Impact of Agricultural Drought on Barley and Wheat Yield: A Comparative Case Study of Spain and Germany

Abstract

Given the growing interest in drought impacts on crops, this work studied the impact of agricultural drought on wheat and barley during the period 2001–2020. The study was carried out in the Spanish regions of Castilla y León and Castilla–La Mancha, with approximate areas of 94,000 km² and 79,000 km², respectively, and in the German regions of Nordrhein-Westfalen, Niedersachsen and Bayern, with approximate areas of 34,000 km², 48,000 km² and 71,000 km², respectively. These are the main cereal-growing regions of Spain and Germany. Soil moisture (SM) in the root zone was extracted from the LISFLOOD model database, and SM anomalies were used as the agricultural drought index. Gross primary productivity (GPP) and leaf area index (LAI) variables were obtained from the Moderate Resolution Imaging Spectroradiometer (MODIS), and the month in which SM is most influential on these crop state variables was identified. Crop yields in Spain and Germany were obtained from the Spanish Ministry of Agriculture, Fisheries and Food and the German Federal Statistical Office, respectively. Agricultural drought years and their impact on cereal yields were determined on a regional scale using three approaches based on the critical month with different time periods. These approaches were the use of the critical month and the two (before or after) and the three months (before and after) around the critical month. Two different analyses were used to identify the critical month, depending on the different environmental conditions in each country. These two approaches consisted of a monthly correlation analysis between SM anomalies and cereal yield in Spain and a monthly trend analysis of SM anomalies in Germany. The results showed a dependence of crop variables on SM in spring months in both countries and in summer months in Germany. Differences were found depending on the environmental conditions. A considerable reduction in cereal yields was obtained in Spain which exceeded 30%. Similarly, a worrying sign was observed in Germany, with a positive agricultural drought trend and a yield reduction of almost 5% in cereal crops. In view of future forecasts of the negative impact of climate change on global food production, this study provides valuable information for water and agricultural management under climate change scenarios. Both in regions that are already threatened and in those that until recently were not affected, it is necessary to study adaptation measures to avoid aggravating the impact of agricultural drought on crops, which could improve water productivity and future food security.

1. Introduction

Global warming has led to rising temperatures, causing environmental changes that have accelerated the water cycle, thus increasing extreme hydrological events and, therefore, reducing water availability and increasing water resource vulnerability [1,2]. Over much of the globe, as a result of climate change, drought has become one of the worst disasters [3], and its frequency and intensity are expected to increase [4,5], particularly in water-limited regions [6]. Droughts adversely affect crops, but the consequences vary according to plants, soils and regions [7]. Thus, drought is considered one of the major natural hazards with significant impacts on the environment, society, agriculture and the economy [8].

Drought is classified into four categories based on its nature: meteorological, hydrological, agricultural and socioeconomic drought [9]. Agricultural drought is a period in which the soil moisture (SM) supply is less than the minimum needs of plants, so the yields of crops and, therefore, their production are negatively affected [10,11]. Studies and assessments of agricultural drought are crucial, as it is considered the most serious concern in many countries from food security, social stability and economic perspectives [12,13].

Agriculture is the main land use type in Europe, which is an activity that requires a significant amount of water, an increasingly scarce resource [14,15]. In fact, SM drought risk is projected to increase in central western Europe and southern Europe under all climate scenarios, and the Mediterranean region is expected to be the most affected region, regardless of the type of drought [14,16,17]. This hazard is especially important, as the European Union (EU) is one of the largest cereal producers in the world, with wheat (Triticum) and barley (Hordeum vulgare) cereals standing out in terms of planted area [18]. As a consequence, net yield losses will reduce the economic output of agriculture in the EU [19].

This scenario is emphasized in rainfed agriculture owing to its increased vulnerability to climate anomalies [20]. Under rainfed conditions, important crops such as wheat and barley often suffer from droughts that cause significant yield losses [21]. Looking ahead to the next decades, it is believed that production levels will stagnate for a variety of reasons, but mainly due to climate change and adverse weather events [18]. Thus, in the face of a changing climate, it is crucial to analyze and understand the impact of climatic extremes on past and present crop yields to ensure and optimize yields. Thus, several studies [22,23,24] have analyzed the drought vulnerability of crops in Spain and the impact on crop yields and consider the need for further analysis to help unravel the climatic mechanisms influencing yield responses to climate in Spain. In addition, with the same approach, several studies have been conducted in Germany [25,26] suggesting the specifical analysis of individual drought years with respect to relevant variables and the performance of a monthly correlation analysis between drought indices and crop yields to determine possible seasonal focal points.

It is increasingly difficult to ignore SM as a key variable of the natural system [27]. SM availability is a nexus of the water and the energy and carbon cycles, as well as a primary process in the climate system [28,29]. SM drought has been shown to alter vegetation processes [30], and in many of them, plant water availability is responsible for this effect [31]. In fact, in agriculture, SM is an essential variable because its scarcity is an obstacle to proper plant productivity [32], reducing crop yields [33].

