Exploring Optimal Cropping System to Improve the Water Use Efﬁciency and Soil Water Restoration after Lucerne-to-Crop Conversion in the Semiarid Environment

: Due to depleting water supplies and the cultivation of high water-demanding crops such as lucerne, the effect of water deﬁcits in crop production has become a major concern, especially in semiarid regions of China. A six-year ﬁeld experiment (


Introduction
Dryland covers a substantial portion of China's northwest, accounting for roughly 56 percent of the country's overall land area [1,2].This region has a loess topography and a large semiarid area, with average annual precipitation ranging from 300 to 600 mm, as well as extreme climatic conditions, such as periodic spring droughts and severe wind and water erosion [3][4][5].Limited and erratic rainfall is the major water resource for agricultural production.As a result, water scarcity is the most significant constraint to agricultural production in this region [3,6,7].The Chinese government began a program in 1998 to convert agriculture on steep slopes to forest and perennial plants, with the ultimate goal of reducing soil erosion [6,8,9].Lucerne (Medicago sativa L.) is one of the most productive forage plants as a result of its high herbage yield and protein concentration, tolerance Agronomy 2022, 12,1905 2 of 12 to drought and cold, and ability to protect against soil erosion.Lucerne is regarded as the preferred species for returning farmland to perennial vegetation in the hilly and gully region of the Loess Plateau [6,[8][9][10].Especially in the last decade, with the rapid development of animal husbandry, the demand for forage has increased significantly, and lucerne production has increased gradually in Northwest China.
Lucerne has a high evapotranspiration rate and water consumption compared to other crops [8,11].Long-term lucerne production depletes soil phosphorus (P) and creates severe soil dryness in deep soil layers, limiting land productivity in arid climates [9][10][11][12].Lucerne can increase soil water demand by at least 50 mm throughout the growing season when compared to other crops [13,14].When the age of lucerne stands reaches ten years, the desiccation of the 0-6 m soil layer reaches an excessive level, and the restoration of soil water is difficult, according to a study [15].Thus, in order to recover soil water in dry soil layers and maintain land productivity, the development of economical cropping systems with higher water use efficiency (WUE) for this semiarid region is required.
In dryland agroecosystems in North America, the adoption of a forage-grain crop rotation dramatically boosted crop productivity by conserving soil water and nutrients [16].In Australia, leguminous forage-crop rotation has permitted the healthy development of agroecosystems [17].A study in northwestern China demonstrated that lucerne-crop systems recover soil water in dry soil layers, and that crop yields in the lucerne-crop systems were comparable to those in a conventional system with fertilizer application, despite the fact that no fertilizer was applied [18].After several years of lucerne production, this evidence suggests that lucerne crop rotation can restore soil water and maintain land productivity [10,19].However, few studies have focused on soil water recovery, soil fertility, and economic return after lucerne-to-crop conversion.Knowledge of the soil water recovery response to the conversion of lucerne to crops in China, and how this is affected by crop species, is paramount for sustainable agricultural production.Such knowledge could serve as a basis for lucerne-crop rotation in semiarid environments.As a result, the goal of this research was to look at soil water recovery and economic returns when lucerne was converted to crops.

Description of the Study Site
From October 2012 to October 2018, a long-term field experiment was undertaken in Dingxi City (35 • 28 N, 104 • 44 E), Gansu Province, China.With an average yearly radiation of 593 J cm −2 , annual sunshine hours of 2477 h, annual mean air temperature of 6.4 • C, maximum air temperature of 38 • C (July), and lowest air temperature of 22 • C, the experimental location has a medium temperate semiarid climate (January).The average annual pan evaporation is 1531 mm [20] and long-term average annual rainfall is 390 mm.The top 0−20 cm soil had 0.256 cm 3 cm −3 field capacity, 0.085 cm 3 cm −3 wilting point, 1.20 g cm −3 bulk density, 10.3 g kg −1 organic carbon, 8.4 pH, 1.1 g kg −1 total nitrogen (N), 1.9 mg kg −1 available P, and 230 mg kg −1 available potassium.

