Evaluating the Impacts of Waterlogging Disasters on Wheat and Maize Yields in the Middle and Lower Yangtze River Region, China, by an Agrometeorological Index

Abstract

Waterlogging disasters severely restrict crop production. The middle and lower Yangtze River region (MLYRR) is an important grain-producing region in China but suffers from severe waterlogging disasters. In this study, an agriculture-specific index called the accumulative humidity index was introduced to analyze the spatiotemporal characteristics of waterlogging during different wheat and maize growth stages in the MLYRR from 1960 to 2020. Additionally, the relationships between waterlogging intensities and crop yield fluctuations were revealed. The results showed that over the past 60 years, the intensity of wheat and maize waterlogging in the central and eastern MLYRR have increased; crop waterlogging was more intense in the 1990s–2010s than during the 1960s–1980s, and waterlogging intensity peaked in the 1990s. For both crops, waterlogging was more intense during the early growth stages, but its yield-reducing impacts were more significant during middle and late growth stages. The southern MLYRR (especially southern Anhui) was the region where both crops were most prone to waterlogging, but yields in this region were not severely affected by waterlogging. Compared with wheat, maize was more prone to waterlogging, and its yield was more significantly reduced by waterlogging. In conclusion, this study provides guidance for agricultural waterlogging risk reduction in the MLYRR.

1. Introduction

Increasing global climate change poses a major threat to agricultural production and food security [1,2]. The frequency of heavy precipitation events is increasing, and the resulting agricultural waterlogging disasters severely affect the growth and yield of food crops. Therefore, it is crucial to conduct a thorough assessment of the impact that waterlogging disasters have on agricultural areas to ensure food security. Moreover, at the regional scale, quantitative research concerning the intensity and impact of waterlogging occurring during crop growth periods (crop waterlogging for short) can provide support for many issues, including optimizing drainage schedules by identifying high-risk regions and periods [3], regional-scale monitoring and early warning of crop waterlogging [4], regional-scale rapid estimation of waterlogging-induced yield loss in the end of crop growing [5], and predicting future crop waterlogging trends in different regions based on future climate simulation datasets, e.g., CMIP6 [6].

At present, hydrometeorological indices have been used extensively to assess the impact of drought and flooding at the regional scale. These indices include simple precipitation-based indices, such as the Palmer’s Z index (Z-index) and standardized precipitation index (SPI) [7,8], as well as more comprehensive indices that account for the water budget, such as the standardized precipitation evapotranspiration index (SPEI) [9]. To date, many of the above indices have been applied to characterize agricultural drought and waterlogging disasters in various regions. For instance, in terms of studies focused on agricultural waterlogging, many scholars have employed commonly used indices such as precipitation, SPI, and SPEI to characterize waterlogging intensities during the growth stages of rice [10], maize [11,12], wheat [13], barley [1], cotton [3], and sugarcane [14], and have further examined the impacts of waterlogging on crop yield variability. However, from the perspective of agricultural waterlogging, even short-term waterlogging events can result in significant crop yield losses [15]. The hydrometeorological indices used in the abovementioned investigations were typically with long-time scales (e.g., monthly); this means that they emphasized the conditions over a long period but ignored the influence of short-term waterlogging events. Therefore, it is suggested to use the indices with short timescales when describing waterlogging intensity during crop growth [16]. In addition, the hydrometeorological indices mentioned above were established based on the concepts of hydrometeorology, thereby largely overlooking the specific characteristics of agriculture and crops, which impacts their applications in regional agricultural waterlogging assessments. Nevertheless, there are some other agricultural factors in addition to the timescale of indices that can affect field moisture conditions, such as persistent soil moisture content and variable crop water demand over growing period. Therefore, to assess the impacts of waterlogging disasters on agricultural crops, it is preferable to use more agriculture-specific indices.

As one of the most important agricultural production areas in China, the middle and lower Yangtze River region (MLYRR) contributes the greatest variability to China’s grain production [12]. Under the influence of the subtropical monsoon, extreme precipitation occurs frequently in this region [17] and its frequency is increasing [18,19]; as a result, this region is known as one of the regions in China most prone to waterlogging [20,21]. Wheat and maize are important food crops in the MLYRR, and they are majorly cultivated in Anhui, Hubei, and Jiangsu provinces (according to the yield statistics over the last 20 years, they have contributed 98.2% of the total wheat yield and 81.9% of the total maize yield in the MLYRR). However, these three provinces have been facing severe agricultural waterlogging disasters. In Hubei Province, approximately 806,200 hm² of farmland were affected by waterlogging disaster every year [22]. In recent years, 70% of the meteorological disasters in Jiangsu Province can be attributed to waterlogging [20]; in the Jianghuai Plain of Anhui Province, waterlogging disasters occur with a frequency of 2.5 years, which results in ~10% of the winter wheat yield being lost [20]. Therefore, it is essential to accurately assess the impacts of crop waterlogging disasters in the MLYRR, especially for the main food-production regions. To date, some scholars have employed meteorological indices and yield data to evaluate the impacts of waterlogging disasters during the growth period of wheat and maize in the MLYRR. For example, taking the Huaihe River Basin (in both Anhui and Jiangsu provinces) as the study area, Gao et al. [13] used the SPEI and SPI to quantify the intensities of drought and flood during different winter wheat growing periods and identified the relationship between meteorological factors and residual crop yields; they found that flooding reduced the crop yield the most during the middle growth period. Liu et al. [21] analyzed the temporal changes in meteorological elements and winter wheat yield in the MLYRR and concluded that heavy rainfall was the main factor that limited winter wheat yields. Song et al. [23] investigated the effects of waterlogging disasters on winter wheat in the MLYRR by using weighted average precipitation; their results demonstrated that excessive precipitation negatively affected the yield and quality of winter wheat in this region. Nevertheless, current investigations in this field still encounter some limitations: (1) Hydrometeorological indices are usually directly used to address agricultural issues, while disregarding agricultural and crop characteristics; (2) Current analysis of waterlogging disasters in the MLYRR has concentrated on winter wheat; in comparison, studies of other major crops that are sensitive to waterlogging, such as maize, are few in number. Needless to say the comparison between different grain crops in terms of waterlogging impacts, remains unresolved; (3) Finally, regional waterlogging investigations should focus on the distinctive impacts of waterlogging disaster occurring at different growth stages, especially in determining how waterlogging impacts the final crop yield losses during different growth periods.