However, despite the importance and interest in agricultural impacts caused by soil moisture deficits, few studies have quantified the impact in terms of crop yields in each zone [34]. Many drought indices have been developed [35], but meteorological drought indices are usually used to assess the impact of drought on agricultural production [36,37]. When agricultural drought indicators are used, SM is usually not used as the primary variable but as a derived variable [38], despite SM being the variable by which agricultural drought is defined. This is so, even though SM has been shown to be the critical variable in the productivity of strategic crops under certain environmental conditions [39].

This study aims to analyze the impact of agricultural drought on wheat and barley crop yields in rainfed systems from different perspectives and under different environmental scenarios. The analysis was performed in the main cereal-growing regions of Spain and Germany, two of the most important countries for the production of these crops in Europe, during the 2001–2020 period. For this purpose, the monthly SM anomalies in the root zone, obtained from the LISFLOOD model database [40], were considered as the agricultural drought index. The study allows for a comparison of the effect of agricultural drought on wheat and barley crops in two areas with different climatic (water-limited vs. energy-limited) conditions over the last two decades. This work can help inform management decisions to face future scenarios, both in water-limited and in energy-limited regions, for these two key cereals.

2. Materials and Methods

2.1. Study Area

Spain and Germany were selected as study areas (Figure 1) for different reasons. The first reason is because several studies have observed significant drought trends in southern and central Europe [41,42]. Second, due to the importance of cereal grain production in Spain and Germany, they were among the top 5 producing countries in Europe in 2022 [43]. Finally, the last reason is due to the different characteristics hindering crop productivity which allow for a comparison of impacts under a wide range of conditions, from the limited water of Spain to the mainly limited energy of Germany [44,45].

In Spain, the regions of Castilla y León (CL) and Castilla–La Mancha (CM) were selected as the areas to be studied (Figure 1) since they are the main cereal-producing regions, accounting for approximately 60% of cereal production [46]. The CL and CM regions consist of 9 and 5 provinces, respectively, and are characterized by a semiarid Mediterranean climate, with cold winters and hot summers, average temperatures ranging between 10 °C and 15 °C and total annual rainfall ranging between 350 mm and 600 mm (Csb climate according to Köppen–Geiger classification [47]). The same approach was used in Germany, where the regions of Bayern (BY), Niedersachsen (NS) and Nordrhein-Westfalen (NW) were selected (Figure 1) as the regions with the highest cereal production, accounting for almost 41% of total production [48]. BY, NS and NW are composed of 7, 4 and 5 districts, respectively, characterized by a temperate continental climate, with cold winters and mild summers, average temperatures around 9 °C and total annual rainfall around 900 mm (Cfb and Dfb according to Köppen–Geiger classification [47]).

2.2. Irrigation and Cereal Cover Mask

To study the areas of rainfed wheat and barley crops, two databases were used to create a mask that filters out all the areas other than the target areas. To exclude irrigated areas, the Digital Global Map of Irrigation Areas of the Food and Agriculture Organization (FAO) was used, which represents the global area equipped with irrigation at a spatial resolution of 5 arc minutes or 0.083 decimal degrees [49]. Additionally, to discard areas with different land cover types, the Climate Change Initiative (CCI) Land Cover (LC) map from the European Space Agency (ESA) was used. It describes Earth’s land surface in 37 original LC classes based on the United Nations Land Cover Classification System (UN-LCCS) [50], with a spatial resolution of 300 m [51].

In this study, ArcGIS v10.8 software (ESRI^®, Redlands, CA, USA) was used to create the mask of irrigated areas, together with land cover maps. First, the irrigation and the land cover maps were resampled in the grids of the soil moisture and the GPP-LAI databases. Then, one mask for each database (soil moisture and GPP-LAI) was calculated by hiding all the pixels with more than 10% irrigated area and those with a land cover other than rainfed cropland, assuming that barley and wheat are the main cereals in those areas.

2.3. Soil Moisture Database

The hydrological rainfall-runoff model LISFLOOD developed by the floods group of the Natural Hazards Project of the Joint Research Centre (JRC) of the European Commission [52,53] was used as the SM database. It is utilized by the European Flood Alert System (EFAS) and the European Drought Observatory (EDO) for flood and drought monitoring, respectively [54,55]. The database has been validated [56] and satisfactorily used in many studies [57,58]. The model provided SM data with a spatial scale of 5 × 5 km and a daily temporal resolution, with data from 1991 to the present [59].

Moreover, it provides SM in three different depth layers in each pixel, but in this work, only the first two layers (0–100 cm) were selected. The SM value of the two layers was averaged to first obtain a daily series of root zone SM, and then a monthly series of root zone SM throughout the study period was obtained from 2001 to 2020.