Experimental Design and Field Management
A nine-year-old lucerne-producing field was divided into two halves in April 2012: one was plowed, and the other was kept under lucerne.The plowed field was divided into five sections to plant selected crops in the following sequences: (1) lucerne-fallow (L-F); (2) lucerne-wheat (L-W); (3) lucerne-corn (L-C); (4) lucerne-potato (L-P); (5) lucerne-millet (L-M) (Table 1).In this study, each treatment was replicated three times, and each plot was 3 m wide × 4 m long.In this study, all field management practices for each crop (i.e., tillage, planting mode, fertilization) were operated according to the local farming practices in the area.Spring wheat (cv.Dingxi 42) and millet (cv.Longgu 11) were sown with a drill (6 rows spaced 20 cm apart) at a density of 470 and 30 plants m −2 ; corn (cv.Xianyu 335) was sown at a density of 52,500 plants ha −1 in a ridge-furrow system with fully plastic film annual mulching [21]; potato (cv.Xindaping) was sown at a density of 52,500 plants ha −1 with a within-row plant spacing of 45 cm without mulching on flat plot.Lucerne was only fertilized at a rate of 45.8 kg P ha −1 as triple superphosphate and 105 kg N ha −1 as urea at planting in 2003.Wheat, potato, and millet received 45.8 kg P ha −1 as triple superphosphate and 105 kg N ha −1 as urea, and maize received 45.8 kg P ha −1 as triple superphosphate and 200 kg N ha −1 as urea.Fertilizers were evenly broadcast on the soil surface and incorporated into the soil with tillage before sowing.Corn, millet, and potato were sown in early May and harvested in early October every year, while spring wheat was sown in late March and harvested in early August every year.All crop residues were removed from plots during harvest.All plots were plowed twice per year except for lucerne, once before sowing and again after harvest.Weeds were hand removed when sighted throughout the growing period.

Measurements and Data Analysis
Meteorological data was obtained from an automatic weather station located at the experimental site.Soil moisture was measured in October of each year, three soil samples of each plot from the 0-10 and 10-20 cm soil depth layers were randomly collected using an auger (4.5 cm inner diameter) to determine soil moisture.A Trime-Pico IPH was used to measure volumetric soil water contents from the 30-50, 50-80, 80-110, 110-140, 140-170, 170-200, 200-250, and 250-300 cm soil layers (Precise Soil Moisture Measurement, IMKO Micromodul technik GmbH, Ettlingen, Germany).No irrigation was applied to crops during the experimental years.The drought index (DI) for yearly rainfall was generated similar to the previous study and used to examine fluctuations and status in rainfall over the six years [2].
Manual harvesting of lucerne took place twice a year, in late June and early October.Plant samples were weighed after being dried for 72 h at 75 degrees Celsius.The entire aboveground biomass taken from both harvests was used to calculate the lucerne annual fodder yield.At physiological maturity, all wheat, corn, potato, and millet plants in each plot were manually harvested.After harvest, the aboveground biomass was weighed before threshing, and a subsample of aboveground biomass was collected from each plot and baked for 30 min at 105 • C, then 48 h at 80 • C, and then weighed to assess moisture content, which was used to compute dry biomass yield.For each plot, the air-dried grain weight was also recorded and utilized to determine grain yield.Each plot's fresh tuber yield was recorded and utilized to calculate potato yield.
Because crop yields are not directly comparable, they must be adjusted to account for differences in economic value per unit of yield weight.As a result, the net economic return on each treatment was determined in this study.Straw was included in the revenue calculations in this study for all crops except potato.According to a previous study, the net economic return was computed [21].Previous studies were used to calculate the prices of grains, pasture, and numerous agricultural inputs [21][22][23][24].
The saturation of available soil water (SASW) for plants was quantified as the ratio of plant-available water storage and available water holding capacity.Information on the SASW in deep profiles is indispensable for understanding soil-plant interactions and soil water storage degree in the region [25].The SASW was calculated according to the previous study [25].The following equation was used to calculate the soil water relative restoration index [18].
where WR is the soil water restoration index (%), WC1 is the amount of soil water storage in October 2018 (mm), and WC0 is the amount of soil water storage in October 2012 (mm).Three soil cores from the surface to a depth of 300 cm were collected from each plot in October 2018 to determine soil organic carbon, total nitrogen, and total phosphorus.Soil organic carbon was determined by wet oxidation with K 2 Cr 2 O 7 and H 2 SO 4 , soil total N was determined by Kjeldahl analysis, and soil total P was determined after digestion with perchloric acid according to the previous study [9].Soil total porosity, bulk density, soil sorptivity, and saturated hydraulic conductivity were measured in October 2015.Soil bulk density and total porosity were measured in a beveled stainless-steel ring using the methods of Xie et al. ( 2020) and Yu et al. (2018) [21,26].Soil saturated hydraulic conductivity was determined using the single head pressure ring infiltrometer method (i.e., the constant head method) according to the previous studies [21,27].Soil sorptivity (S) was assessed as: where S is soil sorptivity, I is the cumulative infiltration, and t denotes the time.

