E ﬀ ects of Soft Rock and Biochar Applications on Millet ( Setaria italica L.) Crop Performance in Sandy Soil

: In arid and semi-arid regions, desertiﬁcation threatens crop production because it reduces the soil’s capacity to retain water and soil nutrients. At two fertilizer levels (90 kg N hm − 2 and 45 kg P hm − 2 and 270 kg N hm − 2 and 135 kg P hm − 2 ), the e ﬀ ects of soft rock (sand: soft rocks = 3:1) and biochar (4500 kg hm − 2 ) applications on soil moisture, soil nutrients, and millet ( Setaria italica L.) photosynthesis, yield, and its agronomic traits (biomass, thousand kernel weight, harvest index) were investigated in a ﬁeld experiment in the Mu Us Sandy Land of China in 2018–2019. The addition of biochar and soft rock singly increased soil water content, alkali-hydrolyzed nitrogen (AN), total nitrogen (TN) and phosphorus (TP), and organic matter content signiﬁcantly, suggesting that their application may increase the nutrient and water holding capacity of soil. Application of biochar and soft rock singly increased the net photosynthesis rate of millet ﬂag leaf, at the ﬂowering stage, from 15.97% to 56.26%. Biochar and soft rock application increased the yield range (2109.0 kg hm − 2 to 5024.7 kg hm − 2 ) from 5.26% to 54.60% under the same fertilizer level. Correlation analyses showed grain yield was signiﬁcantly correlated with photosynthesis rate at the ﬂowering stage, which was signiﬁcantly correlated with soil AN at ﬂowering, soil TP at ﬂowering and harvest, and soil TN at ﬂowering. These results indicated that the application of biochar and soft rock singly could increase soil fertilizer holding capacity to improve the photosynthesis rate at ﬂowering, and, therefore, lead to improvements in crop yield.


Introduction
Land desertification is a serious environmental problem all over the world, affecting one-fifth of the world's population. It is a major threat to the sustainability of agriculture and economic development, especially in arid and semi-arid regions [1,2]. Generally, land degradation mechanisms involve the loss of soil water which further intensifies soil desertification [3]. Decreases in soil water holding capacity and increases in the loss of soil nutrients cause soil to become barren [4]. Desertification induces a decline in soil carbon and nutrients and soil water which affects plant growth and accelerates soil erosion by wind and rain [5,6]. In the future, soil degradation is likely to have a greater impact on agricultural incomes as yields decrease and input costs grow in irrigated, high-quality rain-fed, densely populated, and low-fertility lands.