Given these concepts, this work aimed (a) to characterize the degree that waterlogging impacts wheat and maize in the MLYRR by using an agriculture-specific index, and (b) to further reveal the regional-scale relationships between the intensity of waterlogging at different growth stages and crop yield variability. To achieve these goals, the accumulative humidity index (AHI) was introduced to describe crop waterlogging intensity in the three major wheat and maize producing provinces from 1960 to 2020. On this basis, the correlation between crop waterlogging intensity and detrended crop yield (also known as climatic yield) was analyzed to evaluate the effects that waterlogging has on the wheat and maize yield at different growth stages. According to previous field-scale findings, the yields of wheat [24,25,26] and maize [27,28,29] can be negatively and significantly affected by waterlogging conditions, and their waterlogging responses vary with growth stages. Therefore, in this study, we hypothesize that the waterlogging intensity over wheat and maize growing stages, which is characterized by the AHI, can be significantly and negatively related to crop yield fluctuation in many areas, and, moreover, these negative relations differ among growth stages.

2. Materials and Methods

2.1. Study Region

Hubei, Jiangsu, Anhui provinces, the three major wheat and maize production provinces in the MLYRR (located between 29°05′–35°20′ N and 108°21′–121°57′ E; Figure 1) were selected as the study region. In the study region, the annual precipitation is approximately 700–1600 mm, and the annual mean temperature ranges from 13.6 °C to 18 °C [30]. Over the past 30 years, wheat (Figure 2a) and maize (Figure 2b) yields in the study area have shown increasing trends; in particular, the increases in wheat yields were significant in all provinces. Moreover, the areas that yielded the largest amounts of wheat were concentrated in the eastern MLYRR (Figure 2c), while the high-yield maize-producing areas included the southwestern MLYRR and eastern MLYRR (Figure 2d).

2.2. Data Sources

The meteorological data used in this study mainly include daily precipitation, relative humidity, and air temperature, which were collected from the China National Meteorological Data Service (http://data.cma.cn (accessed on 1 January 2022)). Up to 64 national-level meteorological stations were used (Figure 1). The annual yields of wheat and maize (10 kt/ha) in each district in the MLYRR were collected from the provincial statistical yearbooks of Hubei (1990–2020), Anhui (1990–2020), and Jiangsu (1999–2020). In addition, the starting and ending dates of each growth stage for wheat and maize in the MLYRR were based on an authoritative reference book “Atlas of the Main Crop Growth Periods in China” [31]. For wheat, since winter wheat is unequivocally the dominant type of wheat cultivated across the MLYRR, the wheat growth stages referred to in this study are the winter wheat growth stages. For maize, summer maize (mainly distributed within Anhui and Jiangsu provinces) and spring maize (mainly distributed within Hubei province) are both cultivated in the MLYRR; therefore, the maize growth stages were determined with a regard to the type of maize crops grown locally. The specific divisions are shown in

Table 1. Dates of wheat and maize growth stages in the MLYRR.

Province	Wheat				Maize
	Seedling	Jointing	Heading	Mature	Seedling	Jointing	Mature
Hubei	11.1–12.9	12.10–3.4	3.5–4.9	4.10–5.31	4.1–5.19	5.20–6.19	6.20–8.10
Anhui	11.1–12.4	12.5–3.4	3.5–4.19	4.20–5.31	6.1–6.24	6.25–7.24	7.25–9.10
Jiangsu	11.1–11.29	11.30–3.9	3.10–4.19	4.20–5.31	6.11–7.4	7.5–7.24	7.25–9.10

.

2.3. Research Methods

The methodology followed by this study is outlined in Figure 3. First, daily meteorological data were collected from national meteorological stations to calculate the accumulative humidity index (AHI) at each station. Afterwards, based on the division of wheat and maize growth stages in different provinces, the intensities of waterlogging disasters during different growth stages were quantified and their spatiotemporal characteristics were revealed. Additionally, historical yield data for wheat and maize in various regions of the study area were collected, and the crop yield data were detrended to obtain the climatic yield. Finally, the relationships between the climatic yield and the intensity of waterlogging during different growth stages were analyzed by using Pearson correlation analysis; the results of the correlation analysis were used to assess the impact waterlogging disasters had on wheat and maize in the study region.