Statistical Analyses
SAS software was used to conduct the analysis of variance (ANOVA) (SAS Institute, Inc., Cary, NC, USA).The Kolmogorov-Smirnov test was used to conduct normality and homogeneity tests on the data before the ANOVA.The Kolmogorov-Smirnov test indicated that the distributions of data were a typical normal distribution (p < 0.05).Therefore, all data were statistically analyzed as a completely randomized design with four replications using analysis of variance (ANOVA).The significant differences between the means were estimated at 95% confidence level using Fisher's protected LSD test.

Weather Conditions
There was a significant difference in precipitation between the four growing seasons (Figure 1).Annual precipitation in 2014 and 2017 was similar to the long-term average of 390 mm (1970-2012), while it was wetter than the long-term average in 2013 (471 mm) and 2018 (445 mm), and drier than the long-term average in 2015 (340 mm) and 2016 (300 mm) (Table 2).Seasonal precipitation for wheat was similar to the annual average in 2016-2017 (204 mm), but was higher in 2013 (290 mm), 2015 (235 mm), and 2018 (287 mm), and lower in 2014 (154 mm).Seasonal precipitation for corn, potato, and millet was comparable to the long-term estimate in 2017, but wetter in 2013 (423 mm) and 2018 (366 mm) and drier in 2014-2015 (275 mm) and 2016 (227 mm) (data not shown).

Soil Properties
Soil saturated hydraulic conductivity and soil sorptivity were significantly affected by the lucerne-crop rotation systems (Figure 2).Under lucerne with wheat rotation, the highest soil saturation hydraulic conductivity was found, followed by L-F, L-M, L-L, and L-C, and the lowest under L-P.Soil sorptivity increased by 1.19, 2.31, 0.66, 0.75 (p < 0.05), and 0.25 mm min −1/2 (p > 0.05) under L-F, L-W, L-P, L-M, and L-C crop rotations, respectively, compared to lucerne-lucerne.Crop rotation treatments also affected soil bulk density and porosity (Table 3).With the exception of L-M in the 10-30 cm soil layer, soil bulk density was highest under L-L and lowest under L-M in the 0-50 cm soil layers.In contrast, with the exception of L-M in the 10-30 cm soil layer, total soil porosity was lowest under L-L and highest under L-M in the 0-50 cm soil layers.
water condition for the experimental years at the Rainfed Agricultural Experimental station o Gansu Agricultural University.