Experimental Site and Design
The experimental site was in a field station in the Shaanxi Institute of Sand Control (38 • 19 49" N, 109 • 42 42" E, 1201 m a.s.l. Mu Us Sand Land) with a typical warm temperate monsoon climate, Shaanxi Province, China. The annual average temperature is 8.3 • C and the mean annual precipitation is 365.7 mm. Precipitations are concentrated in July and August. In our experimental period (2018), the maximum precipitation was observed in August. However, the precipitation was 206 mm in July, followed by 246 mm in August, which was 150% higher than usual. The cumulative precipitation during the growing season (from early May to the end of September) was 615.3 mm in 2018 and 421.9 mm in 2019 ( Figure 1). during the growing season (from early May to the end of September) was 615.3 mm in 2018 and 421.9 mm in 2019 ( Figure 1).
The experiment was designed based on two fertilizer levels containing a normal yield pattern (90 kg N hm −2 and 45 kg P hm −2 (NY)) and a high yield pattern (270 kg N hm −2 and 135 kg P hm −2 (HY)). The number of other fertilizers (75 kg K hm −2 , 15 kg Zn hm −2 and 1 kg B hm −2 ) were the same for both fertilizer levels. The fertilizers were applied when the ground was tilled. Four treatments of biochar and soft rock were added to the sandy soil as follows: sand only (CK); sand and biochar (C; 4500 kg biochar hm −2 ); sand and soft rock (SR; sand: soft rock = 3:1); and sand, biochar, and soft rock (SRC; 4500 kg biochar hm -2 and sand: soft rocks = 3:1). Therefore, in our experiment, there were eight groups (Table 1 and Figure S1). A randomized complete block design with a split-plot treatment and four replicates per treatment was used. Each plot was 4-m wide by 5-m long. The plots consisted of homogenous sandy soil prior to our experiment. Before the start of the experiment, the land was bare sand and the land had not been used to grow crops. The soft rock was collected from the Dongsheng district, Erdos city, Inner Mongolia. The biochar was produced from crop straw (585 g kg −1 total C, EC 1.27 mS cm −1 ) by Liaoning Golden Future Agriculture Technology Co., Ltd (Anshan, Liaoning province, China). The properties of the soft rock, sandy soil, and biochar used in our experiment are shown in Table 2. A total depth of 20 cm of the soil mix was added on top of the sandy soil plots.   The experiment was designed based on two fertilizer levels containing a normal yield pattern (90 kg N hm −2 and 45 kg P hm −2 (NY)) and a high yield pattern (270 kg N hm −2 and 135 kg P hm −2 (HY)). The number of other fertilizers (75 kg K hm −2 , 15 kg Zn hm −2 and 1 kg B hm −2 ) were the same for both fertilizer levels. The fertilizers were applied when the ground was tilled. Four treatments of biochar and soft rock were added to the sandy soil as follows: sand only (CK); sand and biochar (C; 4500 kg biochar hm −2 ); sand and soft rock (SR; sand: soft rock = 3:1); and sand, biochar, and soft rock (SRC; 4500 kg biochar hm −2 and sand: soft rocks = 3:1). Therefore, in our experiment, there were eight groups (Table 1 and Figure S1). A randomized complete block design with a split-plot treatment and four replicates per treatment was used. Each plot was 4-m wide by 5-m long. The plots consisted of homogenous sandy soil prior to our experiment. Before the start of the experiment, the land was bare sand and the land had not been used to grow crops. The soft rock was collected from the Dongsheng district, Erdos city, Inner Mongolia. The biochar was produced from crop straw (585 g kg −1 total C, EC 1.27 mS cm −1 ) by Liaoning Golden Future Agriculture Technology Co., Ltd. (Anshan, Liaoning province, China). The properties of the soft rock, sandy soil, and biochar used in our experiment are shown in Table 2. A total depth of 20 cm of the soil mix was added on top of the sandy soil plots.

Crop Management
Changsheng07, a variety of foxtail millet bred by millet research institute Shanxi Academy of Agricultural Science in 2009, was selected for our experiment. Foxtail millet is an important commercial food crop in the local region and Changsheng07 is widely used in local foxtail millet production as it has a strong resistance to desiccation [25]. The seeds were sown on 7 May 2018, and 11 May 2019, respectively. Seeds were spread evenly in the field with a row spacing of 50 cm. When seedlings reached the two-leaf stage, artificial thinning was conducted to maintain a plant spacing of 6-7 cm (300,000 plants hm −2 ). At the jointing stage, a net was placed at 2.5 m above the ground to prevent access to birds; the net was removed at harvest. Because of the dry climate with low rainfall, the plots were irrigated with underground water before sowing and before the seedling stage in July to ensure seed emergence and seedling growth. Plots were not irrigated between July and late September when the harvest was conducted.

