Residual Effect of Bentonite-Humic Acid Amendment on Soil Health and Crop Performance 4–5 Years after Initial Application in a Dryland Ecosystem

Abstract

Degraded soils (including salinized, eroded, and low organic matter) resulting from natural and human effects are universal in arid and semi-arid regions all over the world. Bentonite and humic acid (BHA) are increasingly being tested to remediate these degraded lands, with potential benefits on crop production and soil health. A field study was conducted to quantify the effects of BHA application at six rates (0, 6, 12, 18, 24, and 30 Mg ha⁻¹) on (i) dynamic changes in soil properties and (ii) oat crop productivity parameters in a dryland farming ecosystem. The specific objective of this paper was to determine the residual effects four to five years after a one-time BHA application on soil health and crop performance. The findings demonstrated that with the increasing rates of one-time BHA application, soil profile water storage displayed a piecewise linear plus plateau increase, whereas soil electrical conductivity, pH, and bulk density were all reduced significantly (p < 0.05) in the 0–20 cm and 20–60 cm layers. The improved soil environments gave rise to an increased activity of soil enzymes urease, invertase, and catalase that, respectively, reached peak values of 97%, 37%, and 32% of the control at the rates of 18 to 24 Mg BHA ha⁻¹. In turn, this boosted soil nutrient turnover, leading to a 40% higher soil available P. Compared with the control treatment, application of BHA at the estimated optimum rate (roughly 24 Mg ha⁻¹) increased grain yield by 20%, protein yield by 62%, water use efficiency by 41%, and partial factor productivity of N by 20%. The results of this study indicated for the first time that a one-time BHA application would be a new and effective strategy to combat land degradation and drought, and promote a sustainable soil micro-ecological environment in dryland agroecosystems under a varying climate scenario.

1. Introduction

Globally, soil degradation (including salinized, eroded, low organic matter) resulting from changing climate scenarios and poor farming practices affects more than 75% of Earth’s land surface [1], reducing food production security and increasing irrigation demand in dryland farming regions [2]. Various soil amendments have been evaluated to alleviate degraded soil in dryland regions. The performance of soil amendment applications such as bentonite and humic acid (HA) varies among land uses, soil types, and ecosystems of the world.

Bentonite, an aluminum phyllosilicate clay consisting 85–90% of montmorillonite, can be used as a natural and non-toxic water absorbing soil amendment [3]. Bentonite has been widely used as a lubricant in drilling mud and boreholes, as an agent for stabilization of soil structure in construction sites [4,5], for improvement of soil rheological or sealing properties in geo-environmental applications, and for absorption and fixation of heavy metals in waste water treatment [6]. Some studies on the use of bentonite as an agricultural soil amendment demonstrated a significant improvement in soil water storage [3] and reduction in soil bulk density [7] in the 0–60 cm soil profile in fragile semi-arid agricultural areas.

Bentonite-induced soil quality improvements have been shown to enhance the establishment of agricultural field crops such as maize (Zea mays L.), millet (Panicum miliaceum L.), and squash (Cucurbita pepo L.) [3,7,8].

Humic acid (HA) is a chemical constituent of soil organic matter [9]. Previous pot studies found that HA application with irrigation water had a positive effect on oat (Avena nuda L.), winter rapeseed (Brassica napus L.), and leek (Allium ampeloprasum L.) root and shoot growth [10,11,12]. Application of a combination of the sodium form of bentonite with humic acid (BHA) has been evaluated as a new water saving and ecological restoration strategy on soils with low clay and organic matter in the arid and semi-arid region of northern China [7]. However, the impact of BHA application on soil nutrient cycling and enzyme activity has not been fully tested in soil-crop agroecosystems.

The valuable role of BHA application on soil quality may include the mitigation of soil compaction, enhancement of water retention, and their combined impact on soil microorganisms [13]. BHA has the potential to be used as an adaptive strategy to improve soil microbiological properties in both the top and deeper layers [14]. Some studies have investigated the response of soil microbiological properties to changes of soil physicochemical properties in the surface soil layers following the application of bentonite amendment [15,16].