Soil Properties
Soil saturated hydraulic conductivity and soil sorptivity were significantly affecte by the lucerne-crop rotation systems (Figure 2).Under lucerne with wheat rotation, th highest soil saturation hydraulic conductivity was found, followed by L-F, L-M, L-L, an L-C, and the lowest under L-P.Soil sorptivity increased by 1.19, 2.31, 0.66, 0.75 (p < 0.05 and 0.25 mm min −1/2 (p > 0.05) under L-F, L-W, L-P, L-M, and L-C crop rotations, respec tively, compared to lucerne-lucerne.Crop rotation treatments also affected soil bulk den sity and porosity (Table 3).With the exception of L-M in the 10-30 cm soil layer, soil bul density was highest under L-L and lowest under L-M in the 0-50 cm soil layers.In con trast, with the exception of L-M in the 10-30 cm soil layer, total soil porosity was lowes under L-L and highest under L-M in the 0-50 cm soil layers.1.
Crop rotation systems with lucerne had a significant effect on total nitrogen content, organic carbon content, total P content, mineral N content, and available P content in the soil (Figures 3 and 4).Compared to lucerne-lucerne, all crop rotation treatments decreased soil organic carbon and total nitrogen, while increasing soil total P, mineral nitrogen content, and available P levels at all soil depths.In the 0-10 cm soil layer, lucernefallow, L-W, and L-P significantly reduced soil organic carbon, while L-C significantly reduced soil organic carbon in the 0-10 and 50-110 cm soil layers.In the soil layers of 0-80 cm, lucerne-wheat and L-C significantly reduced total soil N, L-F strongly reduced total N in the soil layers of 0-50 cm, and L-M significantly reduced total N in the soil layers of 50-80 cm.In the soil layers of 0-50 cm, lucerne-crop rotation drastically increased the total P content of the soil.In the soil layers of 0-170 cm, the mineral N content of the soil was significantly higher than in the L-L soil layers, and in the soil layers of 80-170 cm, the mineral N content of the soil was much higher in L-F than in the L-L soil layers.Available soil P content increased significantly in the 0-110 cm soil layers under lucerne-crop treatments and in the 0-80 cm soil layers under lucerne-fallow rotation treatments.L-P significantly decreased available potassium in the 0-50 cm soil layer, L-M significantly decreased available potassium in the 0-30 cm soil layer, and L-C significantly decreased available potassium in the 0-10 cm soil layer compared to L-L.Treatment abbreviations are defined in Table 1.
Crop rotation systems with lucerne had a significant effect on total nitrogen content, organic carbon content, total P content, mineral N content, and available P content in the soil (Figures 3 and 4).Compared to lucerne-lucerne, all crop rotation treatments decreased soil organic carbon and total nitrogen, while increasing soil total P, mineral nitrogen content, and available P levels at all soil depths.In the 0-10 cm soil layer, lucerne-fallow, L-W, and L-P significantly reduced soil organic carbon, while L-C significantly reduced soil organic carbon in the 0-10 and 50-110 cm soil layers.In the soil layers of 0-80 cm, lucerne-wheat and L-C significantly reduced total soil N, L-F strongly reduced total N in the soil layers of 0-50 cm, and L-M significantly reduced total N in the soil layers of 50-80 cm.In the soil layers of 0-50 cm, lucerne-crop rotation drastically increased the total P content of the soil.In the soil layers of 0-170 cm, the mineral N content of the soil was significantly higher than in the L-L soil layers, and in the soil layers of 80-170 cm, the mineral N content of the soil was much higher in L-F than in the L-L soil layers.Available soil P content increased significantly in the 0-110 cm soil layers under lucerne-crop treatments and in the 0-80 cm soil layers under lucerne-fallow rotation treatments.L-P significantly decreased available potassium in the 0-50 cm soil layer, L-M significantly decreased available potassium in the 0-30 cm soil layer, and L-C significantly decreased available potassium in the 0-10 cm soil layer compared to L-L.  1.
Crop rotation systems with lucerne had a significant effect on total nitrogen content, organic carbon content, total P content, mineral N content, and available P content in the soil (Figures 3 and 4).Compared to lucerne-lucerne, all crop rotation treatments decreased soil organic carbon and total nitrogen, while increasing soil total P, mineral nitrogen content, and available P levels at all soil depths.In the 0-10 cm soil layer, lucernefallow, L-W, and L-P significantly reduced soil organic carbon, while L-C significantly reduced soil organic carbon in the 0-10 and 50-110 cm soil layers.In the soil layers of 0-80 cm, lucerne-wheat and L-C significantly reduced total soil N, L-F strongly reduced total N in the soil layers of 0-50 cm, and L-M significantly reduced total N in the soil layers of 50-80 cm.In the soil layers of 0-50 cm, lucerne-crop rotation drastically increased the total P content of the soil.In the soil layers of 0-170 cm, the mineral N content of the soil was significantly higher than in the L-L soil layers, and in the soil layers of 80-170 cm, the mineral N content of the soil was much higher in L-F than in the L-L soil layers.Available soil P content increased significantly in the 0-110 cm soil layers under lucerne-crop treatments and in the 0-80 cm soil layers under lucerne-fallow rotation treatments.L-P significantly decreased available potassium in the 0-50 cm soil layer, L-M significantly decreased available potassium in the 0-30 cm soil layer, and L-C significantly decreased available potassium in the 0-10 cm soil layer compared to L-L.    1.