Sample Collection
When 80% of the plants were in flower (12 August 2018, and 14 August 2019), the aboveground biomass was collected in order to measure dry matter accumulation. Ten plants from each plot were manually cut close to the ground, taken to the laboratory, dried to a constant weight at 75 • C, heated to 105 • C for 20 min so as to deactivate all enzymes, and then weighed. At the flowering stage of growth seasons in 2018 and 2019, soil to a depth of 0-20 cm was collected, and alkali-hydrolyzed nitrogen (AN), available phosphorus (AP), and soil organic matter were measured. During the harvest stage of the 2019 growth season, soil to a depth of 0-20 cm was collected, and AN, AP, and soil organic matter were measured. Soil total nitrogen (TN) and phosphorus (TP) measured during the 2019 growing season. Soil samples from five points per plot, following an s-shaped distribution in the plot, were collected and mixed to give one combined sample for each plot. At the flowering stage, soil samples to depths of 0-20 cm and 21-40 cm were collected using a soil borer to measure soil water content. Fresh soil weight (FW) was weighed immediately after soil collection. After the samples were dried at 105 • C, the dry At harvest time (25 September 2018, and 30 September 2019), the mature vegetation from a 1-m 2 quadrant in each plot was collected manually. Following collection, the spike numbers for each plot were counted then the spike number per hm 2 was calculated. When natural withering was complete, total aboveground biomass was measured and then the plants were artificially threshed to measure the grain yield. Lastly, the harvest index (HI) was calculated according to the following formula [26]: The thousand kernel weight was counted manually and then weighed.

Net Photosynthetic Rate
Net photosynthetic rate is considered to be one of the most important indexes for evaluating the capacity of crop yield formation [26]. In this study, the photosynthetic rate of two growth stages (jointing stage and anthesis stage) were measured. Photosynthetic traits were recorded during the 2018 and 2019 growing seasons. The photosynthetic traits of new fully expanded leaves at the jointing stage and flag leaves at anthesis were measured, in all plots, between 9:30 a.m. and 11:30 a.m. using a Li-6400 portable photosynthesis system (LI-COR Inc., Lincoln, NE, USA). The photosynthetic traits of new fully expanded leaves or flag leaves were calculated as the mean readings for five leaves (one leaf per plant) per plot. All measurements were performed on the middle portions of the leaf exposed to full sunlight, approximately half-way along the length of the leaf.

Soil Properties Measurements
Soil nutrient measurements were conducted according to methods described by Bao [27]. Soil organic matter was measured using the potassium dichromate oxidation heating method. Soil AN was detected using the alkali solution diffusion method. Soil AP was measured using the Olsen method (extraction agent: 0.05 mol/L NaHCO 3 ). Soil TN was detected with Kjeldahl determination and TP was detected using the Mo-Sb anti-spectrophotography method. Soil pH was detected using a benchtop pH meter (SevenExcellence pH Meter Line, METTLER TOLEDO, Greifensee, Switzerland). According to experiment design, the detection was with four replications and every replication held two technological replications.

Data Analysis
The effects of the treatments (CK, C, SR, and SRC), years (2018 and 2019), fertilizer levels (NY and HY), and their interactions on the measured variables were evaluated using one-and three-way ANOVA. When F-values were significant, the least significant difference (LSD) test was used to compare means. Pearson correlation analyses were used to assess correlations between grain yield, soil indexes, and photosynthetic rate. All statistical analysis was performed through SPSS (SPSS Inc., Chicago, IL, USA) software. The figures were created using SigmaPlot 12.5 (Systat Software Inc., San Jose, CA, USA).