Currently, climate change is leading to an increase in the extent of temperate drylands, intensification of drought events, and reduction of plant and microorganism available water in deeper (>20 cm) soils during the crop growing season [17]. However, there is still a lack of information on the amendment effect on soil enzyme activities and, consequently, nutrient availability in the soil profile. Soil enzyme activity has been quoted as a viable bio-indicator of soil microbiological activity because of its very sensitive response to spatial variations of soil health [18]. Soil enzymes (catalase, invertase, and urease) play a vital role in soil ecosystems and contribute to global soil organic matter turnover and nutrient cycling throughout the soil profile [13,19]. Thus, data on enzyme activity as a soil microbiological index in soils treated with BHA or other similar materials might provide an insight into the contribution of the natural water-absorbing amendments in enhancing soil quality.

Changes of soil physicochemical characteristics, for instance, soil pH and nutrient status, following amendment treatments of biochar and manure were not always observed [18,20]. The effectiveness of BHA in enhancing soil quality, however, depends on a number of factors, including soil texture, application rate, and accompanying soil and crop management [21]. It has been suggested that excessive tillage and repeated irrigations may degrade the effectiveness of bentonite [3,7,22]. On the other hand, bentonite and HA are extremely resistant to microbial degradation, which implies that, although the effectiveness of BHA may diminish over time, annual reapplication of BHA might not be essential. Residual effects of bentonite application on soil water-holding capacity and water status have been shown in a few field studies [3,7]. Rollins and Dylla estimated that the effectiveness of a single bentonite application declined at a rate of about 14–20% per year, but this varied with frequency of damage by wind erosion and wetting and drying cycles [22]. In contrast, manure and biochar are both intensively studied soil amendments, but the correlation between their long-term impacts and the short-term effects on soil physicochemical and microbiological characteristics have not been determined [18]. The difference is a result of the aging processes and the development of unbalanced conditions in chemical exchange and biological activity in the soil amendment systems [23]. However, information is still scarce on residual effect over time of a one-time BHA application.

Oat, as a valuable health-food, is gradually being exploited for human consumption worldwide [24,25,26]. It has generally been regarded a low-input cereal crop and has been conventionally grown on farms with moderate or low management [24]. Meanwhile, oat has an acute sensitivity to water stress that occurs with different and unpredictable frequencies and length across the growing season [27,28]. Declining water reserves in deep soils with the rise of drought are severely curtailing productivity of oat and other cereal crops [29]. One potential method of managing the risk of oat crop losses in dry seasons and optimizing the crop performance in favorable seasons is the application of BHA [7] which both conserves water and releases nutrients into their available forms [10].

We hypothesized that the effects of a single low-rate BHA application on crop water use efficiency and soil water storage might be ephemeral because of the breakdown of HA and the loss of bentonite through wind erosion and possible dilution by tillage or transport to deeper depths by water percolation through the soil profile during heavy rainstorms. In contrast, a higher rate of BHA application may lead to a longer-term improvement of soil enzyme activity, water storage, and nutrient cycling throughout the soil profile and, therefore, result in a positive effect on crop productivity even at 4–5 years after application. To examine this hypothesis, a 5-year field experiment with oat crop was conducted in a typical dryland farming area to compare one-time application of BHA at six rates.

The specific aim of this research was to assess the effectiveness of a single BHA application at different rates in a dryland farming ecosystem for improving (i) soil health indicators (soil water storage, nutrient availability, and enzyme activities), (ii) oat crop performance parameters (grain yield, protein, partial factor productivity of nitrogen and water use efficiency) at 4 and 5 years after the initial application, and (iii) determine the optimum rate of one-time BHA application.

