CHU Zhen-dong ,MlNG Bo,Ll Lu-luXUE JunZHANG Wan-xuHOU Liang-yu XlE Rui-zhiHOU PengWANG Ke-ruLl Shao-kun

1 Agricultural College,Shihezi University/Key Laboratory of Oasis Eco-Agriculture,Xinjiang Production and Construction Crops,Shihezi 832000,P.R.China

2 Institute of Crop Sciences,Chinese Academy of Agricultural Sciences/Key Laboratory of Crop Physiology and Ecology,Ministry of Agriculture and Rural Affairs,Beijing 100081,P.R.China

3 Heilongjiang Bayi Agricultural University,Daqing 163000,P.R.China

Abstract A high grain moisture content at harvest has been an important problem in the high latitude region of Northeast China,and it is closely related to the genotypes of varieties,local meteorological factors and planting management. However,delayed harvest at a low temperature could not effectively reduce the grain moisture content. In this study,we continuously observed the grain drying during the late stage of different maturing types of maize varieties in Daqing,Heilongjiang Province,China in 2016 and 2017. A two-segment linear model was used to analyze the different stages of the drying processes:1) Twosegment linear model fitting can divide the grain drying process of all varieties into two separate linear drying processes with different slopes. 2) During the rapid drying stage,the drying was faster at a higher temperature. The rate of slow drying was influenced by air vapor pressure. 3) The moisture content and meteorological factors when the drying rate turns from one stage into the other were not consistent between varieties and years. After entering the frost period,temperatures below 0°C will significantly reduce the rate of grain drying. 4) Due to the short growth period of early-maturing varieties,the drying time was prolonged,and the grain moisture content was lower than that of the mid-late maturing varieties. Local meteorological conditions do not allow the drying of mid-late maturing varieties to achieve a lower moisture content. When the temperature falls below 0°C,the drying rate of grain decreases markedly. Therefore,one feasible way to solve the problem of high moisture content is to replace the early-maturing varieties and implement the corresponding cultivation techniques.

Keywords:grain drying,maize,Northeast China,two-segment linear model

1.lntroduction

In order to meet the huge market demand with limited farmland,selecting relatively late maturing varieties and making full use of regional heat resources have been universal strategies for increasing yields in Northeast China in the past decades. Maize usually reaches physiological maturity in a period of short days before the first frost,so high grain moisture content (～30%) is a common phenomenon at harvest (Chaiet al.2017),and it is also an important factor influencing grain quality and post-harvest management(Jennings 1974;Cloningeret al.1975;Plett 1994;Wang and Li 2017b;Liet al.2018). Many farmers delay the harvest until after the frost to reduce grain moisture in the field. However,the effectiveness of grain drying at a low temperature in the field needs research-based justification.

Previous studies have shown that grain drying is affected by both genotype and environment (Schmidt and Hallauer 1966;Daynard and Duncan 1969;Carter and Poneleit 1973;Brooking 1990;Borráset al.2009;Elmore and Abendroth 2010;Nielsen 2011;Wang and Li 2017a;Gaoet al.2018). A study by Xiang (2011) on many inbred maize lines found that the grain moisture content was significantly correlated with precipitation in only five of the lines,but it was significantly correlated with crop heat units (CHU) in all the inbred lines.Daynard and Kannenberg (1976) also reported that CHU was significantly correlated with grain drying rate,and proposed using the equationy=c+dx2to first predict the grain drying rate and then the grain moisture content at harvest.In this equation,yis the grain moisture content,cis the initial grain moisture content,dis the maize grain drying rate (percentage decrease in grain moisture content per CHU),andxis the accumulated CHU after silking. Studies by Daynard and Kannenberg (1976) and Afuakwaet al.(1984) suggest that low temperatures induce physiological maturity in high moisture maize,while Daynard (1972) found that the mean maximum and minimum temperatures for maize to begin physiological maturation are 18 and 7–8°C,respectively. Newton and Eagles (2006) found that the mean maximum temperature during physiological maturation of maize in the cold highlands of Mexico was 18.3–18.9°C and that the mean minimum temperature was 8.5–9.1°C;while the grain moisture content was 30–37%. Pordesimoet al.(2006) observed that the grain moisture contents of two maize varieties had either stopped decreasing or began decreasing very slowly by October 18 in Tennessee,USA.During this period (October 15–October 21),the highest temperature was 15.3°C,the lowest temperature was 3.3°C and the average temperature was 9.3°C.