2.4. Wheat and Barley Crop Data

Yield data for wheat and barley crops were obtained on an annual scale, for every province of the study area, from 2001 to 2020. Provincial and regional yield data for the Spanish regions CL and CM were obtained from the Statistical Yearbook of the Ministry of Agriculture [46], which provides provincial and regional grain yields (kg/ha) of wheat and barley in rainfed systems for the period 1904–2020. The crop yield data for the NW, NS and BY regions of Germany were obtained from the German Federal Statistical Office [60], which provides winter wheat and barley grain yields for administrative districts and federal states from 1999 to 2022.

It was assumed that there was no need to eliminate the trend in the crop yield series because the increase in yields due to technological improvements mainly occurred in the 20th century [24,61], prior to the study period.

The gross primary production (GPP) and leaf area index (LAI) were also used, as they are two important indicators of vegetation growth and biomass and, therefore, have become essential for studying vegetation and climate change interactions [62,63]. The GPP describes the amount of carbon dioxide fixed by plants through photosynthesis, a key component of the terrestrial carbon cycle [64]. The LAI quantifies leaf area in an ecosystem and is therefore a fundamental variable in processes such as respiration, rainfall interception and photosynthesis [63]. Thus, both variables were obtained from the Moderate Resolution Imaging Spectroradiometer (MODIS). The product MCD15A2H was used to obtain the annual LAI, and the products MOD17A2H and MYD17A2H were used to obtain the annual GPP. The MCD15A2H product version 6 derived from the MODIS Terra/Aqua combined product has a temporal resolution of eight days and a spatial resolution of 500 m [65]. The MOD17A2H and MYD17A2H version 6 products are derived from the MODIS sensors onboard NASA’s Terra and Aqua satellites, respectively, with a temporal resolution of eight days and a spatial resolution of 500 m [66,67].

In this study, only the months of the phenological cycle of wheat and barley characteristic of each region for each year were considered. For this purpose, the sowing and harvesting months for wheat and barley crops in the regions of Spain were extracted from the sowing, harvesting and marketing calendar [68], which provides data at the provincial and regional scales. For the choice of dates, those corresponding to semihard and soft wheat (Triticum aestivum) and malting barley (two races, Hordeum distichum) were considered since they are predominant in CL and CM [46]. In addition, the calendar provides for each month of the year a percentage of sowing and harvesting occurrence. Thus, the months with the highest percentage of sowing and harvesting were selected to establish the beginning and end of the phenological cycle. Due to the high degree of similarity between the provincial and regional phenological cycles in each study region, the regional phenological cycle of each crop was considered for both scales. Thus, the phenological cycle of barley in CL and CM is from November to July, and the phenological cycle of wheat is from October to July in CL and from November to July in CM.

In the federal states of Germany, the sowing and harvesting months of wheat and barley crops were extracted from the phenological database of Germany’s national weather service [69]. It provides the dates of the sowing, emergence, earing, grain filling, milky ripening and harvesting phases, but in this work, only the dates of the sowing and harvesting phases were considered. For the choice of dates, an average of the available dates in the study period was made, and the resulting month was selected. In addition, as in the Spanish regions, the regional phenological cycle of each crop was considered for each region under study and its provinces. Thus, in BY, NS and NW, the phenological cycle of barley is from September to July, and that of wheat is from October to August.

2.5. Agricultural Drought Index: Soil Moisture Anomalies

SM anomalies have been successfully used in many works [70,71,72] and have been proven to be a good indicator of agricultural drought [73,74]. In the present study, monthly SM anomalies at the provincial scale in the root zone were used to analyze the impact of agricultural drought on cereal yields. For this purpose, first, a provincial-scale spatial average of SM in the selected cereal areas was obtained. Then, provincial-scale SM anomalies were calculated as follows:

$$\mathrm{SM}\mathrm{anomalies}={\mathrm{SM}}_{\mathrm{t}}-\overline{SM}/{\mathrm{\delta }}_{\mathrm{SM}},$$

(1)

where SM_t is the monthly SM series for year t, $\overline{SM}$ is the monthly average using the entire study period and δ_SM is the standard deviation of SM in the study period at the provincial scale [70,71,72].

2.6. Analysis of Biophysical Indicators: GPP and LAI

The GPP and LAI data series were used to identify the state of cereals as a whole since it was not possible to characterize wheat and barley crops separately. The phenological cycle considered was that in which the months of the wheat and barley phenological stages coincided. Although differentiation at the crop scale was not feasible, the study focuses mainly on wheat and barley cereal crops, which are the main rainfed crops in the study regions.

For the study of the month in which the two biophysical parameters of cereals are most affected by agricultural drought, a Pearson correlation analysis was performed between the cereal GPP and LAI and the agricultural drought index using MATLAB v.R2023b software (MathWorks^®, Natick, MA, USA). The analysis was performed at the provincial scale from 2001 to 2020, thus obtaining a monthly correlation coefficient (R) value during the phenological cycle of the cereal. In each province, the month with the highest absolute R value was considered the critical month. To obtain the critical month for each region, the most frequent critical month at the provincial scale was selected.