2. Materials and Methods

2.1. Study Site

The experiment was conducted from 2011 to 2015 at the Yijianfang village of Qingshuihe County Research Centre, Hohhot, Inner Mongolia, in north central China (Long. 111.7 °E, Lat. 39.5 °N, ca. 1374 m above sea level). The experimental site has a sandy loam soil texture. The average annual rainfall is 365 mm. The descriptive baseline soil properties are shown in

Table 1. Selected soil chemical properties in the experimental site.

Property	Value
pH	7.8
SOM (g kg−1)	10.3
AN (mg kg−1)	45.1
AP (mg kg−1)	7.4
AK (mg kg−1)	124

. Further detail on the soil properties and climate can be found in our previous paper [30].

The region is part of the Loess Plateau along the Great Wall in north central China, where the main broad group of soils formed is the Huangmian soil (Calcic Cambisols in the FAO soil classification system), ranging in depth from 30 to 80 m. The detailed description of the soils of the Loess Plateau in China is given in a previous paper [31].

2.2. Characteristics of BHA Soil Amendment

The commercial BHA amendment was a mixture of bentonite, humic acid (HA), Na₂CO₃, and cellulose with a ratio of 91:6:2:1 [7]. Briefly, bentonite was mined from a local source by Trirock Co., Ltd., Naiman County, Tongliao, Inner Mongolia of China. It contained (wt/wt): 73.2% SiO₂, 11.4% Al₂O₃, 2.67% CaO, 2.58% K₂O, 1.05% MgO, 0.31% Na₂O, and 0.29% Fe₂O₃. Chemically, the free HA fraction occupies 38.3% (wt/wt) of the total HA content. The BHA mixture is marketed to local farmers as a soil amendment for coarse-textured sandy soils.

2.3. Experimental Set-Up and Field Management

The experiment consisted of six treatments and was carried out as a randomized complete block (RCB) design with three replications under the condition of no irrigation, relying upon natural rainfall. The treatments were (1) control or check with no applied BHA and application of BHA at rates of (2) 6 Mg ha⁻¹, (3) 12 Mg ha⁻¹, (4) 18 Mg ha⁻¹, (5) 24 Mg ha⁻¹, and (6) 30 Mg ha⁻¹. The BHA was broadcast uniformly on the soil surface and incorporated into the 0–15 cm soil profile by a rotary cultivator before planting on May 2011. All plots, with or without BHA application, received the same management practice of annual spring cultivation at about 15 cm depth with a rotary cultivator. Each plot had an area of 6 m × 5 m.

Oat seeds were sown at the depth of approximately 4 cm in late May and harvested in mid-September in each year (2011–2015). Each plot consisted of 20 rows of oat crop with row spacing 25 cm and sowing density of 375 seeds m^–2. In each year, fertilizer N as urea at 34.5 kg N ha⁻¹ and P as diammonium phosphate (containing 27 kg N ha⁻¹ and 69 kg P ha⁻¹) were broadcast and incorporated in all plots prior to seeding. Additional urea (69 kg N ha⁻¹) was manually applied as broadcast at GS 31.

The dates of key phenological stages were recorded for each plot when 50% of the oat crop achieved the certain stage, according to Zadoks’ scale. These included seedling (GS 12), jointing (GS 31), heading (GS 55), grain filling (GS 75), and maturity (GS 92) [32]. The experiment followed local management recommendations for weed control and other agronomic practices.

2.4. Sampling and Measurements

We sampled soils five times during the growing season (GS 12, GS 31, GS 55, GS 75, and GS 92) in 2014 and 2015. Samples were collected for three replicates for each treatment; three samples were taken midway between two rows within each plot. After removing any visible surface debris, a 5 cm diameter handheld auger was used to collect soil samples at depths of 0–10, 10–20, 20–40, 40–60, 60–80, and 80–100 cm. The cores from each plot were combined to obtain one composite sample for each layer (i.e., 0–10, 10–20, 20–40, 40–60, 60–80, and 80–100 cm). Each soil sample was thoroughly mixed, homogenized, and then divided into two parts. One part was air-dried, sifted through a sieve (2 mm), and stored for further tests of soil electrical conductivity, pH, AN, AP, AK, and SOM. The second part was immediately stored in a cooler (4 °C), transported to the laboratory, and assessed for the activities of catalase, invertase, and urease within two weeks after soil sample collection.