In high latitude regions,the frost-free period is short and temperatures drop rapidly beginning in autumn. The grain drying rate is greatly affected by temperature. However,long-term drying in the field often causes plants to undergo lodging (Xunet al.2017),which increases the difficulty of harvesting and causes yield loss (Thomison 2011).Therefore,it is very important to determine the relationship between grain drying and meteorological factors in order to select the ripening period of varieties according to the local climate characteristics,to establish planting plans and to determine the most suitable date for mechanical harvesting.The objectives of this study were:(1) to determine the drying characteristics of different maize varieties in Daqing City,Heilongjiang Province,China,and their relationships with temperature;and (2) to provide a basis for variety selection and mechanized harvesting scheduling in high latitude areas.

2.Materials and methods

2.1.Study site and varieties

The experiment was carried out at an agricultural experiment station affiliated with Heilongjiang Bayi Agricultural University in Daqing,Heilongjiang Province,China. The geographic location of the test station is 125°10′4′′E,46°34′41′′N,which lies in a typical northern temperate continental area of a monsoon climate. The average annual temperature is 4.2°C,average temperature of the coldest month is–18.5°C,extreme minimum temperature is–39.2°C,average temperature of the hottest month is 23.3°C,and extreme maximum temperature is 39.8°C. The annual frost-free period is 143 days,average annual wind speed is 3.8 m s–1,annual precipitation is 427.5 mm,ranging from 352 to 480 mm during the maize growing season (May–September),annual sunshine duration is 2 726 h,and sunshine duration during the growing season is 1 300 to 1 350 h. The soil is meadow soil,with the main physical and chemical properties of the 0–20 cm plow layer as follows:pH 8.15,organic matter 26.59 g kg–1,available nitrogen 154.28 mg kg–1,rapidly available phosphorus 19.25 mg kg–1,and available potassium 142.93 mg kg–1.

Eight representative varieties of maize that are typically planted in the region were selected for testing,including three early maturing varieties (DMY1,HT4 and DMY3) and five mid-late maturing varieties (NH103,SD636,XY335,XY998,and ZD958). The characteristics of the experimental varieties planted in 2016 and 2017 are shown in Table 1.We used a randomized block design in which each plot was 27.3 m2and contained six rows of 7 m in length. Each treatment was replicated three times. Plots were artificially seeded on April 28,2016 and May 8,2017. Planting density was 67 500 plants ha–1,with 65 cm equal line spacing. Other management practices were based on local field production practices.

Table 1 Planting dates and key growth and development stages of the different maize varieties for research trials conducted at Daqing,Heilongjiang Province,China in 2016 and 2017

2.2.Variations in meteorological conditions during the maize growth periods

The average temperature at the study site from May to December shows unimodal curve variation (Fig.1-A),rising gradually beginning on May 1,reaching a maximum between late July and early August,and then decreasing gradually until it reaches the minimum at the end of December. The active accumulated temperatures above 0°C were 3 222.4 and 3 279.4°C in 2016 and 2017,respectively,and the effective accumulated temperatures above 10°C were 1 566.0 and 1 554.3°C. The average temperature between late July and early September in 2016 was higher than that during the same period in 2017;while the average temperature from late September to mid-November was lower in 2016 than in 2017. Relative humidity levels were 64.74 and 58.66% from May to December in 2016 and 2017,respectively. The daily relative humidity was above 40% for most days from September to December,and the daily relative humidity was above 60% over the past 30 years(Fig.1-B). The GAB model (Labuza and Altunakar 2007)was used to analyze the experimental years and normal equilibrium moisture content in Daqing. Analysis of the meteorological data over the past 30 years showed that the equilibrium moisture content increased gradually afterSeptember,as the temperature dropped and the relative humidity remained at a high level. The perennial equilibrium moisture content in the drying phase was over 15%. Due to the high relative humidity in 2016,the equilibrium moisture content was significantly higher than that in 2017 (Fig.1-C).