2.3.1. Characterizing Degree of Waterlogging during Crop Growth Stages

The accumulative humidity index (AHI) was adopted because it considers crop evapotranspiration in addition to precipitation. Additionally, it includes the carry effect of previous field moisture conditions. This index has been applied by some previous works [16,32,33,34]. The AHI originates from a classical index, i.e., the relative humidity index, which is a commonly used index recommended in the Chinese National Meteorological Drought Level guidelines [35] and can be calculated as follows:

$${\mathrm{M}}_{\mathrm{i}}=\frac{\mathrm{P}-\mathrm{E}{\mathrm{T}}_0}{\mathrm{E}{\mathrm{T}}_0}$$

(1)

where M_i, P, and ET₀ indicate the relative humidity index, precipitation (mm), and ET₀ (mm) during the calculation period, respectively. In this study, this was calculated through the Penman–Monteith equation recommended by FAO [36]; detailed description is available at https://www.fao.org/3/X0490E/x0490e06.htm#equation (accessed on 1 January 2022).

Agricultural drought and flood indices are expected to describe field moisture conditions over crop growing periods. In this regard, we used the potential evapotranspiration of field crops (ET_c) to replace the reference crop evapotranspiration (ET₀) in the relative humidity index, resulting in more specific evapotranspiration in the fields. Thus, the Mi can be calculated as follows:

$${\mathrm{M}}_{\mathrm{i}}=\frac{\mathrm{P}-\mathrm{E}{\mathrm{T}}_0}{\mathrm{E}{\mathrm{T}}_{\mathrm{m}}}$$

(2)

$$\mathrm{E}{\mathrm{T}}_{\mathrm{c}}={\mathrm{K}}_{\mathrm{c}}\times \mathrm{E}{\mathrm{T}}_0$$

(3)

where ETc is the potential evapotranspiration of crops (mm) during the calculation periods. K_c is the crop coefficient during corresponding periods, which depends on factors such as crop type, growth stage, and soil conditions. The calculation of K_c in this study followed the single crop coefficient method; specifically, based on the recommended K_c of wheat and maize by FAO, local data, including wind speed and humidity, were used to modify the K_c [36].

To reflect the accumulative influence of previous soil moisture conditions on current moisture status, the accumulated humidity index (AHI for short), which was derived from the relative humidity index and had 10-day temporal resolution, was established to account for the long-term effect of previous moisture conditions [32,33]. The AHI consists of two components: the humidity index for the current 10-day period and the humidity index for previous periods. In addition, a climatic characteristic variable α is included to reflect the relationship between water supply and demand; the weights of the humidity index for the current 10-day period and for the previous periods are α and (1 − α), respectively. Since crop evapotranspiration is associated with air temperature, α is determined by the 10-day average air temperature. In [32,33], the appropriate α under different air-temperature conditions was determined by using 10-day agricultural meteorological reports and actual soil moisture conditions (

Table 2. Weighting coefficient under different air temperatures.

Average Temperature (T/°C)	T ≥ 25	25 > T ≥ 20	20 > T ≥ 15	15> T ≥ 10	T < 10
α	0.7	0.6	0.5	0.4	0.3

Note: α is an empirical coefficient to reflect the relationship between water supply and demand.

).

The influence that previous field moisture conditions have on current conditions can be characterized from two aspects. On the one hand, the number of previous periods considered was dependent on the conditions of the current season. Specifically, summer corresponds to high air temperatures and vigorous crop growth, which increases field evapotranspiration and demonstrates the moisture from the previous season had only a short-term influence (i.e., fewer previous periods should be considered). In comparison, winter corresponds to low air temperatures and slow crop growth, which suppresses field evapotranspiration and shows that the moisture from the previous season had a long-term influence (i.e., more previous periods should be considered). On the other hand, the time interval between the previous periods and the current period is another important factor; a closer time interval indicates that the previous moisture conditions exert a greater influence on the current status. Given these concepts, an empirical formula was derived to describe the accumulative impacts of previous moisture conditions:

$${\displaystyle \sum _{\mathrm{i}=1}^{\mathrm{n}}\left(\frac{\mathrm{n}+1-\mathrm{i}}{{\displaystyle {\sum }_{\mathrm{i}=1}^{\mathrm{n}}\mathrm{i}}}\right)}\times {\mathrm{M}}_{\mathrm{i}}$$

(4)

where n is the number of previous 10-day periods to be considered, which is set to five for winter, four for spring/autumn, and three for summer. M_i is the humidity index for the i-th period preceding the current period. $\frac{\mathrm{n}+1-\mathrm{i}}{\sum _{\mathrm{i}=1}^{\mathrm{n}}\mathrm{i}}$ is a dynamic weighting coefficient that reflects the contribution of the soil moisture conditions during the ith previous period. It is evident that the sum of all the weights of n previous periods equals 1.