2.4.1. Determination of Soil Physicochemical Properties

SBD (g cm^–3) was measured using a 10 cm diameter and 5 cm high cutting ring. SEC (mS cm⁻¹) and pH were determined in suspensions of distilled water and soil (5:1); respective measurements were made using hand-held conductivity and pH meters. SWC (%) was determined by oven-drying (105 °C) samples until constant weight. SWS (mm) at each layer was evaluated according to Equation (1):

$$\mathrm{SWS}\left(\mathrm{mm}\right)=\mathrm{SWC}\left(\%\right)/100\times \mathrm{SBD}\left({\mathrm{g} \mathrm{cm}}^{–3}\right)/{\rho }_w\left({\mathrm{g} \mathrm{cm}}^{–3}\right)\times \mathrm{d}\left(\mathrm{mm}\right)$$

(1)

where SWC is soil water content, $\mathrm{SBD}$ is soil bulk density, ${\rho }_w$ is water density, and $\mathrm{d}$ is soil depth.

AP, AN, and AK were quantified by NaHCO₃-molybdenum antimony colorimetric, NaOH alkali diffusion, and NH₄OAc-flame photometry methods, respectively; SOM was measured by oxidation with the K₂Cr₂O₇ and exothermic heating method [33]. All of the reagent chemicals had a purity of ≥99% and were purchased from Tang-Shan SanYou Co., Ltd., Tangshan, China.

2.4.2. Soil Enzyme Activities

Soil urease activity was determined by measuring the reduction of urea following incubation [34]; the urease activity was calculated and expressed as μg NH₃-N g⁻¹ soil (24 h)⁻¹ [35].

Soil invertase activity was assayed by monitoring the consumption of sucrose following incubation [34]; invertase was expressed as mg glucose g⁻¹ soil (24 h)⁻¹ [36].

Soil catalase activity was determined by measuring the reduction of H₂O₂ for soil suspended in a reaction solution [34]; the catalase was expressed as 0.1 mol L⁻¹ KMnO_{4 mL} g⁻¹ soil (30 min)⁻¹ [37].

2.4.3. Plant Productivity and Grain Protein

To measure plant aboveground BY and GY, two 1.0 m² areas of each plot were harvested in 2014 and 2015 after the oat reached physiological maturity. Grain samples of each plot (5 g) were pulverized to pass through a 1 mm mesh screen. After being dried at 75 °C until constant mass, a subsample of 1 g was digested for the assessment of total N content, using the Kjeldahl digestion as well as analytical method. GP was determined by multiplying the grain N concentration by 6.25 [38]. TGP was estimated as the product of GY and GP concentration. Total growing season ET (mm) was calculated according to Wang et al. (2011). WUE (kg ha⁻¹ mm⁻¹) was calculated from GY and ET [39]. PFPN was calculated by dividing the GY by applied N fertilizer [40].

2.5. Data Analysis

One-, two-, or three-way analyses of variance (ANOVA) were performed, respectively, for one-time point variables (e.g., yield, protein), multi-time point variables (soil nutrient dynamics), and variables involving soil depth and growing season (e.g., enzyme activity), with the MIXED procedure of SAS, using the Type III and REML estimation methods (SAS Institute Inc., Cary, NC, USA). Rate of one-time application of BHA in 2011 was considered as a continuous variable for analyzing trends, and the measured dependent response variables were fitted to piecewise linear plus plateau models using the NLIN procedure in SAS. The BHA rate at which a plateau was reached was considered the optimal rate for the measured dependent response variable; additional BHA would not increase the response variable. Redundancy analysis (RDA) methods based on the ggvegan package of the R software (R Core Development Team, R Foundation for Statistical Computing, Vienna, Austria) were performed to elucidate the relationships among soil parameters and plant characteristics.