Fig.1 Daily temperature (TEMP,°C) (A),relative humidity (RH,%) (B) and equilibrium moisture content (Meq,%) (C) during the maize growth periods (May to December) at the Daqing site,Heilongjiang Province,China. The experimental years (2016–2017)shown in circles and crosses,respectively;the 30-year average of meteorological data (1988–2017) shown in black solid line.

2.3.Sample and data procurement

We recorded the dates of sowing,seedling emergence,silking,and physiological maturity of the seedlings,then selected 200 plants for experimental monitoring that had the same growth,times of silking and pollination and pest free status at the silking stage. The date that each selected plant reached physiological maturity was recorded based on the disappearance of the milk line and the complete formation of the black layer.

In 2016,sampling was conducted for 89 d,from September 6 to December 4;the sampling intervals were 2 d in September,4 d in October and 7 d in November and December. In 2017,sampling was conducted from September 6 to December 9,and the sampling intervals were once every 5 or 7 d. Five representative ears were collected from each variety at sampling. When precipitation occurred,the sampling was delayed by one day. Ear samples were picked out by hand threshing. Samples of 100 grains in the middle of the sampled ears were weighed for the fresh weight (Wfresh). The grains were then dried at 85°C to a constant weight,and their weight was taken as the dry weight (Wdry). Grain moisture content (MC (%)) was calculated according to the following equation:

2.4.Calculation of the critical grain moisture content

Based on the dynamic trends of grain drying of different varieties obtained experimentally,a two-segment linear model was used to analyze the different stages of the drying processes. Two-segment linear model fitting divided the grain drying process into two separate linear drying processes with different slopes. MC was calculated daily according to the following equation:

whereTis an ordinal date,typically consisting of a day of the year ranging between 1 and 365 (starting on January 1);dis the constant of the segmentation model;gis the ordinal date of the turning point of the two-segment linear model.WhenT≤g,eis the slope during the rapid drying phase,which can be regarded as the daily average drying rate during the rapid drying phase;whenT＞g,his the slope during slow drying,which can be regarded as the average daily drying rate during the slow drying phase. Based on the nonlinear regression method provided by SPSS Software,the Levenberg-Marquardt iteration method was used to estimate the parameters in the two-segment linear model.

2.5.Data analysis

Excel 2013 was used to calculate and plot the data. SPSS 21.0 was used to fit the data for the variation of maize grain moisture content and statistically analyze the data.

3.Results

3.1.Maize grain moisture content

The observations of grain moisture content for both experimental years began in early September and ended in early December. During these periods,with the extension of drying time,the grain moisture content decreased monotonously (Fig.2). However,the drying process of a given variety was found to fluctuate,which may be related to a deficiency of sample representativeness. Because of the diversity of maturity dates and grain-filling processes in the different maize varieties,the grain moisture contents and drying processes were significantly different among the varieties. Three early-maturing varieties had grain moisture contents that were lower than those of the midlate-maturing varieties. In the same determination process,the differences of grain moisture content between earlymaturing and late-maturing varieties were between 20–26%in 2016 and between 15–21% in 2017. Compared with 2016,the average final grain moisture content was slightly lower in 2017.

Fig.2 Dynamics of maize grain drying of the different maize varieties in the research trials conducted at Daqing,Heilongjiang Province,China,in 2016 and 2017. MC,maize grain moisture content.

3.2.Drying process differences among the varieties

The results of nonlinear model regression analysis showed that the drying process of different varieties could be divided into two distinct linear drying processes with different slopes(Figs.3 and 4). The first phase has a steeper slope,so it can be called the rapid drying phase. The slope of the second phase was obviously less steep than that of the first phase,so it can be called the slow drying phase. The twosegment linear model could accurately simulate the changes of grain moisture content in 2016 and 2017 (Table 2). In general,most varieties obtained good fitting results,as the determination coefficients (R2) were greater than or equal to 0.95. Only XY998 had a coefficient lower than 0.89 in 2016.