Grain Yield and Related Agronomic Traits
Grain yield was significantly affected by treatment, year, fertilizer, and the interactions (Table 3). In the 2018 growing season, the yield over all treatments ranged from 2109.0 kg hm −2 (CK + NY) to 5024.7 kg hm −2 (C + HY). Under the normal fertilizer treatments, the C treatment increased the yield significantly, while the SR and SRC treatments did not increase yield significantly. In the high yield Agronomy 2020, 10, 669 6 of 19 treatment, the C and SR treatments increased yield significantly compared with CK, while there was no significant difference between the CK and SRC treatments. The increases in yield were 46% and 16.8%, respectively. In the 2019 growing season, the yield ranged from 2772.7 kg hm −2 (CK + NY) to 4898.7 kg hm −2 (C + HY). Compared with the 2018 season, the average yield increased by 18.6%. In 2019, C and SR treatments produced a higher yield than the CK treatment at the same fertilizer level. In 2019, the C treatment produced the highest yield of all treatments in the same fertilizer level.
Biomass was significantly affected by treatment, fertilizer but not year (Table 3). Therefore, the data were pooled over the two years. The C and SR treatments increased the biomass significantly under both fertilizer levels. There were no significant differences between the SRC and the CK treatments or between the SRC and C and SR treatments at the same fertilizer level.
HI was significantly affected by fertilizer, year, and interactions (Table 3). In the 2018 season, under both fertilizer levels, Treatment C had the highest HI (25.9% and 30.0%) of all the treatments. In 2018, the lowest HI (22.2% in NY and 26.2% in HY) was found in the SR and SRC treatment, respectively. In 2019, there was no significant difference between treatments under the normal fertilizer level. Under the high fertilizer level, the SRC had the highest HI (37.0%), while Treatment C had the lowest HI (30.5%). Overall, the HI in 2019 increased by 25.6% when compared with the HI in 2018.
One thousand kernel weight expresses the degree of fullness of the grain. In 2018, under the normal fertilizer level, the SR treatment produced the highest thousand kernel weight index (2.35 g), which is 14.1% higher than that of CK ( Figure 2). Under the high fertilizer level, there was no significant difference in thousand kernel weight between treatments. In 2019, under the normal fertilizer level, the SR treatment had the highest thousand kernel weight (2.47 g), while the C treatment had the lowest thousand kernel weight (2.10 g). CK had the maximum thousand kernel weight under the high fertilizer level. Under the high fertilizer level, the application of soft rock and biochar singly increased spike number significantly compared with CK in both seasons (p < 0.05, Figure 3), and the increase ranged from 9.1% to 46.3%. Under the normal fertilizer level, the SRC treatment in 2018, and the C treatment in 2019 produced the most spikes. There was no significant difference among the C, SR, and SRC treatments under the normal fertilizer level.
To detect dynamic changes in millet growth, the biomass at anthesis was measured ( Figure 4). In 2018, the SR treatment had the highest biomass accumulation under both fertilizer levels (17.1 g per plant under NY, and 21.8 g per plant under HY), while there was no significant difference among SR, C, and SRC treatments. In 2019, under the normal fertilizer level, the C treatment produced the highest biomass (19.7 g per plant), while under the high fertilizer level the SR treatment produced the highest biomass accumulation (27.6 g per plant).

Soil Properties
Because of the necessary irrigation of plots completed before July, it was not possible to measure natural soil moisture during irrigation. Therefore, we measured soil water content at anthesis on 12 August 2018 and 14 August 2019. The C treatment decreased the soil water content (at a depth of 0-20 cm) compared with the other three treatments under the normal fertilizer levels in 2018 but in high fertilizer level treatment C had the highest soil water content compared with the other treatments (Table 4). In 2019, these three treatments held higher water contents than that of CK at a depth of 0-20 cm. At the depth of 21-40 cm, the SR treatment had the highest water content, followed by the C, SRC, and CK treatments under the normal fertilizer level in 2018. At the 21-40 cm soil depth, all three treatments had considerably higher water contents (more than 200%) than that of CK. To understand how soft rock and biochar help the soil retain nutrients, we measured soil nutrient content (Table 5, Figure S2). Results showed that, at the flowering stage under both fertilizer levels, SR increased the AN and AP content significantly compared with the CK treatment in both growth stages and increased TN and TP content, especially in normal fertilizer level treatment in the 2019 growing season. The C treatment did not allow the soil to maintain significantly more AP compared to the CK treatment in either season and fertilizer level, while C treatment increased TN and TP content significantly compared with CK treatment. The SRC treatment had not exhibited higher TP content than that of CK (p > 0.05). At the normal fertilizer level, SRC had the second-highest AN content after the SR in 2018 and the third-highest AN content after SR and CK in 2019. However, under the same fertilizer level, the soil AN and AP contents in 2019 did not greatly increase compared with the AN and AP contents in 2018. In the C and SR treatments, organic matter content, in 2018, and both fertilizer treatments, increased significantly compared with the CK treatment. The C and SR treatments both produced a high organic matter content and there was no significant difference between them. The other treatments, SRC and CK, though lower than C and SR, still had higher organic matter contents four times greater than the low content in both fertilizer levels. In 2019, organic matter content greatly increased. As in 2018, the C and SR treatments produced a higher organic matter content than the other treatments (SRC followed by CK) and there was no significant difference between the two treatments (under both fertilizer levels). At the harvest stage, the SR treatment increased soil P, N, and organic matter significantly compared with the CK treatment and there was no significant difference between the SR and C treatment under either fertilizer level. 3.3. Photosynthetic Response.
The net photosynthetic rate ranged from 23.06 to 33.62 µmol CO 2 m −2 s −1 in 2018 at the jointing stage (Table 6). At this stage, the C and SR treatment caused a significant increase in the net photosynthetic rate compared with CK. At the flowering stage, the increase in the net photosynthetic rate in SR and C treatments was significant compared with CK. In 2019, at the jointing stage, CK held the highest photosynthetic rate among these four groups under high fertilizer level, while under normal fertilizer level, the SR and C treatments had a higher photosynthetic rate than CK. At the flowering stage, the photosynthetic rate of SR and C treatments was higher than that of CK.