3. Results

The ANOVA showed that BHA treatment, soil depth, growth stage, year, and their interactions all had a highly significant effect (p < 0.01) on all of the soil parameters measured at each depth and growth stage (SWS and three enzyme activities), soil parameters measured once a year (at GS 92) at different depths (SEC, pH, SBD AN, AP, AK, and SOM) (

Table 2. Analysis of variance (ANOVA) for the effects of BHA treatments and soil depths on soil pH, SBD, SEC, and nutrients.

Soil Parameter	Year	T (Treatments)	D (Depths)	T×D (Interaction)
pH	2014	**	**	**
	2015	**	**	**
SEC	2014	**	**	**
	2015	**	**	**
SBD	2014	**	**	**
	2015	**	**	**
AN	2014	**	**	**
	2015	**	**	**
AP	2014	**	**	**
	2015	**	**	**
AK	2014	**	**	**
	2015	**	**	**
SOC	2014	**	**	**
	2015	**	**	**
SOM	2014	**	**	**
	2015	**	**	**

** indicates highly significant effect (p < 0.01).

and

Table 3. Analysis of variance (ANOVA) for the effects of BHA treatments, soil depths, and growth stages on soil water storage and enzyme activities.

Factor	SWS		Urease		Invertase		Catalase
	2014	2015	2014	2015	2014	2015	2014	2015
T	**	**	**	**	**	**	**	**
D	**	**	**	**	**	**	**	**
T×D	**	**	**	**	**	**	**	**
GS	**	**	**	**	**	**	**	**
T×GS	**	**	**	**	**	**	**	**
D×GS	**	**	**	**	**	**	**	**
T×D×GS	**	**	**	**	**	**	**	**

** indicates highly significant effect (p < 0.01). Factors: T—treatment; D—depth; GS—growth stage.

), and crop parameters measured once per year (GY, TGP, ET, WUE, PFPN, and BY).

The relationships between all of the measured parameters and BHA treatment rate displayed a linear or piecewise linear plus plateau nature. The values reported in the results section are the parameters of the linear or piecewise linear models of the respective dependent variable against level of BHA treatment rate. The slopes of the linear sections have units of change in the dependent variable per Mg BHA ha⁻¹; the units of the plateaus were the dependent variable at the plateau level of BHA. The term plateau is used to describe the region of no further change in the dependent variable with further increase in BHA rate; depending on the dependent variable, plateaus occurred following an increase or decrease of the dependent variable.

3.1. Enzyme Activity

The enzyme activities generally tracked SWS in both growing seasons (

Table 4. The parameters of the 2014 and 2015 linear or piecewise linear models of soil water storage (SWS), urease, invertase, and catalase activity in different depths plotted against one-time bentonite-humic acid (BHA) application rate.