Soil Water Recovery
After the six-year experimental period, L-L had significantly less soil water content in all soil layers than other treatments (Figure 5).In the 0-180 cm soil layers, L-W, L-C, L-P, L-M, and L-F substantially enhanced soil water content compared to L-L; however, soil water content between L-W, L-C, L-P, L-M, and L-F was not significantly different in all soil layers.L-F had the highest saturation of available soil water at a 0-300 cm soil depth, followed by L-W, L-C, L-P, and L-M, and L-L had the lowest saturation of available soil water at a 0-300 cm soil depth (Figure 6).In soil layers 0-110 and 110-200 cm, the rate of soil water recovery was higher than in soil layer 200-300 cm (Figure 7).In soil layer 0-110 cm, the rate of soil water reclamation was 3.3, 16.7, 11.0, 10.6, 10.9, and 9.9 mm per year; in soil layer 110-200, the rate was 2.6, 13.9, 11.1, 12.2, 9.7, and 10.4 mm per year; and in soil layer 200-300, the rate was 2.3, 8.9, 10.9, 10.9, 9.9, and 8.8 mm per year.In the soil depth 0-300 cm, the rate of soil water recovery was 8.2, 39.5, 33.0, 33.7, 30.5, and 29.1 mm per year.L-F had the highest soil water absolute restoration index, preceded by L-C, L-W, L-P, and L-M and L-L had the lowest value (Table 4).The absolute soil water restoration indices at 0-300 cm soil depth was 116, 179, 166, 167, 161, and 158 percent among L-L, L-F, L-W, L-C, L-P, and L-M, respectively, during the six-year experimental period.* Treatment abbreviations are defined in Table 1.# Different lowercase letters in the same column represent significant differences at p ≤ 0.05 between treatments.  1.

Soil Water Recovery
After the six-year experimental period, L-L had significantly less soil water content in all soil layers than other treatments (Figure 5).In the 0-180 cm soil layers, L-W, L-C, L-P, L-M, and L-F substantially enhanced soil water content compared to L-L; however, soil water content between L-W, L-C, L-P, L-M, and L-F was not significantly different in all soil layers.L-F had the highest saturation of available soil water at a 0-300 cm soil depth, followed by L-W, L-C, L-P, and L-M, and L-L had the lowest saturation of available soil water at a 0-300 cm soil depth (Figure 6).In soil layers 0-110 and 110-200 cm, the rate of soil water recovery was higher than in soil layer 200-300 cm (Figure 7).In soil layer 0-110 cm, the rate of soil water reclamation was 3.3, 16.7, 11.0, 10.6, 10.9, and 9.9 mm per year; in soil layer 110-200, the rate was 2.6, 13.9, 11.1, 12.2, 9.7, and 10.4 mm per year; and in soil layer 200-300, the rate was 2.3, 8.9, 10.9, 10.9, 9.9, and 8.8 mm per year.In the soil depth 0-300 cm, the rate of soil water recovery was 8.2, 39.5, 33.0, 33.7, 30.5, and 29.1 mm per year.L-F had the highest soil water absolute restoration index, preceded by L-C, L-W, L-P, and L-M and L-L had the lowest value (Table 4).The absolute soil water restoration indices at 0-300 cm soil depth was 116, 179, 166, 167, 161, and 158 percent among L-L, L-F, L-W, L-C, L-P, and L-M, respectively, during the six-year experimental period.Treatment abbreviations are defined in Table 1.Treatment abbreviations are defined in Table 1.
Figure 5. Volumetric soil water content in the 0-300 cm soil depths after lucerne-to-crop c in October 2018.Horizontal bars denote Fisher's protected least significant difference a Treatment abbreviations are defined in Table 1.Treatment abbreviations are defined in Table 1.T abbreviations are defined in Table 1.Treatment abbreviations are defined in Table 1.
Figure 5. Volumetric soil water content in the 0-300 cm soil depths after lucerne-to-crop c in October 2018.Horizontal bars denote Fisher's protected least significant difference a Treatment abbreviations are defined in Table 1.Treatment abbreviations are defined in Table 1.T abbreviations are defined in Table 1.  1.   1.

