather conditions 21 . However, daily mean precipitations in the 2016-2017 crop season were much less than those in an average season (Table 5). These results suggested that low values for grain number per m −2 , spike number, and grain number per spike in the 2016-2017 cropping season could be caused by high air temperatures at tillering.  www.nature.com/scientificreports www.nature.com/scientificreports/ Late-emerging tillers are considered to grow by using extra nutrients that cannot be translocated to the grain when its limited capacity. At all sowing dates and N application rates at heading, NSC concentration, NSC amount m −2 , and NSC amount per grain number at heading in the 2016-2017 cropping season were clearly higher than those in the 2017-2018 cropping season (Tables 3 and 4). Furthermore, late-emerging tiller had strong negative and positive correlations with grain number m −2 and NSC amount, respectively, regardless of sowing date (Figs 4  and 5). Thus, the emergence of a large number of late-emerging tillers in the 2016-2017 cropping season might be caused by applying N to plants with a large amount of extra NSC due to the limited sink capacity.
For two-rowed barley, grains with a test weight ≥709 g L −1 and a whiteness ≥40 are highly evaluated in Japan 22 . In the present study, Grain protein concentration increased as sowing date delayed and N application rate increased (Tables 1 and 2). N application at heading increased test weight but also protein concentration (Tables 1  and 2 and Fig. 1). It was known that N application at heading is highly effective to increase grain protein concentration 13,14,23 . In addition, plants sown at late were increased protein concentration due to their high nitrogen concentration at heading in the 2017-2018 cropping season (Tables 2 and 4). Since a negative correlation between grain protein concentration and whiteness, one of the indexes of grain quality, has previously been indicated 10 , a large amount of N application is not recommended in late sowing in the production area.
In conclusion, N application at heading in the 2016-2017 crop season could be used for the production of late-emerging tillers, whereas that in the 2017-2018 crop season could be used more effectively to increase grain yield because of the adequate sink capacity. At all sowing dates, grain yield and test weight in the 2017-2018 crop season were markedly increased by N application when plants ripened under high air temperatures (Tables 2  and 6). Therefore, we can propose this weather-adaptive N application technique to improve grain yield and test weight. In particular, when high temperature is expected at ripening, N should not be applied to plants at heading grown under high air temperatures at tillering, whereas it should be applied to plants at heading grown under low air temperatures at tillering.
Grain number m −2 had the strongest positive correlation with grain yield regardless of sowing date (Fig. 2). This result was in close agreement with results from previous reports [24][25][26] . In high latitude region of Japan, future surface air temperature is predicted to particularly increase in winter rather than in summer 16,27 , suggesting that grain number m −2 may decrease in response to increasing air temperatures in winter. To facilitate steady and stable barley grain production under conditions of climate change, the development of techniques or varieties with the ability to produce sufficient grain numbers under high air temperature conditions in winter is needed. Treatments included three sowing dates and three topdressings at heading, which were arranged as a split-plot experiment with three replicates in a randomised complete block design. The main plot and subplot were sowing date and N application rate at heading (i.e., when 90% of spikes have been headed), respectively.

Materials and
The day before sowing, plots were manually supplied with 50 kg ha −1 N, 43 kg ha −1 P 2 O 5 , and 43 kg ha −1 K 2 O in the form of synthetic fertiliser in both crop seasons. The fertilisers were incorporated into the soil by ploughing. A two-rowed hulled barley variety 'Haruka-Nijo' , which had recently been developed by Kyushu Okinawa Agricultural Research Center, NARO 28 and has rapidly extended its acreage in southwestern Japan, was used. The seeds, which were wrapped with seeder tape, were drill-sown by hand at 158 seeds m −2 on 7 November (early), 25 November (normal), and 12 December (late) in 2016 and on 7 November (early), 24 November (normal), and 12 December (late) in 2017. Plants were manually supplied with 40 kg ha −1 N, 34 kg ha −1 P 2 O 5 , and 34 kg ha −1 K 2 O and 20 kg ha −1 N, 17 kg ha −1 P 2 O 5 , and 17 kg ha −1 K 2 O at Zadoks growth scale 14 (four leaves emerged) and 31 (first node detectable) 29 , respectively, in the form of synthetic fertiliser in both crop seasons. In addition, plants were manually supplied with 0, 30, or 60 kg ha −1 N in the form of ammonium sulfate at heading. After trimming, each plot was 4.7 m wide × 2.8 m long and 6.8 m wide × 2.8 m long in the 2016-2017 and 2017-2018 crop seasons, respectively, with one ridge containing four rows spaced at 30 cm. sampling and measurement. At heading, plants from 1.12 m 2 (1.4 m wide × 0.8 m long) were sampled from each plot in both crop seasons. The number of spikes was counted and approximately 10% of the plants were separated into green leaf blades, dead leaf blades, leaf sheaths, leaf sheath plus stems, and spikes. After the area of green leaf blades was measured with a leaf area metre (LI-3050A/4, Li-COR, Inc., Lincoln, NE, USA), each plant part was dried at 80 °C in a ventilated oven for 2 d along with the remining plants to determine dry weight. The dried samples of leaf sheath plus stem were ground to a powder with a vibrating sample mill (TI-1001, CMT Co., Ltd., Tokyo, Japan) to measure NSC concentrations. The concentrations of NSC were determined as described by Ohnishi and Horie 30 .
At maturity, plants from 2.52 m 2 (1.4 m wide × 1.8 m long) were harvested in both crop seasons and air-dried until they reached a constant weight. Late emerging spikes were separated. The number of spikes was counted and the plants were threshed to determine their grain weight. The grains with a thickness ≥2.5 mm were recorded for grain weight. The number of grains with a thickness ≥2.5 mm required to make up 40 g was counted with a multi auto counter (KC-1M5, Fujiwara Scientific Co., Ltd., Tokyo, Japan) and the 1000-grain weight was calculated from this value. The test weight was measured with an instant multiple moisture tester (PM830-2, Kett Electric Laboratory Co., Ltd., Tokyo, Japan). Grain yield and 1000-grain weight were corrected to a 12.5% moisture concentration basis. The grain protein concentration was determined by Kjeldahl method (N × 5.83). In addition, the number of late-emerging spikes were counted.