		Intercept		Slope *		Plateau Initiation
		2014	2015	2014	2015	2014	2015
SWS	0–10	10.2 ± 0.99	6.8 ± 0.87	0.05 ± 0.007	0.03 ± 0.005	24 ± 3.2	-
	10–20	9.5 ± 0.62	8.6 ± 0.66	0.10 ± 0.004	0.07 ± 0.005	24 ± 9.9	29 ± 1.7
	20–40	21.3 ± 1.22	16.7 ± 1.71	0.27 ± 0.011	0.19 ± 0.015	22 ± 6.81	23 ± 1.4
	40–60	18.7 ± 0.89	15.5 ± 0.92	0.18 ± 0.005	0.09 ± 0.005	-	-
Urease	0–10	0.83 ± 0.04	0.98 ± 0.06	0.013 ± 0.003	0.02 ± 0.005	24 ± 4.8	20 ± 4.0
	10–20	0.67 ± 0.02	0.72 ± 0.06	0.017 ± 0.002	0.03 ± 0.001	24 ± 2.1	16 ± 3.3
	20–40	0.39 ± 0.03	0.33 ± 0.02	0.016 ± 0.003	0.008 ± 0.001	24 ± 3.1	-
	40–60	0.35 ± 0.02	0.28 ± 0.01	0.011 ± 0.002	0.005 ± 0.001	24 ± 2.8	-
Invertase	0–10	24.9 ± 1.48	21.0 ± 0.60	0.33 ± 0.08	0.33 ± 0.03	-	-
	10–20	17.9 ± 0.81	15.9 ± 1.22	0.18 ± 0.04	0.50 ± 0.07	-	-
	20–40	7.5 ± 0.52	5.5 ± 0.41	0.12 ± 0.05	0.19 ± 0.02	24 ± 7.4	-
	40–60	6.5 ± 0.26	5.3 ± 0.39	0.07 ± 0.01	0.12 ± 0.02	-	-
Catalase	0–10	15.7 ± 0.45	13.9 ± 0.37	0.08 ± 0.03	0.11 ± 0.02	24 ± 9.4	-
	10–20	14.5 ± 0.35	12.8 ± 0.26	0.11 ± 0.02	0.12 ± 0.01	-	-
	20–40	12.5 ± 0.47	10.7 ± 0.42	0.12 ± 0.04	0.12 ± 0.04	23 ± 6.6	24 ± 5.8
	40–60	12.7 ± 0.34	10.7 ± 0.46	0.13 ± 0.02	0.10 ± 0.03	-	-

* Slopes of the regression lines were significant (p < 0.05) for all parameters. Units of intercept are the units of the corresponding dependent variable; units of slope are quotient of dependent variable and rate of BHA; units of intersection of linear and plateau sections (plateau initiation) are Mg BHA ha⁻¹. Plateau initiation for models that did not converge to piecewise linear are indicated by “-”.

and Figure 1B–D). Specifically, with each Mg BHA ha⁻¹ of application, the improved activities ranged from 5 to 30 μg NH₃-N g⁻¹ soil (24 h)⁻¹ for urease, from 0.1 to 0.5 mg glucose g⁻¹ soil (24 h)⁻¹ for invertase, and from 0.05 to 0.19 mol L⁻¹ KMnO_{4 mL} g⁻¹ soil (30 min)⁻¹ for catalase (Table 4).

Plateaus in enzyme activities were reached more often than in SWS (Table 4 and Figure 1). In all cases, the highest enzyme activity occurred at approximately 24 Mg BHA ha⁻¹ (ranging from 16 to 28 Mg BHA ha⁻¹). At this rate, the highest corresponding increase compared to the control was up to 97% for urease, 39% for invertase, and 32% catalase in the wet year (2014).

Overall, BHA application showed a large effect on most of the enzyme activity parameters, particularly in the 20–40 cm soil layer, during the mid- to late-growing season. It corresponded to the rapid crop development stages in both 2014 and 2015.

3.2. Soil Physicochemical Properties

The regression of SWS displayed a significant piecewise linear (slope plus plateau) shape against BHA application rates (Figure 1A), with the slopes varying from 0.05 to 0.27 mm/(Mg BHA ha⁻¹) in 2014 and from 0.03 to 0.19 mm/(Mg BHA ha⁻¹) in 2015 (Table 4). The response curves did not always show plateaus, but when there was a plateau, it often started at the BHA application rate of about 24 Mg ha⁻¹ (range of 22 to 29 Mg BHA ha⁻¹). On average, the maximum increase in SWS at the highest BHA application rate was 12 to 29% over the no BHA control treatment.