2.7. Identification of Agricultural Drought Years

The countries studied in the present work have different environmental conditions. Specifically, Spain is a territory with mainly water-limited conditions, whereas Germany mainly has energy-limited conditions. Several studies [75,76] have analyzed drought by differentiating zones according to the predominant limiting variable, thus dividing the territory between water-limited and energy-limited zones. Therefore, due to the different environmental conditions and limiting factors, a different procedure was followed in each country for the identification of agricultural drought years.

2.7.1. Spain

In Spain, the critical month for crop yield was determined by performing a Pearson correlation analysis between wheat and barley yields and the agricultural drought index from 2001 to 2020, thus obtaining a monthly R value during the phenological cycle of wheat and barley. The analysis was performed at the provincial level, and subsequently, the most frequent critical month in each region was extracted, considering the month with the highest correlation as critical.

Once the critical month for CL and CM was obtained, it was used to identify the drought years. One of the most commonly used methods for drought detection is based on the definition of a threshold level below which drought is said to have occurred [77]. This approach was applied in the present work, using a percentile as a threshold. A percentile is defined as the value that divides a linearly ordered dataset, so that it indicates the value below which a percentage of the dataset is equal to or less than that value [78]. A common value adopted in the literature to detect agricultural drought is the 20th percentile [79,80,81]. Accordingly, the 20th percentile was used as the threshold level, below which the onset of agricultural drought was defined.

In addition, three approaches were studied for wheat and barley with different time periods as the criteria to identify drought years. The critical month (M) and two (2M, the critical month plus the next or the previous month) or three (3M, the critical month plus the previous and next months) months around the critical month were used. Thus, for the identification of drought years for wheat and barley crops in each Spanish region and for each year of the study period, the threshold level was applied to the three approaches.

2.7.2. Germany

Germany is a temperate country in which agriculture is mainly energy-limited. However, environmental and climate conditions are changing, and positive drought trends have recently been observed in many regions of central Europe [41,82]. Consequently, for the selection of the critical month for the yield variable and, therefore, the detection of drought years, a provincial and monthly scale analysis of agricultural drought index trends was performed. The Mann–Kendall (MK) statistical test is a nonparametric test used to identify trends in time-series data [83,84]. It is widely used to detect whether there are statistically significant increasing or decreasing trends in hydrometeorological time series [85,86]. The ability of this test to detect trends in hydrology studies has been demonstrated [87], and it has been applied in several agricultural drought studies [73,88,89]. Many studies have reported that correlations in the time series may affect the results of the MK test [90,91]; nevertheless, in this study, the anomalies of the SM series were studied, which, according to [92], is a technique for avoiding these inconveniences. In this study, the MK test was used to detect whether there were statistically significant increasing or decreasing trends in the agricultural drought index during 2001–2020 to identify the critical month (month in which statistical significance predominates) of administrative districts. Under the null hypothesis of no trend Ho, the MK test statistic (S) was calculated as follows [83,84]:

$$S=\sum _{i=1}^{n-1}\sum _{j=i+1}^{n}sgn(x_j-x_i)$$

(2)

where

$$\mathit{sgn}(x)=\left\{\begin{array}{cc}1& ifx>0\\ 0& ifx=0\\ -1& ifx<0\end{array}\right.$$

(3)

For an upward trend, the S statistic is increased by +1, while it is decreased by −1 for a downward trend. The S statistics remain unchanged for a zero difference. The statistical parameter Z allows us to determine whether a trend is significant:

$$Z=\left\{\begin{array}{ll}\frac{S-1}{\sqrt{var\left(S\right)}}& ifS>0\\ 0& ifS=0\\ \frac{S+1}{\sqrt{var\left(S\right)}}& ifS>0\end{array}\right.$$

(4)

Throughout this study, a p value of 0.05 (confidence level of 95%) was used as the criterion for the statistical significance of a trend. Thus, for an absolute value of Z greater than 1.96, a significant trend was considered. The month with the most districts with statistically significant trends was considered the critical month for all German regions.

Once the critical month for the yield variable in the German regions was obtained, the years with agricultural drought were identified using the same procedure explained above for the Spanish regions. Thus, for the three approaches studied (M, 2M and 3M), the 20th percentile was used as the threshold level, below which years with agricultural drought were defined.