Discussion
High-water-use plants such as lucerne remove a lot of moisture from the soil when rainfall is limited and unpredictable, resulting in soil desiccation [10,28,29].As a result, soil productivity cannot be maintained for many years by continuously cultivating lucerne because it results in a dried soil layers [28].To avoid soil drying after several years of lucerne cultivation, some appropriate field management methods must be implemented [6].In this research, soil moisture increased dramatically in all rotations during the experiment.Soil water restoration was the most efficient after lucerne-to-fallow conversion, mainly because annual water consumption was less by 6-31 mm under lucernefallowing compared to other treatments (data not shown).Interestingly, although corn had higher evapotranspiration than others crop [1,21,29], lucerne-to-corn conversion efficiently enhanced soil water restoration.This is most likely because the ridge-furrow system for corn with annual all-plastic film mulching had significant benefits in terms of furthering water infiltration, reducing soil evaporation [30,31], and replenishing soil moisture in the autumn (July to September), thereby increasing soil water storage in the surface soil moisture, agreeing with findings in a previous study [28].Under wheat, saturated soil hydraulic conductivity and soil sorptivity were highest, reflecting the fact that the field was plowed after wheat harvest, resulting in a significant loss of soil evaporation due to the exposed surface during fallowing.Due to the limited and erratic rainfall in this region, soil water recovery in the soil layers of 30-180 cm was significantly higher in the lucerneto-fallowing conversion than under the lucerne-to-crop conversions, which means that  1.

Discussion
High-water-use plants such as lucerne remove a lot of moisture from the soil when rainfall is limited and unpredictable, resulting in soil desiccation [10,28,29].As a result, soil productivity cannot be maintained for many years by continuously cultivating lucerne because it results in a dried soil layers [28].To avoid soil drying after several years of lucerne cultivation, some appropriate field management methods must be implemented [6].In this research, soil moisture increased dramatically in all rotations during the experiment.Soil water restoration was the most efficient after lucerne-to-fallow conversion, mainly because annual water consumption was less by 6-31 mm under lucerne-fallowing compared to other treatments (data not shown).Interestingly, although corn had higher evapotranspiration than others crop [1,21,29], lucerne-to-corn conversion efficiently enhanced soil water restoration.This is most likely because the ridge-furrow system for corn with annual all-plastic film mulching had significant benefits in terms of furthering water infiltration, reducing soil evaporation [30,31], and replenishing soil moisture in the autumn (July to September), thereby increasing soil water storage in the surface soil moisture, agreeing with findings in a previous study [28].Under wheat, saturated soil hydraulic conductivity and soil sorptivity were highest, reflecting the fact that the field was plowed after wheat harvest, resulting in a significant loss of soil evaporation due to the exposed surface during fallowing.Due to the limited and erratic rainfall in this region, soil water recovery in the soil layers of 30-180 cm was significantly higher in the lucerne-to-fallowing conversion than under the lucerne-to-crop conversions, which means that the infiltration of soil water in the autumn is primarily in the shallow soil layer.It also implies that bare land may not be conducive for water infiltrating into deep soil; in contrast, root channels formed by plant root growth might be beneficial for water infiltration into deep soil, agreeing with findings in a previous study [32].
After the conversion of lucerne to crops, the soil bulk density was reduced in all soil layers, mainly because the soil was plowed every year, leading to increased soil total porosity.However, after conversion from lucerne to cropland, soil organic carbon, total N, and mineral nitrogen decreased, consistent with previous research findings [33][34][35].This was most likely due to the fact that soil respiration rates were greater under long-term intensive cropping in lucerne-crop rotation systems than in sole lucerne crops [18].In addition, carbon loss during lucerne-to-crop conversions was significantly greater than that following lucerne-lucerne conversion, implying a greater opportunity for N loss [36].On the other hand, because of its ability to fix atmospheric N, lucerne cultivation can be a renewable and environmentally friendly source of nitrogen [37], resulting in higher soil total N and mineral N content under L-L than under other treatments.Soil P content was accumulated by lucerne and depleted through forage harvesting, resulting in a lower soil P content under lucerne-lucerne, which is consistent with previous research findings [9,19,38].Phosphate fertilization, on the other hand, provided timely soil P replenishment for the crop productivity of lucerne crops.As a result of the conversion of lucerne crops, the total and available soil P increased.Therefore, soil total P and available P contents were increased after the conversion of lucerne to crops.
For a long time, the main crops in the dryland farming on the western Loess Plateau were potato and spring wheat, supplemented by other short seasonal crops including millet and flax (Linum usitatissimum L.); these short seasonal crops are widely planted to avoid famines caused by wheat failures [5].Corn has been considered one of the leading crops on the semiarid western Loess Plateau of China in the past 20 years due to the extensive use of the fully mulched ridge-furrow system and excellent yields.Wheat yields were low in this region, with harvested spring wheat yields in our study ranging from 903 to 2263 kg ha −1 (data not shown), which was due to the spring drought severely limiting wheat yield and economic returns; this is consistent with previous research findings [1,39].In this study, the net economic return of corn was RMB 5377−29,443 ha −1 yr −1 , consistent with the findings from previous studies with plastic mulching in this region [21] and without plastic mulching in northeastern China [23]; fresh tuber yields of potato varied from 6905 to 15,517 kg ha −1 , with an average yield of 10,921 kg ha −1 (data not shown), which supports previous reports [40,41].Interestingly, although continuous production of lucerne caused severe soil dryness in deep soil layers, it did not significantly reduce the yields of potato and corn, presumably because their critical growth period coincides with the main rainfall period, thereby maintaining a high yield.The results of this study demonstrated that lucerne-to-corn conversion had a greater soil water restoration and higher net economic return than other lucerne-crop rotation systems.These results highlight that when considering the economic benefits and soil water restoration, corn sown under the ridge-furrow system with fully plastic film annual mulching is the most suitable rotation species for sustainable land production.