At the 0–60 cm soil layer, SEC, pH, and SBD displayed a decreasing trend with increasing BHA application rates in both 2014 and 2015, although the rates of decline varied among soil depths and between the two years (Figure 2A–C). Specifically, SEC of 24 Mg BHA ha⁻¹ treatment was reduced, respectively, by 0.107 and 0.133 mS cm⁻¹ in 0–10 cm, 0.296 and 0.291 mS cm⁻¹ in 10–20 cm, 0.194 and 0.161 mS cm⁻¹ in 20–40 cm, and 0.137 and 0.123 mS cm⁻¹ in 40–60 cm soil layers, compared with those of the no BHA treatment (Figure 3A). Average soil pH was reduced by 0.2 units in both 2014 and 2015 for treatments of BHA application compared to the no BHA treatment (Figure 3B). SBD of the 0–60 cm layer showed a consistent decreasing trend in 2014 and 2015 with increasing rates of BHA application (Figure 3C). For example, in the 10–20 cm soil layer, SBD was significantly reduced by 0.01–0.09 g cm^–3 in 2014 and 0.01–0.19 g cm^–3 in 2015 for all BHA treatments, compared with the no BHA treatment. The estimated BHA application rate for the most effective response often occurred at 21 Mg ha⁻¹ (ranging from 7 to 30 Mg BHA ha⁻¹). At this rate, the corresponding reduction was up to 34% for SEC, 5% for pH, and 12% for SBD, compared with the no BHA treatment.

3.3. Soil Available Nutrients

In both 2014 and 2015, soil nutrient concentrations and SOM generally tracked SWS (Figure 3A–D), with increment rates varying from 0.4 to 0.9 mg kg⁻¹ for AN, 0.1 mg kg⁻¹ for AP, 0.4–1.8 mg kg⁻¹ for AK, and 0.03–0.14 g kg⁻¹ for SOM, with each Mg BHA ha⁻¹ application. Plateaus were not reached for AN and only sometimes for AP, AK, and SOM. When there was a plateau, it often occurred at the rate of about 20 Mg ha⁻¹ (range 5 to 30 Mg BHA ha⁻¹). At this rate, the corresponding increases were up to 44% for AP, 17% for AK, and 45% for SOM, compared with the control.

3.4. Yield and Grain Protein

Over the two growing seasons, GY, TGP, and BY (Figure 4A,B,F) were all influenced by BHA treatments. There was a general linear increasing trend, with the slopes of 22 kg GY ha⁻¹ and 8 kg TGP ha⁻¹ in both 2014 and 2015, while the slope of BY was 54 kg ha⁻¹ in 2014 and 48 kg ha⁻¹ in 2015. All slopes are respective units of the dependent variable per Mg BHA ha⁻¹. The estimated highest GY, TGP, and BY were reached, respectively, at 22, 24, and 24 Mg BHA ha⁻¹ in 2014 (a wet year) and at 25, 26, and 27 Mg BHA ha⁻¹ in 2015 (a dry year). At this rate of BHA application, GY, TGP, and BY were, respectively, 17, 53, and 15% in 2014 and 20, 62, and 15% in 2015, greater than the control.

3.5. Water and Nitrogen Use

With increasing the rate of BHA application, WUE was increased linearly with the slope of 0.2 kg yield ha⁻¹ mm⁻¹ in both 2014 and 2015 (Figure 4D). PFPN displayed the same response trend as the WUE, while ET showed the opposite trend, with the corresponding reduction of ET at 3.1 mm in 2014 significantly greater than the 1.4 mm reduction in 2015 (Figure 4C). As pointed out previously, these slopes are all per Mg BHA ha⁻¹. At the estimated shoulder application rate of 16 and 19 Mg BHA ha⁻¹, the WUE was, respectively, enhanced by 41% and ET was reduced by 17% in the wet year, compared to the control (Figure 4C, D). The variation of PFPN (Figure 4E) was consistent with GY.