2.8. Yield Reduction Calculation

Once the years with agricultural drought were identified, the percentage reduction in wheat and barley yields was calculated for those years in each region and for the three criteria. A normal year for crop yield is considered when the drought index is between the 40th and 60th percentiles [93], and the average yield in normal years is estimated and used as a reference. In years classified as agricultural drought years, the annual yield reduction rate is calculated as follows:

$${\mathrm{Y}\mathrm{i}\mathrm{e}\mathrm{l}\mathrm{d}}_{\mathrm{r}\mathrm{e}\mathrm{f}}=\frac{{\mathrm{Y}\mathrm{i}\mathrm{e}\mathrm{l}\mathrm{d}}_{\mathrm{n}1}+{\mathrm{Y}\mathrm{i}\mathrm{e}\mathrm{l}\mathrm{d}}_{\mathrm{n}2}+\cdots +{\mathrm{Y}\mathrm{i}\mathrm{e}\mathrm{l}\mathrm{d}}_{\mathrm{n}\mathrm{x}}}{\mathrm{x}}$$

(5)

$${\mathrm{Y}\mathrm{R}}_{\mathrm{m}}=\frac{{\mathrm{Y}\mathrm{i}\mathrm{e}\mathrm{l}\mathrm{d}}_{\mathrm{r}\mathrm{e}\mathrm{f}}-{\mathrm{Y}\mathrm{i}\mathrm{e}\mathrm{l}\mathrm{d}}_{\mathrm{d}\mathrm{m}}}{{\mathrm{Y}\mathrm{i}\mathrm{e}\mathrm{l}\mathrm{d}}_{\mathrm{r}\mathrm{e}\mathrm{f}}}\times 100\%$$

(6)

where Yield_ref represents the benchmark yield of the selected region, x represents the number of normal years in the study period and Yield_nx (x = 1, 2, …, 4) represents the yield of year x. YR_m is the yield reduction rate of the selected region in the m^th year (m = 1, 2, …, 20), and Yield_dm (m = 1, 2, …, 20) represents the yield in year m identified as dry [34].

3. Results and Discussion

3.1. Biophysical Variables versus Agricultural Drought

The average R values for the relationships of the agricultural drought index with GPP and the LAI for the predominant month are given in

Table 1. Predominant month and mean R of the predominant month for each study region (Castilla y León, CL; Castilla–La Mancha, CM; Bayern, BY; Niedersachsen, NS and Nordrhein-Westfalen, NW) resulting from the correlation analysis between biophysical parameters and the agricultural drought index.

Region	Predominant Month		Average R
	GPP	LAI	GPP	LAI
CL	May	May	0.50	0.69
CM	April	April	0.29	0.69
BY	May	July	−0.57	0.16
NS	June	July	−0.44	0.57
NW	May	July	−0.48	0.49

. The results of the correlation analysis between GPP and the agricultural drought index (Table 1) show that the months of the reproductive stage (April, May and June) of cereals are predominant in the provinces of all the regions considered. Moreover, the average R value (resulting from averaging the R values of the critical month of the provinces belonging to the region studied) shows a clear difference between the Spanish and German regions, presenting a direct relationship between SM anomalies and GPP in the Spanish regions (0.50 in CL and 0.29 in CM) and an inverse relationship in the German regions (−0.57, −0.44 and −0.48 in BY, NS and NW, respectively). These results are consistent with those of [94], which show that in times of drought, in areas where soil water content is high, an increase in plant GPP is promoted, as occurred in this study in the regions of Germany. Thus, moderate drying of wet soil causes an increase in the carboxylation capacity of plants [94]. Several studies have reported improved vegetation conditions during drought in wet regions [95,96]. In contrast, in areas where the water content is below a threshold, GPP decreases under drought conditions, as is the case in regions of Spain. Vegetation in arid regions reacts quickly to SM deficiency, which would imply greater drought stress in vegetation, as this is a limiting factor of vegetation functioning in water-limited environments [97].

Regarding the relationship between the LAI and agricultural drought index (Table 1), it is observed that in Spanish regions, the predominant months correspond to the reproductive stage of the plant, as in the results obtained with GPP. In the German regions, July is the predominant month, and the mean R values (0.16, 0.57 and 0.49 in BY, NS and NW, respectively) are lower than those in the Spanish regions (0.69 in both regions, CL and CM). Both cereal variables, GPP and the LAI, show different behavior in the regions of the two countries, in accordance with the fact that one has water-limited and the other energy-limited conditions. In BY, NS and NW, there is an inverse relationship between SM anomalies and GPP and a direct relationship with the LAI. Conversely, in CM and CL, there is a direct relationship in both cases. The results are consistent with those obtained by the authors of [98], who demonstrated decoupling between the LAI and GPP as aridity decreases. Hence, areas in water-limited environments, such as CM and CL, showed a stronger link between GPP and the LAI. In contrast, in the BY, NS and NW regions, under energy-limited conditions, a decreased LAI in drought years could facilitate carbon uptake by smaller leaves and consequently enhance GPP [98].