Finally, the AHI formula can be written as:

$$\mathrm{A}\mathrm{H}\mathrm{I}=\mathrm{\alpha }\times {\mathrm{M}}_0+\left(1-\mathrm{\alpha }\right)\times \left[{\displaystyle \sum _{\mathrm{i}=1}^{\mathrm{n}}\left(\frac{\mathrm{n}+1-\mathrm{i}}{{\displaystyle {\sum }_{\mathrm{i}=1}^{\mathrm{n}}\mathrm{i}}}\times {\mathrm{M}}_{\mathrm{i}}\right)}\right]$$

(5)

Similar to other meteorological indices established according to the principles of water balance (i.e., SPEI), higher AHI values indicate wetter moisture conditions. Considering that the growth stages of wheat and maize consisted of several 10-day periods, the AHI values over each growth stage were averaged to represent the water conditions of this period. Moreover, since this study focused on the impacts of waterlogging, the influence of drought should be minimized when characterizing waterlogging intensity and when analyzing the impact of waterlogging on crop yield. To address this concern, this study followed previous practices [14,37] and classified the years considered in every district as ‘wet conditions (the top 30%)’, ‘near-normal conditions (the middle 40%)’, and ‘dry conditions (the lowest 30%)’; this classification was performed based on annual wheat/maize AHI values. Through this filtering process, the impacts of dry years were generally excluded. Then, when quantifying the waterlogging intensity of wheat and maize at each station, the AHI in the near-normal and wet years was employed; higher AHI values indicate more intense waterlogging conditions during the crop growth periods.

2.3.2. Spatiotemporal Characteristics of Crop Waterlogging Disasters

The linear trend method was used to describe the changing trend of the waterlogging intensity:

$$\mathrm{Y}=\mathrm{k}\times \mathrm{t}+\mathrm{b}$$

(6)

where Y is the waterlogging intensity index (i.e., AHI). t indicates the years of the calculation period. k is the regression coefficient and k × 10 represents the climate tendency rate of waterlogging intensity. k > 0 and k < 0 refer to increasing trends and decreasing trends of waterlogging intensity, respectively. Additionally, the significant result of a regression model (p < 0.05) indicated that significant change in waterlogging intensity occurred during the calculation period.

In the spatial analysis, ArcGIS software (version 10.2; ESRI, Redlands, CA, USA) was used to illustrate the spatial distribution of crop waterlogging intensities (i.e., AHI) during different crop growth stages. Based on the AHI results at each station, the result for the entire study region was obtained by spatial interpolation; the inverse distance weighting method was used for interpolation, and the spatial resolution of computations was 0.1° × 0.1°.

2.3.3. Climatic Yield

Time series of crop yield can be primarily divided into trend yield and detrended yield. The trend yield was determined by nonclimatic factors, such as advances in agricultural technology and improvements in field management. The detrended yield (also known as climatic yield) refers to short-term yield fluctuation, which is induced by meteorological factors, especially by natural disasters such as drought and flooding. Currently, there are many methods proposed for detrending crop yields; however, this issue was not a focus in this study. Thus, we employed the popular five-year moving average method [10,38] to determine trend yields (Y_tr) of wheat and maize. Finally, the climate yield (Y_cl) of wheat and maize was the difference between the actual crop yield (Y_act) and the trend yield:

$${\mathrm{Y}}_{\mathrm{c}\mathrm{l}}={\mathrm{Y}}_{\mathrm{a}\mathrm{c}\mathrm{t}}-{\mathrm{Y}}_{\mathrm{t}\mathrm{r}}$$

(7)

2.3.4. Relationships between Climatic Yield and Waterlogging Intensity

After computing the climatic yields of wheat and maize, the relationships between climatic yields and the waterlogging intensity during crop growth stages were investigated by performing Pearson correlation analysis. In the following, these relationships are referred to as “Y_cl-AHI correlations”. Due to the differences in waterlogging intensity and crop waterlogging sensitivity between different growth stages, the Y_cl-AHI correlations were calculated for different growth stages. Then, by examining the correlation coefficients in each region, the impacts that waterlogging disasters have on crop yield in each region were assessed. Furthermore, when the relationships were found to be negative and significant (p < 0.05), it was considered that local waterlogging disasters had a significant negative impact on crop yield during that growth stage [14].

3. Results

3.1. Temporal Trends

During the seedling stage of wheat (Figure 4a), the changing trends of wheat AHI commonly increased in the central and eastern MLYRR (i.e., Anhui and Jiangsu) but decreased in the western MLYRR, i.e., Hubei. At the subsequent jointing stage (Figure 4b), almost all the stations witnessed increasing trends of wheat AHI, particularly in Anhui and Jiangsu, in which most stations detected significant increases in wheat AHI. However, during the following heading stage (Figure 4c), decreasing trends occurred at many stations in the northeastern MLYRR. Furthermore, at the final mature stage (Figure 4d), the decreasing trends of wheat AHI spread to the whole study region; up to 14 stations reported significant decreases in wheat AHI. In terms of the whole growth period (Figure 4e), most areas of the MLYRR saw increasing trends of wheat AHI, especially for central MLYRR, i.e., Anhui. Therefore, it can be concluded that the changing trends of wheat waterlogging proneness varies greatly with crop growth stages. Furthermore, it can be noted that the predominant trend indicates that wheat has an increasing proneness to waterlogging.