The analysis of the drying process results for the earlymaturing varieties is shown in Fig.3. During the rapid drying phase,the average moisture content of the three earlymaturing varieties on the first sampling date was 37.04%,which dropped by 17.19 to 19.84% at the end of the rapid drying phase. The average daily drying rate was 0.87% d–1.In 2017,the average moisture content of the three earlymaturing varieties on the first sampling date was 38.50%,which dropped by 19.82 to 18.68% at the end of the rapid drying phase. The average daily drying rate was 0.61% d–1.

At the turning point of the two phases,the grain moisture contents of different varieties were between 15.48 and 20.29%,which were higher than the equilibrium moisture contents between 3.05 and 7.85%. During the slow drying phase of more than 50 d,the grain moisture content of the three early-maturing varieties dropped by only 3 to 5%.The average daily drying rate was 0.03–0.12% d–1. When the experiments finished,the grain moisture content of the early-maturing varieties finally decreased,and became very close to the equilibrium moisture content in the two experimental years.

The two-segment linear model regression analysis showed that the average grain moisture content of the midlate-maturing varieties was 54.38% at the beginning of the experiment in 2016,which dropped by 19.87 to 34.51% at the end of the rapid drying phase. The average daily drying rate was 0.93% d–1. In 2017,the average grain moisture content of the mid-late-maturing varieties was 52.31%,and the grain moisture content dropped by 24.83 to 27.48% at the end of the rapid drying phase. The average daily drying rate was 0.55% d–1(Fig.4).

Fig.4 Maize grain moisture content (MC) of mid-late maturing varieties (MLMVs) as predicted with the two-segment linear models for the research trials conducted at Daqing,Heilongjiang Province,China,in 2016 (left column) and 2017 (right column).The coefficients of the two-segment linear models are listed in Table 2.

In the slow drying phase,the grain moisture content of the five mid-late-maturing varieties dropped by 8.14% in 2016 and by 6.9% in 2017. The average daily drying rates of most varieties were between 0.14 and 0.21% d–1,and only SD636 and ZD958 had drying rates in 2016 that were less than 0.1% d–1. At the end of the experiments,MC of the mid-late-maturing varieties was 6.9–21.19% above the equilibrium moisture content in 2016,and it was higher (at 3.45–13.39%) in 2017.

By analyzing the turning point between the rapid and slow drying stages,the inflection of the daily average drying rate appears very close to the ordinal date of the first frost. Based on the relative root-mean-square error (RRMSE;Despotovicet al.2015),the differences between the turning point and FFD or physiological maturity in both years were analyzed.The results in Table 3 show that the differences between the turning point of drying rate and the first frost period in each year was significantly smaller than that between the turning point and the physiological maturity date. The slow drying phase of the early-maturing varieties began before the FFD in both years (October 5,2016 and October 15,2017;Fig.3),which was about 2–7 days before the FFD in 2016 and 1–6 days in 2017. For the mid-late-maturing varieties,the slow drying phase began around the FFD in both years.The differences of periods between the turning points and the FFD were between–3 and 7 in 2016 and between–1 and 7 in 2017.The turning point of all mid-late-maturing varieties was 2.7 days later than the FFD on average. But,the turning point of the drying rate of ZD958 occurred much earlier than those of the other varieties in 2016.

Table 2 Coefficients of the two-segment linear model for predicting grain moisture content (MC,%) of the different maize varieties in the research trials conducted at Daqing,Heilongjiang Province,China,in 2016 and 20171)

Fig.3 Maize grain moisture content (MC) of early-maturing varieties (EMVs) as predicted with the two-segment linear models for the research trials conducted at Daqing,Heilongjiang Province,China,in 2016 (left column) and 2017 (right column).The coefficients of the two-segment linear models are listed in Table 2.

3.3.Relationships between drying dynamics and the main meteorological indicators

The two-segment linear model divides the drying dynamics of different varieties into the rapid drying,slow drying and transition phases. The dynamic characteristics of grain drying in different varieties can be summarized as the slopes of the rapid and slow drying phases. Meanwhile,the grain moisture content at the turning point between the two drying phases is also an important dynamic characteristic of drying.