Discussion
Sandy soils, which are low nutrient content, poor water, and fertilizer retention, occupy a large proportion of global arable lands [13]. As we know, lots of approaches are used to improve sandy soil crop productivity, such as fertilization, returning straw to field, cultivation modes, and so on. Among those approaches, soil conditioners can improve soil properties, increase soil water and fertilizer retention capacity, and crop productivity [12,13,15]. In our two-year research study, it was demonstrated that the application of biochar and a local soft rock can increase the grain yield of millet significantly. The application of biochar in non-sandy agriculture fields of average fertility has been widely researched and is known to increase yields [28,29]. Our results confirmed that, in sandy soil, biochar can increase grain yield. In past research, it was found that biochar applications can increase the yield and water-use efficiency of Sorghum bicolor L. [30]. Based on our combined results, it could be concluded that biochar is also suitable for use in desert sandy soil management. In our study, the addition of soft rock increased yield significantly, which was consistent with previous results for wheat, potato, and maize [31]. In many test indexes, soft rock application achieved similar results to biochar application.

Soft Rock Increases the Soils Ability to Retain Nutrients from Fertilizers
Previous studies have proved that soil conditioners, such as polyacrylamide, biochar can effectively increase the water holding capacity, decreasing deep percolation, and reducing evaporation in sandy soils [21,32]. Soft rock applications increase grain yield by improving the soil water retention capacity in sandy soil [8,11]. Here, the application of soft rock could maintain a higher soil water content especially at depths of 21-40 cm after long periods without irrigation, while in the first year, biochar treatment had not increased soil water content compared with CK at depth of 0-20 cm. This is because the application of soft rock on sandy soil may decrease the infiltration rate and saturated hydraulic conductivity [12]. Usually, the sand soil holding high soil saturated hydraulic conductivity and bulk density was averse to hold water. According to a previous study, the application of soft rock could decrease these indexes [11]. Mixing soft rock and sandy soil decreased the soil saturated hydraulic conductivity and bulk density to increase water holding capacity. It was found that biochar application increased soil water content (0-20 cm) with aging, which was consistent with a previous study, in which authors found that biochar application decreased the soil water content in the first year [33]. This may be because the application of biochar increased the soil porosity and decreased the soil water holding capacity [33]. Lately, the improvement of organic matter induced by application of biochar increased soil water holding capacity, as previous research reported [34]. Therefore, it could be concluded that soft rock and biochar applications to sandy soil increase soil water, which could be beneficial to improving nutrient holding capacity.
In sandy soil, nutrient leaching is a major threat to agricultural production [35]. Soil organic matter and nutrient level are important indicators for evaluating soil quality, and they are also affected by the application of soil conditioners [36,37]. Sandy soil has greater permeability coefficients and low fertilizer retention and soft rock has more fine particles with hydrophilic and absorbent properties. In our result, the organic matter was increased with the application of soft rock (Table 5). Bacterium experiment showed that mixing soft rock and sand could increase the diversity, uniformity, and richness index of bacteria was significantly compared with control soil of only sand and increase accumulation of soil organic matter [38]. This indicated that the application of soft rock could affect the bacteria community to regulate the soil organic matter accumulation, and this could explain the increase of organic matter in our experiment. Biochar was found to increase soil fertilizer retention [21], while in our results the biochar had not exhibited a strong ability to hold AP and AN (Table 4). In fact, biochar application in sand soil was found to decrease soil bulk density and increase soil porosity [33]. The huge soil porosity could induce the fertilizer missing into deep soil under flooding irrigation and heavy rainfall conditions in our experiment. While the soft rock held strong capacity in absorbing water, this could be the main reason for mixing soft rock with sand which increased the soil AN and AP.