For maize, during the seedling stage (Figure 5a), the AHI increased in the central and eastern MLYRR (i.e., Jiangsu and Anhui) but generally decreased in the western MLYRR (i.e., Hubei). During the subsequent jointing stage (Figure 5b), the increasing trend of maize AHI increased in Jiangsu and Anhui; up to nine stations witnessed a significant increase in AHI. During the final mature stage (Figure 5c), the changing trends of maize AHI at different stations were still primarily increasing; however, the number of significantly decreasing trends was less than that at the jointing stage. Finally, during the entire growth period (Figure 5d), the maize AHI in the eastern MLYRR (i.e., Jiangsu and Anhui) showed large increasing trends; at seven stations, significantly increasing trends in the AHI were observed. In comparison, the changing trends of maize AHI in Hubei decreased in most areas. Hence, over the past 60 years, the proneness of maize waterlogging has increased in the central and eastern MLYRR, with significant increasing trends being observed at many stations.

In terms of different provinces, the changing AHI trends during maize and wheat growth stages are provided in

Table 3. Climatic tendency rates of AHI during wheat and maize growth stages.

Province	Wheat Growth Stage					Maize Growth Stage
	Seedling	Jointing	Heading	Mature	Entire Stage	Seedling	Jointing	Mature	Entire Stage
Hubei	0.0207	0.0604	−0.0411	−0.0973 *	−0.0143	0.0207	0.0604	−0.0973 *	−0.0143
Anhui	−0.0013	0.1159 *	−0.0030	−0.1046 *	0.0018	0.1755	0.2503	0.0039	0.1433
Jiangsu	0.0525	0.1116 **	−0.0508	−0.0875 *	0.0065	0.0315	0.1626 *	0.0172	0.0704 *

Note: Bold fonts indicate that the climatic tendency rate of AHI was significant; * and ** indicate p < 0.05 and p < 0.01, respectively.

. For wheat, it is worth noting that the AHI significantly increased during the jointing stage in Anhui and Jiangsu but significantly decreased during the mature stage in all provinces. For maize, the climatic tendency rates of AHI in Jiangsu were great, and the increasing trend of AHI over the entire growing period in Jiangsu was significant, reconfirming our previous results, shown in Figure 5d. In addition, there was a significant decreasing trend in maize AHI during the mature stage in Hubei.

3.2. Interdecadal Analysis

The intensity of wheat and maize waterlogging in different decades is illustrated in Figure 6. It is apparent that the last three decades (i.e., 1990s, 2000s, and 2010s) experienced more intense waterlogging than those before; additionally, the intensity of wheat and maize waterlogging reached a historical high in the 1990s, followed by the most recent decade, i.e., the 2010s. This finding implies a near-term high risk that waterlogging disasters posed to wheat and maize. Comparing the AHI values of wheat and maize, it was found that in each decade, maize suffered more intensive waterlogging than wheat. Finally, in terms of different provinces, the intensities of wheat and maize waterlogging were similar in Anhui and Hubei; both of which were much higher than that in Jiangsu.

3.3. Intergrowth-Stage Distribution

As shown in Figure 7a, wheat was more prone to waterlogging at early growth stages (especially the seedling stage) than at later growth stages, except for the mature stage in Hubei. The intensity of wheat waterlogging was lower in Jiangsu than in the other two provinces. For maize (Figure 7b), the differences in waterlogging intensity between different growth stages varied greatly with region. In Hubei, maize suffered severe waterlogging disasters at all growth stages, especially at the initial growth stage. In Anhui, maize waterlogging disasters were concentrated within the first two growth stages. By comparison, in Jiangsu, the proneness of maize waterlogging was much lower than in the other provinces, occurring only during the middle stage (jointing). In summary, both wheat and maize were less prone to being waterlogged during late growth stages. In addition, wheat and maize waterlogging was less intensive in Jiangsu than in Hubei and Anhui.

3.4. Spatial Characteristic Analysis

During the wheat seeding stage (Figure 8a), the frequency of waterlogging was generally high in the study region, and the southern MLYRR faced more serious waterlogging disasters. During the jointing stage (Figure 8b), the distribution of waterlogging events was similar to that at the seedling stage, but waterlogging intensity generally decreased in northern regions. During the subsequent heading and mature stages (Figure 8c,d), the waterlogging risk increased in the southern MLYRR (especially the southern part of Anhui); a similar result was also found over the whole growth stage of wheat (Figure 8e). The above results indicated that wheat waterlogging intensity was generally higher during the early growth stages; during the late growth stages, waterlogging was concentrated in the southern MLYRR. In particular, the southern part of Anhui was the most waterlogging-prone region during all growth stages.

According to Figure 9a, the regions most prone to waterlogging during the maize seedling stage were mainly distributed in the southern MLYRR, especially in southern Anhui. In the following jointing stage (Figure 9b), the proneness of maize waterlogging increased in southern Anhui and decreased in other areas. During the final mature stage (Figure 9c), maize waterlogging intensity was low across the whole study region. In terms of the whole growth period (Figure 9d), southern Anhui was obviously the region where the risk of maize waterlogging was high.