3.2. Critical Month Identification

In Spain, the critical month for cereal crops in CL and CM (Figure 2) is predominantly May, except for wheat in CM, where it is April. In both regions and for both crops, the critical month is in the spring phenological stages, i.e., during the reproductive and maturation phases of the crop. The R values are within the range of 0.60 to 0.80, which shows that SM is a fundamental variable for the development of cereals. These results are also consistent with those of [99], which observed that in the main cereal-growing areas of Spain, the critical periods of impact on the wheat and barley yields of SM were concentrated in the spring phases. Several studies [100,101] have shown that grain yields are strongly affected by water stress during anthesis in May. In addition, [102] observed that the dry conditions in those months in Spain led to a marked decrease in yields.

Analyzing the agricultural drought trends for each month of the year in all districts of the study regions in Germany (Figure 3), a generalized negative trend is observed in all months and areas. This result is consistent with that reported in the study [41], in which the authors observed a trend toward drier conditions in central Europe over the last three decades. The month with the maximum number of districts with a significant trend was April, followed by May. Thus, April was selected as a critical month in the regions of Germany to identify drought years. Several studies [103,104] prove that for winter cereals that dominate agricultural production in central Europe, the development stage in early spring is crucial for determining grain yield, as the cereals are very susceptible to drought in that period. Thus, [105] studied the effect of water stress on winter wheat production in central Europe and found that during the month of April, the crop is very sensitive to water stress, causing low yields.

3.3. Agricultural Drought Year Detection

For the identification of drought years, the 20th percentile of the agricultural drought index was calculated, using it as the threshold below which years were considered dry years (Figure 4). As a general result, in all regions except CM, dry years were coincident in wheat and barley crops since the critical month used for drought year selection was the same for both crops. Of the three approaches applied (M, 2M and 3M) for wheat and barley, in the case of 2M, April and May were detected in all regions. Furthermore, for the three temporal approaches, very similar patterns were observed for each crop and region. Hence, the three criteria used mostly adequately identify the predominant dry years in CM and CL, as well as in BY, NS and NW. In the case of German regions, the years with agricultural drought were grouped in the last decade of study, in accordance with the soil moisture trend observed in this study and others [106]. Therefore, it could be said that the three criteria used are equally valid for the identification of agricultural drought years.

In the CL region, the dry years identified (2005, 2009, 2017 and 2019) coincide for the three approaches used. In the CM region, there is variation in dry years in the three periods as well as between the two crops. Thus, the years 2005, 2012 and 2017 coincide mostly for both crops and time periods in CM. The years 2006 and 2015 were especially dry for wheat and barley crops, respectively. The drought episodes identified in CL and CM have also been detected and studied by other authors, but mostly from a meteorological drought perspective. The study [107] describes that 2005 was characterized by extremely dry conditions on the Iberian Peninsula, which had a significant impact on cereal production, decreasing it by 60% on average. The 2015 event first appeared in early spring (May and earlier) in southern France and on the Iberian Peninsula [108,109]. Several studies have identified 2012 as a year with drought [110,111]. Ref. [112] analyzed the drought that affected Europe from July 2016 to June 2017. They found that during that period, most of Western Europe suffered a major drought event resulting in severely affected crops, especially cereals in Spain. Ref. [113] indicates that Spain suffered major droughts in 2005, 2012 and 2017. Moreover, the consequences of drought events for vegetation, such as those observed in 2015, 2018 or 2019 in most parts of Europe, are productivity losses [114].

In the German regions, a cluster of dry years was observed in the second decade of study. Almost 90% of the dry years identified in the German regions occurred from 2010 onward, with a clear incidence in 2011, 2012 and 2020, showing a clear climatic trend, which led to an increase in the frequency of droughts. These results are consistent with those obtained by the authors of [115], who reported that droughts identified in southwestern Germany from 1801 to 2018 based on meteorological indices were distributed over the study period, but in the last decade, drought episodes showed greater severity. Ref. [116] also classified the period from 2003 onward as the one with the highest incidence of severe drought events in southwestern Germany. Ref. [117] argued that the most abrupt change in increasing warm season droughts in Europe has been observed in central Europe in the last 15 years, with dry events identified in agreement with those of the present study.

Agriculture is facing changing climatic conditions, and severe droughts are expected to increase in the coming years [3,118]. The results obtained show an increase in droughts in the most recent decade of study in the German regions. Despite the apparent lower dependence on irrigation in temperate regions, the impacts of drought on agriculture in these areas are considered a major abiotic stress [119]. One of the actions aimed at reducing drought risks is increasing water supply [120,121]. Irrigation can greatly mitigate adverse impacts resulting from water stress by maintaining higher SM requirements [122]. This fact becomes even more evident and corroborates the evolution of the agricultural drought years identified in this study when analyzing the evolution of the irrigated area in German regions. According to data obtained from the German Federal Statistical Office [60], from 2010 to 2020, the irrigated area increased by 86%, 27% and 86% in BY, NS and NW, respectively. Therefore, there are worrying signs of increasing water stress in areas where this variable was not previously a limiting factor.