Conclusions
Based on the analysis of long-term experimental results, this study shows that compared with long-term lucerne planting, the continued lucerne crop rotation decreased soil bulk density, soil organic carbon, total N, and mineral N, whereas total P and available P increased.Lucerne-corn rotation achieved the greatest economic return and rate of soil water recovery among four lucerne-to-crop conversion treatments.Considering the aforementioned factors such as economic benefits and soil moisture recovery, lucernecorn rotation can be regarded as a suitable rotation mode when corn is sown under the ridge-furrow system with fully plastic film annual mulching.

Figure 1 .
Figure 1.Monthly precipitations in the years from 2013 to 2018 and 1997-2011 average for the lu cerne-to-crop conversion treatment of the study at the Rainfed Agricultural Experiment Station o Gansu Agricultural University.

Figure 1 .
Figure 1.Monthly precipitations in the years from 2013 to 2018 and 1997-2011 average for the lucerne-to-crop conversion treatment of the study at the Rainfed Agricultural Experiment Station of Gansu Agricultural University.

Figure 2 .
Figure 2. Soil saturated hydraulic conductivity (a) and soil sorptivity (b) under lucerne-to-crop conversion in October 2015.For a given dependent variable, bars with different letters indicate significant difference (p ≤ 0.05).Error bars indicate standard deviations of the means (n = 3).Treatment abbreviations are defined in Table1.

Figure 3 .
Figure 3. Soil organic carbon (a), total nitrogen (b), and total phosphorus (c) after lucerne-to-crop conversion in October 2018.Horizontal bars denote Fisher's protected least significant difference at p ≤ 0.05.Treatment abbreviations are defined in Table1.

Figure 2 .
Figure 2. Soil saturated hydraulic conductivity (a) and soil sorptivity (b) under lucerne-to-crop conversion in October 2015.For a given dependent variable, bars with different letters indicate significant difference (p ≤ 0.05).Error bars indicate standard deviations of the means (n = 3).Treatment abbreviations are defined in Table1.

Agronomy 2022, 12 , 1905 6 of 12 Figure 2 .
Figure 2. Soil saturated hydraulic conductivity (a) and soil sorptivity (b) under lucerne-to-crop conversion in October 2015.For a given dependent variable, bars with different letters indicate significant difference (p ≤ 0.05).Error bars indicate standard deviations of the means (n = 3).Treatment abbreviations are defined in Table1.