According to the above results concerning both crops, even though the growing months of maize and wheat within a year differed completely, the waterlogging intensity of the two crops in the MLYRR showed similar geographical distribution characteristics. The southern MLYRR was more prone to crop waterlogging than the northern MLYRR, and southern Anhui was the region where the increased proneness of both crops to waterlogging overlapped.

3.5. Relationships of Wheat and Maize Climatic Yield vs. Waterlogging Disaster Intensity

Figure 10 shows the correlation between the climatic yield of wheat and the AHI during each growth stage in the study area (hereafter referred to as Y_cl-AHI correlations). For the seedling stage (Figure 10a), the Y_cl-AHI correlations were almost all negative in the eastern MLYRR (i.e., Jiangsu) but were mostly positive in the central and western MLYRR (Hubei and Anhui). In the subsequent jointing stage (Figure 10b), numerous negative Y_cl-AHI correlations appeared, especially in Anhui. Up to three districts had significant Y_cl-AHI correlations, which were all negative. At the later heading stage (Figure 10c) and mature stage (Figure 10d), negative Y_cl-AHI correlations became more widespread than during previous growth stages; additionally, the negative correlations were stronger, and many significantly negative results were detected (i.e., seven and six for the heading stage and mature stage, respectively). In terms of the whole growth period (Figure 10e), the Y_cl-AHI correlations were often negative, and the negative correlations appeared stronger in the southern MLYRR. In conclusion, from the jointing stage, the effect that waterlogging had on wheat yield was mainly negative; especially during the heading stage and mature stage, wheat yield in many districts was significantly reduced by waterlogging. In addition, among the three provinces, Anhui was subjected to the most severe waterlogging-induced yield reduction.

During the maize seedling stage (Figure 11a), the Y_cl-AHI correlation coefficients in various districts were a mix of negative and partly positive values; the only significant correlation was found to be negative. At the jointing stage (Figure 11b), the number of significant negative Y_cl-AHI correlations increased to six; moreover, the strength of the negative correlations increased. For the final mature stage (Figure 11c), the number of significant negative correlations increased to nine. In terms of the entire growth period (Figure 11d), most districts witnessed negative Y_cl-AHI correlations; northern MYLRR was dominated by negative correlations, and three of them were significant. In conclusion, the impacts that waterlogging had on maize yield occurred mainly from the jointing stage to the mature stage. In addition, maize yield was most affected by waterlogging disasters in the northern MLYRR, especially in the northern parts of Anhui and Jiangsu.

From the provincial scale, the Y_cl-AHI correlations for wheat and maize were also computed (

Table 4. Correlation analysis between waterlogging intensity and climatic yield of wheat and maize in three provinces in the study area from 1990 to 2020.

Province	Wheat Growth Stage					Maize Growth Stage
	Seedling	Jointing	Heading	Mature	Entire Stage	Seedling	Jointing	Mature	Entire Stage
Hubei	0.065	−0.205	−0.286	−0.111	−0.199	−0.206	0.006	0.064	−0.127
Anhui	0.066	−0.326	−0.387	−0.527 *	−0.463 *	0.082	−0.199	−0.204	−0.089
Jiangsu	−0.076	−0.122	−0.085	−0.278	−0.241	0.180	−0.398	−0.711 **	−0.421

Note: Bold fonts indicate that the climatic tendency rate of AHI was significant; * and ** indicate p < 0.05 and p < 0.01, respectively.

). For wheat, the Y_cl-AHI correlations were negative at all growth stages except during the seedling stage; in particular, the correlations in Anhui were significantly negative during the mature stage. In terms of the whole growth period, the Y_cl-AHI correlations were negative in all provinces; in particular, Anhui witnessed a significant waterlogging-induced wheat yield reduction, which was consistent with previous district-level conclusions. For maize, negative Y_cl-AHI correlations were more common than positive correlations, and the only significant correlation was negative (i.e., mature stage in Jiangsu). During the whole growth period, the Y_cl-AHI correlations were negative in all provinces.

According to the above findings concerning the AHI of wheat and maize, waterlogging disasters in the MLYRR posed great threats to the yield of both crops. In addition, the negative impact the waterlogging had on maize was more significant than that on wheat, as demonstrated by more significantly negative correlations and the strongly negative correlations (r < −0.5) of maize (Figure 10e and Figure 11d).

4. Discussion

4.1. Characteristics of Wheat and Maize Waterlogging Disasters in the MLYRR

According to our results, as shown in Figure 4e and Figure 5d, over the past 60 years, the waterlogging proneness of wheat and maize has generally increased in the central and eastern MLYRR (i.e., Anhui and Jiangsu). This is consistent with a previous report [39], which showed that extreme precipitation in Anhui has been more frequent over the past 50 years. In addition, the results shown in Figure 6 demonstrate that for both wheat and maize, waterlogging disasters occurred more frequently over the last three decades (since the 1990s) than in previous decades. This finding was consistent with many existing investigations into extreme precipitation in the MLYRR. Jiang et al. [40] found an upward trend in summer runoff and flood flow in the lower reaches of the Yangtze River over the past 40 years; Su et al. [41] studied extreme precipitation events in the MLYRR from 1960 to 2003 and indicated that the frequency of extreme precipitation events in the MLYRR increased significantly after 1990; and Zhang, et al. [19] found that the frequency and intensity of extreme precipitation events in the Yangtze River region increased during 1961–2002. In addition, Figure 6 indicates that the 1990s was the decade most prone to waterlogging for wheat and maize in the study region. Similarly, previous reports demonstrated that the frequency of heavy rain in the MLYRR increased significantly in the 1990s [18,30]; in 1998, the well-known catastrophic flood hit the whole Yangtze River region [40].