Soft Rock Increases Photosynthesis and Yield of Millet
Net photosynthetic rate is an important index used to evaluate crop yield production ability [26,[39][40][41] and promoting the photosynthetic rate or yield are the most reliable indicators for evaluating the performance of soil conditioners. In the present study, soft rock and biochar application had increased flowering photosynthetic rate and yield significantly ( Table 6), suggesting that soft rock has great potential as a useful soil conditioner in the sandy soil. We found that the flowering photosynthetic rate was significantly correlated with yield and this indicates that in millet the flowering photosynthesis is important for yield production in millet. Photosynthesis at jointing had little effect on grain yield accumulation compared to photosynthesis at flowering. These results are consistent with results obtained for wheat [41,42]. Much research on soil fertilizer management has focused on photosynthesis [43,44]. A higher nitrogen supply contributes to a higher photosynthesis rate in wheat [45]. We found that a higher photosynthesis rate (which was positively correlated with grain yield) was significantly related to soil nutrient content, especially the soil total phosphorus. In many previous studies, phosphorus was found to play an important role in maintaining high photosynthesis rates [46,47]. In fact, as soil amendments, potassium polyacrylate, polyacrylamide, and humic acid had been found to have a highly significant effect (p < 0.01) on potato photosynthesis characteristics [48] and biochar application increased the soil organic carbon to improve the leaf photosynthesis [49]. Taking all these together, it should be found that no matter higher fertilizers supply or soil amendments application could increase crop photosynthesis. However, in our study, the photosynthesis rate was not correlated with soil available phosphorus, while total phosphorus was significantly correlated with grain and photosynthesis rates. This may be because sandy soil has strong phosphorus adsorption rates, which may have decreased the effectiveness of the phosphorus fertilizer. Therefore, the application of biochar and soft rock may increase flowering photosynthesis through improving soil hydrolyzable nitrogen and total phosphorus to increase the grain yield.

Conclusions
Application of soft rock as a soil amendment in sandy land improved the capacity of soil holding water and fertilizer. This study suggested that the application of soft rock could increase the soil AP, AN TP, and TN, while the improvement in AP and AN from biochar treatment was not significant. Because of the improved fertilizer holding capacity, the application of soft rock increased the photosynthesis rate at flowering, which is an important factor in improving grain yield. This study showed the potential of soft rock as an amendment to improve soil properties and millet yield in the Mu Us Sandy land. Further research is needed to investigate how soft rock changes the soil properties to improve the nutrient holding and water holding ability of the soil.
Supplementary Materials: The following are available online at http://www.mdpi.com/2073-4395/10/5/669/s1, Figure S1: Schematic diagram of the field layout. NY, 90 kg N hm −2 and 45 kg P hm −2 and HY, 270 kg N hm −2 , and 135 kg P hm −2 . CK, sand only; C, sand and biochar; SR, sand and soft rock; SRC, sand, biochar, and soft rock. The blue line indicated the PVC pipe for irrigation, and slash indicated the buffer zone; Figure S2: Effects of biochar and soft rock treatments on soil total nitrogen (TN) and phosphorus (TP) at two fertilizer levels in 2019. CK, sand only; C, sand and biochar; SR, sand and soft rock; SRC, sand, biochar, and soft rock; NY, normal yield pattern; HY, high yield pattern.