Figure 3 .
Figure 3. Soil organic carbon (a), total nitrogen (b), and total phosphorus (c) after lucerne-to-crop conversion in October 2018.Horizontal bars denote Fisher's protected least significant difference at p ≤ 0.05.Treatment abbreviations are defined in Table1.

Figure 3 .
Figure 3. Soil organic carbon (a), total nitrogen (b), and total phosphorus (c) after lucerne-to-crop conversion in October 2018.Horizontal bars denote Fisher's protected least significant difference at p ≤ 0.05.Treatment abbreviations are defined in Table1.

Figure 5 .
Figure 5. Volumetric soil water content in the 0-300 cm soil depths after lucerne-to-crop conversion in October 2018.Horizontal bars denote Fisher's protected least significant difference at p ≤ 0.05.Treatment abbreviations are defined in Table1.

Figure 6 .
Figure 6.The saturation of available soil water in the 0-300 cm soil depth in October 2018 letters indicate treatment means that are significantly different (p ≤ 0.05).Error bars indic ard deviations of the means (n = 3).Treatment abbreviations are defined in Table1.

Figure 7 .
Figure 7. Rate of soil water recovery of lucerne-to-crop conversion in October from 201 Different letters indicate treatment means that are significantly different among treatme 0-300 cm soil depth (p ≤ 0.05).Error bars indicate standard deviations of the means (n = 3).T abbreviations are defined in Table1.

Figure 6 .
Figure 6.The saturation of available soil water in the 0-300 cm soil depth in October 2018.Different letters indicate treatment means that are significantly different (p ≤ 0.05).Error bars indicate standard deviations of the means (n = 3).Treatment abbreviations are defined in Table1.

Figure 6 .
Figure 6.The saturation of available soil water in the 0-300 cm soil depth in October 2018.letters indicate treatment means that are significantly different (p ≤ 0.05).Error bars indic ard deviations of the means (n = 3).Treatment abbreviations are defined in Table1.

Figure 7 .
Figure 7. Rate of soil water recovery of lucerne-to-crop conversion in October from 201 Different letters indicate treatment means that are significantly different among treatme 0-300 cm soil depth (p ≤ 0.05).Error bars indicate standard deviations of the means (n = 3).T abbreviations are defined in Table1.

Figure 7 .
Figure 7. Rate of soil water recovery of lucerne-to-crop conversion in October from 2012 to 2018.Different letters indicate treatment means that are significantly different among treatments in the 0-300 cm soil depth (p ≤ 0.05).Error bars indicate standard deviations of the means (n = 3).Treatment abbreviations are defined in Table1.

Figure 8 .
Figure 8.Average net economic return of lucerne-to-crop conversion in October from 2012 to 2018.Different letters indicate treatment means that are significantly different (p ≤ 0.05).Error bars indicate standard deviations of the means (n = 3).Treatment abbreviations are defined in Table1.

Figure 8 .
Figure 8.Average net economic return of lucerne-to-crop conversion in October from 2012 to 2018.Different letters indicate treatment means that are significantly different (p ≤ 0.05).Error bars indicate standard deviations of the means (n = 3).Treatment abbreviations are defined in Table1.

Table 1 .
Details of experimental treatments.

Table 3 .
Soil bulk density and soil total porosity of the crop rotation treatments in October 2015.
* Treatment abbreviations are defined in Table1.# Different lowercase letters in the same colum represent significant differences at p ≤ 0.05 between treatments.

Table 2 .
Annual rainfall from 2013 to 2018 and average of 1997-2011, drought index (DI), and soil water condition for the experimental years at the Rainfed Agricultural Experimental station of Gansu Agricultural University.

Table 3 .
Soil bulk density and soil total porosity of the crop rotation treatments in October 2015.
* Treatment abbreviations are defined in Table1.# Different lowercase letters in the same column represent significant differences at p ≤ 0.05 between treatments.

Table 4 .
Average soil water absolute restoration index of the crop rotation treatments in October from 2012 to 2018.

Table 4 .
Average soil water absolute restoration index of the crop rotation treatments in October from 2012 to 2018.Treatment abbreviations are defined in Table1.# Different lowercase letters in the same column represent significant differences at p ≤ 0.05 between treatments. *