Our results concerning waterlogging proneness indicate that Anhui (especially in the southern part of the province) is a region with a very high risk for wheat waterlogging (Figure 8). This conclusion can be supported by many regional-scale analyses of wheat waterlogging disasters. Chen et al. [30] considered that spring precipitation in Anhui and Jiangsu (central and eastern MLYRR) had significant impacts on winter wheat yield, and waterlogging occurred most frequently in the southern MLYRR and along the Huaihe River in Anhui. Chen, et al. [20] analyzed the spatial characteristics of winter wheat waterlogging in Hubei, Anhui, and Jiangsu provinces and found that the high-risk waterlogging areas were concentrated in the southern part of the study region. Similarly, Hu [42] employed a crop damage index to describe wheat waterlogging in three provinces of the MLYRR and found that wheat waterlogging occurred more frequently in the southern parts of every province. In addition to the above conclusions regarding wheat waterlogging, the present work also revealed that the southern MLYRR, especially southern Anhui, was the region most highly prone to maize waterlogging (Figure 9). As explained by [42], for southern Anhui, the high likelihood of waterlogging is due to topographic changes in the local mountainous areas; additionally, warm moist air also helps form heavy precipitation [43].

4.2. Responses of Wheat and Maize Yields to Waterlogging Disasters

According to the results shown in Figure 10 and Figure 11, for both wheat and maize, waterlogging during the middle and late growth stages (i.e., after the seedling stage) had more significant negative effects on crop yield than early stage waterlogging. We found that a previous investigation obtained similar results: Gao et al. [13] employed the SPEI and SPI to analyze the effects of drought and flooding conditions on winter wheat yield in the Huaihe River basin of the MLYRR, and they found that waterlogging during the middle wheat growth period (i.e., jointing to heading stage) had greater impacts than waterlogging at other growth stages. In fact, current relevant regional-scale research scarcely considers the wheat yield response at different growth stages, but numerous field experiments including wheat waterlogging at different growth stages can help explain our findings. de San Celedonio et al. [24] experimentally found that the late growth stage (around the flowering stage) was the most waterlogging-sensitive period for wheat. Ding et al. [25] concluded that waterlogging that occurred during the middle growth stage (i.e., stem elongation) was more harmful to wheat yields than waterlogging that occurred at the late growth stage (anthesis or heading stage). Similarly, Ploschuk et al. [26] believed that for wheat, waterlogging during the late growth stage reduced the quality of roots, stems, and seeds more significantly than waterlogging during the early growth stage. However, for maize crops, current views in field experiments are more controversial. Huang et al. [29] and Huang et al. [28] conducted multiyear field experiments in central China and found that the negative effect waterlogging stress had on summer maize yield was greater during the middle growth stage than during the early and the later stages. However, Tian et al. [27] conducted spring maize field experiments in northeast China and found that spring maize was most sensitive to waterlogging stress during the early growth stage (seedling stage), followed by the middle and late growth stages. The above discrepancy may be attributed to the influence of maize varieties, as well as the influence of soil and climate environment.

The results of Figure 10d and Figure 11e suggest that in the MLYRR, maize yield was more significantly reduced by waterlogging events than wheat yield. In line with this conclusion, Cao et al. [44] studied the influence of main meteorological elements (e.g., precipitation and temperature) on crop yield in the MLYRR, and they found that excessive rainfall caused greater variability within maize yields than within wheat yields. In addition, in this work, the greater yield reduction of maize induced by waterlogging may be related to the greater intensity of maize waterlogging. Specifically, during different decades and different growth stages in the MLYRR, maize suffered more intense waterlogging events than wheat did (Figure 6 and Figure 7), and intense waterlogging events tended to induce more significant crop losses. Finally, in terms of different regions, the negative effect waterlogging had on wheat was found to be more significant in Anhui than in other provinces, as demonstrated by the Y_cl-AHI correlations (Figure 10 and Table 4). This conclusion can be supported by a previous rainfall-assessment report on wheat in the MLYRR [21]; in that report, the winter wheat yield in Anhui was found to be severely affected by extreme precipitation (yield decreased by 26.9%), and the yield reduction was the greatest in the MLYRR.