As shown in Table 3,the average temperature,the relative humidity and the equilibrium moisture content for each of the three phases were calculated. In the rapid drying phase,from the start of field observations to the drying turning point,the drying rate was positively correlated with average temperature,relative humidity and equilibrium moisture content (Meq). However,Meqis calculated from temperature and relative humidity,so it has a very significant autocorrelation with those two factors.Therefore,the stepwise regression analysis of temperature,relative humidity and drying rate (e) shows that temperature is the main factor affecting the rapid drying phase,withR2=0.719. Similarly,relative humidity andMeqwere positively correlated with drying rate during the slow drying phase,and the stepwise regression analysis of temperature,Meqand the difference between the turning point moisture content (MTP),Meqand drying rate (h) shows that the vapor pressure conditions and the differential between the kernel and outside environments are the main factors affecting the slow drying phase,withR2=0.643. During the period of slow drying,the average temperature dropped below 0°C,markedly lower than that during the rapid drying phase.The drying rates for the different varieties in this phase are significantly lower than those in the rapid drying phase.Under the influence of low temperature and high relative humidity in winter,the equilibrium moisture content in the slow drying phase increased,which was less favorable to grain drying. At the end of the experiment,the grain moisture content of the early-maturing varieties was very close to the equilibrium moisture content.

At the turning point,the different varieties changed from the rapid drying phase to the slow drying phase,with grain moisture contents from 16.6 to 41.8% in 2016 and from 15.4 to 34.9% in 2017. There was no consistency in the varieties between the different years. After tracing the meteorological factors within 7 days before and after the turning point,a period which can be called the transition phase,the average temperature when the turning point occurred was not consistent among the varieties,nor was it consistent between the years. In contrast,the other meteorological factors in the transition phases of different varieties were relatively consistent,but they still differed between the years.Therefore,temperature,relative humidity and equilibrium moisture content were not the determinants for the change in the drying rate (Table 4). As mentioned in Section 2.2,the turning point of the drying process in most varieties and both years occurred around the time of the first frost.

4.Discussion

The grain drying of maize is mainly affected by heat units before physiological maturation,while the atmospheric water demand should be considered after physiological maturation.In the two experimental years,since the temperature during the sampling phase in 2016 was higher than that in 2017,the drying rates of varieties in 2016 were faster during the rapid drying period. But the relative humidity in 2016 was also significantly higher than in 2017,especially in the slow drying phase,leading to a lower drying rate at the end of drying in 2016 than in 2017,and the grain moisture content at the end of the experiment was also higher than that in 2017.The relationships between the dynamic characteristics of drying and meteorological factors are consistent with the results of previous studies (Table 5).

Table 3 Analysis of the differences between the turning points (g) of the two-segment linear model and the physiological maturity(PM) and the first frost day (FFD)

Table 4 The dates and grain moisture contents (MC,%) of the turning points of the two-segment linear model,and averages of temperature (TEMP),relative humidity (RH) and equilibrium moisture content (Meq) during different grain drying phases of different maize varieties in the research trials conducted at Daqing,Heilongjiang Province,China in 2016 and 2017

Table 5 The correlations between slope and meteorological factors in the two different drying phases

After physiological maturity,a black layer appears and it blocks sucrose transport from the plant to the kernel.The drying of the kernel becomes a physical process of water loss. The reference model for this mechanical drying indicates that the declining moisture content per unit time is determined by the difference between the actual and equilibrium moisture contents (Strohman and Yoerger 1967;Thompsonet al.1968;Maioranoet al.2014;Martinez-Feriaet al.2019). Moreover,the equilibrium moisture content depends on the air vapor pressure (Earle and Earle 2004).Due to the short plant growth duration,early-maturing varieties can enter into the mechanical drying at an earlier date. This study found that the drying rate of the three earlymaturing varieties slowed down when the grain moisture content was close to the equilibrium moisture content in 2016. The division of the drying process is influenced by the water vapor pressure. Due to the differences in weather conditions between 2016 and 2017,the equilibrium moisture content in the slow drying phase was about 1.6% higher in 2016 than in 2017. At the end of experiment,the moisture content of the early-maturing varieties was about 2.3%higher in 2017 than in 2016,thus indicating that the end of the drying process of early-maturing varieties was affected by the water vapor pressure,and the grain moisture content of early-maturing varieties in this study could be reduced to the equilibrium moisture content.