3.4. Impact of Agricultural Droughts on Grain Yield

According to the identification of dry years in each region and for wheat and barley crops, the yield reduction for these years was calculated by averaging the 4 years identified in each region for each of the three approaches used (

Table 2. Average percentage of yield reduction in drought years in Castilla y León (CL) and Castilla–La Mancha (CM) for wheat and barley and for the three criteria (M, 2M and 3M).

Month Period	CL		CM
	Wheat	Barley	Wheat	Barley
M	32.8	28.8	28.5	33.7
2M	34.9	32.2	31.5	33.9
3M	34.9	32.2	8.4	34.8

and

Table 3. Average percentage of yield reduction in drought years in Bayern (BY), Nordrhein-Westfalen (NW) and Niedersachsen (NS) for wheat and barley and for the three criteria (M, 2M and 3M).

Month Period	BY		NS		NW
	Wheat	Barley	Wheat	Barley	Wheat	Barley
M	−11.57	−11.38	1.31	3.69	2.85	−0.33
2M	−0.09	4.78	2.90	−0.79	1.88	4.41
3M	0.82	−2.49	11.80	16.18	−4.62	1.35

).

In drought years, the Spanish regions of CL and CM (Table 2) suffered a significant decrease in crop yields, which was evident in all cases. The average yield reduction in both crops was greater than 30%, which demonstrates the importance of soil water in the root zone for wheat and barley yields [39]. These results are in agreement with the evidence obtained by the authors of [123], confirming that in semiarid agricultural areas, the most important limiting factor for crop productivity is water stress caused by reduced soil water availability and, therefore, agricultural drought. Comparing the two regions, a greater reduction is observed for wheat in CL and for barley in CM. However, the difference in crop reduction in both regions is small, usually less than 5%.

There is an interesting debate as to whether wheat or barley is better adapted to drought. Some authors [124,125] found better adaptation of barley to drought conditions, displaying a better yield potential than wheat in the Mediterranean basin. In contrast, several studies [126,127] did not find consistent differences between the yields of the two crops. In general, in the present work, in both regions and crops, a similar yield reduction of between 29% and 35% predominates. These reductions are larger than those obtained in the projection of [128] for winter wheat in Spain, with a predicted average yield decrease of 21% by the end of the 21st century.

These results highlight the susceptibility of rainfed crops to the increasing intensity of soil drought and warn of the severe impacts on crop yields that are already occurring. It is well known that the Mediterranean area is among the most vulnerable regions to climate change, and this is likely to worsen in the future [129]. Indeed, climate models project a decrease in SM and an increase in the duration and intensity of droughts in the Mediterranean region [3,17]. Therefore, this observed reduction in crop yields could be exacerbated in the future, and in the face of growing global food needs, this finding is of great concern and may aggravate the global food crisis [130,131].

In the case of Germany, there is notable variability in cereal yield variations in all three regions (Table 3). In BY, an increase in yield was observed, reaching 11% when the critical month approach was applied for both crops. In contrast, the NS and NW regions showed mostly yield reductions. In NS, the average yield reduction was 5% for wheat and 10% for barley, reaching 11% and 16%, respectively, with the 3M approach. The NW region showed smaller values than NS, with an average yield reduction of 2% for wheat and 3% for barley.

Yield increases in years of agricultural drought identified in BY can be explained by drought periods also being characterized by increased energy availability [96]. In this way, the main limiting variable for crops in this area increases, favoring the conditions for crop development and, therefore, yield. Furthermore, in the same context, this result can also be explained by consequences resulting from climate change, which have already been seen to influence crop yields in Europe, affecting them differently according to region and a variety of other factors [132]. Thus, due to global warming, energy factors such as the average annual temperature in Europe in the period 2006–2015 increased by approximately 1.52 °C [133] and were found to cause crop yields to increase, except in southern Europe [134]. This is consistent with the slight increase in crop yield obtained in dry years in BY, where the yield increase implied by an increase in energy variables probably outweighs the yield reduction caused by agricultural drought. However, some regions with little or small positive impacts of climate change could reverse this circumstance, leading to more severe drought situations [135,136]. This idea is further reinforced as climate change is projected to worsen in the near future [137].

In northern Germany, there are other factors that could explain the results obtained for NS and NW. In those regions, the soils are mostly sandy [138] and therefore have a low water-holding capacity. The sand fraction has an inverse relationship with important soil water properties such as water-holding capacity [139]. This circumstance increases the susceptibility to soil drought in the regions of NS and NW, located in northwestern Germany, where decreases in wheat and barley yields have been reported. Therefore, this finding warns of the deleterious consequences already occurring for the yield of the two main exported cereals from Europe [140] in areas where SM was not previously a limiting factor for crop development. Ref. [141] found several years ago that those agricultural regions of northern Germany could lose productivity if they are subjected to more frequent and long-lasting droughts without irrigation.