4.3. Perspectives for Waterlogging Disaster Risk Reduction

For wheat, the southern MLYRR was most likely to suffer waterlogging disasters (Figure 8e), while the southwestern and northeastern MLYRR had the greatest yield losses induced by waterlogging (Figure 10). Therefore, the southwest MLYRR is a key area for waterlogging prevention and control. By comparison, waterlogging in the northern MLYRR (especially northern Anhui) had great impacts on wheat yield (Figure 10) but was not frequently seen; thus, in this region, although wheat waterlogging was infrequent, its occurrence would lead to significant wheat yield losses. The features of this region have been explained by a previous waterlogging study in the MLYRR [20]. In that study, the authors considered that the areas that were highly sensitive to waterlogging disasters were generally distributed in flat areas where soils had a high proportion of clay, and they concluded that northern Anhui was more susceptible to waterlogging. Finally, there were some areas in which waterlogging disasters were frequent, but wheat yields were only minimally impacted by waterlogging, such as the southwestern part of the study area (Figure 8 and Figure 10). It is supposed that some other factors reduced the negative impact of waterlogging, such as highly efficient drainage. For maize in the MLYRR, the area most prone to waterlogging was the southern MLYRR (Figure 9d), while the areas with the greatest waterlogging-induced yield loss were distributed in the northern MLYRR (Figure 11), primarily in the northeastern part. It is noteworthy that in the northeastern MLYRR, although maize waterlogging was not intense, it could pose a substantial threat to maize yield once it occurs; this region was an insidious focus for waterlogging disaster prevention.

Among the different provinces in the MLYRR, field drainage in Anhui and Jiangsu provinces should be the focus of increased attention. First, these two provinces are highly productive wheat- and maize-producing areas (Figure 2). Second, waterlogging intensity in these two provinces has shown an increasing trend over the last 60 years, especially among maize waterlogging, which was significant in many places (Figure 4e and Figure 5d). Finally, waterlogging disasters during the wheat and maize growth stages in these two provinces had significant negative impacts on yields (Figure 10e and Figure 11d). These facts both determined the importance of wheat and maize drainage in Anhui and Jiangsu.

In terms of different crop growth stages, the middle and late growth stages (Figure 10 and Figure 11) were the periods most sensitive to waterlogging for both wheat and maize. More importantly, previous studies revealed that precipitation in the MLYRR showed significantly increasing trends in June–July [40,41] and January [40], which corresponded to the middle and late growth periods of wheat and maize, respectively. Therefore, more attention should be given to preventing the waterlogging disasters during the middle and late stages of wheat and maize. Finally, our results (Figure 6) demonstrated that wheat and maize waterlogging disasters in the MLYRR over the last three decades were stronger than before; additionally, the most recent decade, the 2010s, reached a historic high, second only to the 1990s. In this regard, current drainage scheduling in the MLYRR should account for the potential increasing risk that waterlogging poses to crops in the future.

Under future climate scenarios, crop losses due to waterlogging are expected to increase in many areas [45]. Our results regarding the temporal trend in wheat and maize waterlogging (Figure 4, Figure 5 and Figure 6 and Table 3) strongly imply that the waterlogging crisis in the MLRYR is expected to worsen in the future. Considering that different varieties of wheat [46,47] and maize [48,49] differ greatly in their resistance to waterlogging stress, more waterlogging-tolerant varieties of wheat and maize are suggested to cultivate in the MLRYR, especially in the areas witnessing significant yield-reducing impact of waterlogging (Figure 10 and Figure 11); this measure is considered beneficial for counteracting the risks arising from future waterlogging [1]. In addition, more precise agricultural management practices, such as adjusting the planting date of crops and appropriate nutrient management, can also help counteract the negative consequence of waterlogging [45,50].

5. Conclusions

In this work, an agriculture-specific index was introduced to characterize wheat and maize waterlogging during different growth stages in the MLYRR. Furthermore, the relationship between waterlogging intensity and climate-induced fluctuations in crop yield was revealed. The main findings are as follows:

(1) Over the past 60 years, the proneness of wheat and maize waterlogging disasters showed generally increasing trends in the central and eastern MLYRR; in particular, the proneness of maize waterlogging significantly increased at many stations. Wheat and maize waterlogging during the last three decades was more intensive than during previous decades; in addition, the 1990s was the historical high, followed by the most recent decade (2010s). Among the different provinces, Jiangsu suffered the least intense wheat and maize waterlogging.

(2) For both wheat and maize, waterlogging was more likely to occur in the southern MLYRR than in the northern MLYRR, especially in southern Anhui. The relationships between waterlogging intensity and the climatic yield of both wheat and maize were extensively negatively correlated across the study region; moreover, the observed significant relationships were almost all negative, which indicates the significant effect that waterlogging has on reducing crop yields.

(3) In the MLYRR, both wheat and maize were more likely to suffer from waterlogging during the early growth stage than during later stages. However, compared with the early growth stage, the middle and late stages (after the seedling stage) were the periods when the negative effects of waterlogging were significant. In addition, compared with wheat, maize was more prone to suffer from waterlogging conditions, and the negative effects waterlogging had on its yield were greater.

In summary, at the regional scale, scientific and rational assessment of the impacts that waterlogging has on crops is important to ensure high and stable agricultural yields as well as to take anti-flooding measures. The agrometeorological index used in this paper (AHI) can act as a powerful tool for monitoring and early warning of crop flooding, as well as identifying the key regions and periods highly affected by flooding, thus providing guidance for optimizing drainage schedules and cultivating flooding-tolerant crop varieties.