We used a two-segment linear model to analyze the different stages of the drying processes. The late drying stage of different varieties can be divided into two phases:a rapid drying phase and a slow drying phase. However,the segmentation of the drying process was not related to the physiological maturity. Physiological maturity is reached when the grain moisture content drops to 30–35% (Borrás and Westgate 2006;Salaet al.2007). The results of our tests and relevant analysis showed that the grain moisture contents at the turning points were between 16.6 and 41.8% in 2016,and between 15.4 and 34.9% in 2017. But the turning point of the drying process occurred around the FFD in both years. The differences in the period between the turning point and the FFD were between–3 and 7 d in 2016 and between–1 and 7 d in 2017,respectively. Frost stress seems to affect the drying processes physiologically,which is more dramatic in varieties that have not yet reached physiological maturity.

This study was conducted in the Songliao Plain,which is an important maize producing area in China. Under the influence of a continental climate,autumn is frequently affected by a cold wave,and the temperature fluctuates drastically. However,the farmers there are inclined to plant mid-late-maturing varieties with higher yield potential,to increase yields under global warming. There was notenough time or heat units to complete the grain filling and drying processes. When the early frost stress occurs,even the mid-late maturing varieties have difficulty in reaching normal physiological maturity.In these cases,the grain moisture content is high and the grain quality is poor at harvest,and it is difficult to achieve mechanized grain harvesting (Liet al.2021). Three early-maturing varieties(DYM1,HT4 and DMY3) were studied in this paper. Their growth durations were shorter than those of other mid-late maturing varieties,thus the stages of grain filling and drying occurred earlier,with higher temperatures and shorter durations (Tanet al.2008). Therefore,the grain moisture content of the early-maturing varieties reached a low level in the early stage of sampling. However,there was a significant positive correlation between the maturity of varieties and yield (Fig.5),but determining how to effectively increase the yield of early-maturing varieties to replace the mid-late maturing varieties is an important research direction,and will be important for popularizing the application of mechanical grain harvesting technology in the future.

Fig.5 The relationship between yield and maturity of different varieties in the research trials conducted at Daqing,Heilongjiang Province,China.

5.Conclusion

The response of maize grain drying characteristics to meteorological conditions is widely documented,however,the dynamics of maize grain drying at low temperature has not been studied extensively. Our study provides a new insight regarding the impact of low temperature on high moisture content grain drying in the field. The grain drying characteristics of earlymaturing varieties under higher temperature conditions after physiological maturity are consistent with most results from previous studies. However,the grain drying rate of mid-late maturing varieties slowed down significantly after the frost,even though the grain moisture content was still high. This coincides with air temperatures falling below 0°C,or later than the 3rd wk of October in Heilongjiang Province,and the grain drying during the frost period is slow. Delaying maize harvest makes it difficult to reduce the grain moisture content effectively,but can raise the risk of yield loss due to stalk lodging. Overall,farmers in that region should consider replacing the commonly used varieties with ones in which the physiological maturity may be achieved earlier than 10 to 20 d before the frost period,to meet the requirement of lower grain moisture content. Earlymaturing varieties can provide more time and better temperature conditions for grain drying,thereby reducing the grain moisture content by 5 to 15% at harvest,which saves drying costs and avoids the risk of grain mildew.

Acknowledgements

This work was financially supported by the National Key Research and Development Program of China(2016YFD0300110),the National Natural Science Foundation of China (31971849),the China Agriculture Research System of MOF and MARA (CARS-02-25),and the Agricultural Science and Technology Innovation Program(CAAS-ZDRW202004).

Declaration of competing interest

The authors declare that they have no conflict of interest.