Evidence of all this was observed in 2011, a year of particularly intense agricultural drought in the German regions (Figure 4), causing an average yield reduction of more than 9% in cereals (

Table 4. Average percentage of yield reduction in the drought year 2011 in Bayern (BY), Nordrhein-Westfalen (NW) and Niedersachsen (NS) for wheat and barley and for the three criteria used (M, 2M and 3M). ND means not a drought year.

Month Period	BY		NS		NW
	Wheat	Barley	Wheat	Barley	Wheat	Barley
M	ND	ND	4.67	12.02	6.16	11.87
2M	3.26	12.30	5.60	10.36	2.70	12.78
3M	5.75	12.10	10.63	21.57	0.96	11.59

). This result is consistent with that of [142], which classified the 2011 drought in the EU as having the largest spatial extent on record. In the same context, [115] studied droughts using meteorological indices for southwestern Germany and classified 2011 as one of the most extreme years with precipitation shortages accompanied by high temperatures. Thus, 2011 was identified as a dry year in Germany due to a precipitation deficit in early April, ending the month with a soil moisture deficit and leading to vegetation stress in mid-May and consequent reductions in crop yields, especially for cereals [40]. Consistent with the aforementioned research, the present study finds that in 2011, identified as an agricultural drought year in Germany, both cereals showed a marked reduction in yield for all regions and approaches (Table 4). Barley showed an average yield reduction of 13%, which was much larger than that of wheat, with an average reduction of 5% (Table 4). Therefore, this shows that in areas where, until now, soil water content was not a limiting factor, a soil moisture deficit in April is promoting a reduction in cereal yields, with barley being more vulnerable than wheat.

Several studies have shown that in recent decades, there has been a general trend of change toward drier conditions, mainly in central and eastern Europe [41,102]. Some studies projected a progressive increase in climate threats in Europe, mainly driven by a trend toward more likely and severe extreme soil drought events in central Europe under future warming scenarios [16,42].

The detrimental effect observed for wheat and barley yields in Spain due to the decrease in SM may be an indication of what may happen to areas where soil water content was not previously a crop constraint. In Germany, an area where SM was not historically a limiting factor, an observed and significant reduction trend in SM reported in recent years is starting to cause reductions in crop yields that, according to future projections, could increase. In fact, this likely hypothesis is reinforced by a recent study [143] conducted for the period 1980–2100, in which a widespread shift in the ecosystem from energy- to water-limited conditions was found, which the authors attributed to global warming. This finding warns against a possible increase in the frequency of agricultural drought events and consequent yield loss in cereals, such as that already detected in this study for Germany in the last decade.

4. Conclusions

The impact of agricultural drought on wheat and barley crops in the main cereal-growing regions of Spain and Germany during the 2001–2020 period was analyzed. In general, the results of all the analyses carried out in this study show the importance of SM in cereals, especially in the months of the reproductive and ripening phases of the crops. Moreover, with respect to the three approaches studied, the results obtained showed similar patterns. Therefore, it can be stated that the three criteria were useful for the identification of drought years and for the calculation of yield loss.

The analysis of the GPP and LAI variables revealed similar behaviors in water-limited environments and the inverse in energy-limited environments, showing the influence of soil water content on the development of these variables. The use of SM anomalies evidenced their suitability for the identification of years with agricultural drought, in agreement with previous findings. Furthermore, a cluster of drought years was observed in the second decade of study in Germany, which is a clear indicator of warning in terms of SM deficit in areas where this variable is not the main constraint.

As expected, in water-constrained areas, a considerable reduction in crop yields has been observed, exceeding 30% yield reduction, because of agricultural drought. Unexpectedly, in regions located in energy-constrained areas, where water is not the main limiting factor, a singular yield reduction of around 5% has been observed for both crops. Thus, the results suggest that, in the face of increased droughts, the worsening of cereal yield losses is likely, which will increase notably in southern regions, as is happening in Spain. Similarly, Germany, where the SM was not considered a constraint until now, is starting to show increasing water stress that may lead to unprecedented reductions in cereal yields.

This finding has important implications for agriculture, as it was proven that in recent years, in areas where water was not the main limiting factor, it is now having detrimental effects on crop yields. Although the yield decrease detected in the German regions shows moderate values, it is an indicator of change in environmental conditions and will have negative consequences on agricultural production. Indeed, in view of climate change impacts, the results obtained warn against a new uncertain scenario for crop yields. This likely transformation suggests that in areas where SM has not yet been a limiting factor, the situation can reverse and lead to a significant decrease in cereal yields. This possible scenario could be similar to that currently being observed in Spanish regions, where yields have declined by more than 30% in drought years over the last two decades due to soil water deficits. The results of this study are of interest for adaptation and/or mitigation strategies to cope with these detrimental impacts on crop yields due to agricultural drought.