3.6. Relationships between Crop Performance Parameters, Soil Physicochemical Properties, and Enzyme Activity

There were highly significant (p < 0.01) correlations among the crop performance parameters (GY, TGP, BY) and soil health parameters (SWS, urease, invertase, catalase, nutrients, SOM). Soil chemical and physical parameters SEC, pH, and SBD also displayed highly significant correlations with each other and correlated with other soil health parameters, soil water storage, available nutrients and enzyme activity, and with crop performance parameters. TGP had the closest correlation with the activity of soil catalase in 2014 and with AN and AP in 2015. BY and GY were more closely correlated with urease, AN, and AP in both 2014 and 2015 (Figure 5). Bi-plots from the redundancy analysis (RDA) showed a consistent trend where increasing rates of the BHA application increased the magnitude of RDA1 (Figure 5), demonstrating that application of BHA created a healthier soil environment and improved crop performance. A notable exception was greater RDA1 for BHA 24 than BHA 30 in 2014.

4. Discussion

4.1. Impact of BHA on Soil Health Parameters

Increase in SWS was deemed the most important effect of the BHA application, and it influenced other soil and crop response parameters. In this study, we demonstrated a significant effect of BHA at 4–5 years after its initial application, and such an effect was greater in 2014 (a wet year) than in 2015 (a dry year) (Table 4 and Figure 1). There was also a significant improvement in SWS in the 0–20 cm soil layer and at the deeper depths (20–60 cm) and corresponding reduction of water loss out of the rooting zone to the ground water (Table 4). The results of this study indicate that there was likely a significant amount of BHA remaining in the soil 4–5 years after initial application. The greater improvement in SWS of the deep soil layer, notably in the wetter year (2014) (Table 4), was mainly attributed to the sufficient rainfall during the growing season to supply the continuous water uptake by the crop with the excess percolating through the rooting zone to deeper depths. To a lesser extent, the increased SWS in both 2014 and 2015 may have reflected the corresponding retention of soil water from previous years (e.g., annual rainfall was 549.6 mm in 2013 and 456.2 mm in 2014). Some earlier studies on the long-lasting impact of BHA (“tiny reservoir”) application on sandy soil water storage [7,41] support our conjecture. At the high rates of application, BHA in the soil could absorb a large amount of water and suppress surface evaporation [42], resulting in reduced evapotranspiration (Figure 4C). This would foster greater and longer soil water retention in our semi-arid region than in other sites where greater rainfall or different soil textures and pedogenesis have diminished the effectiveness of the applied BHA in a few years after application [7].

Collectively, the increased SOM and improved soil physicochemical properties could provide a healthy and suitable living condition for soil microorganisms [13,43]. Improvement in soil microorganisms likely contributed to more timely and efficient nutrient turnover and cycling, leading to higher plant available N, P, and K concentrations [44,45,46]. The lower SBD in both the layer of incorporation (0–20 cm) and adjacent deeper layers (20–60 cm) (Figure 2) was likely associated with the migration of BHA to subsurface soil layers [47]. This indicates the importance of physicochemical adjustment for soil micro-ecological environment improvement in the dryland agroecosystem.

4.2. Plant Response to BHA

In this field study, we demonstrated a clear link between BHA induced improvements in SWS and other soil health indicators (enzyme activity and available soil nutrients) and grain yield and grain protein (Figure 5). The improved crop response to BHA in both wet and dry years was likely both a direct effect of the improved plant-water status and nutrient uptake due to the increased transpiration [48] and an indirect effect through an improvement in soil health parameters such as increased microbial activity that led to increased plant available nutrients [13]. Increasing BHA application rates resulted in higher yields (Figure 4). This in turn led to both improved water use efficiency (WUE) and improved partial factor productivity of nitrogen (PFPN) in both wet and dry years (Figure 4D,E).

4.3. Optimization of the BHA Application Rate

In this study, some of the dependent variables exhibited a linear response to the rate of BHA application, with a plateau, while others exhibited only a linear response. The absence of a plateau indicated that the maximum response had not yet been achieved with the maximum BHA rate of 30 Mg ha⁻¹ used in our experiment. However, it has been noted that too much BHA would likely be detrimental, as demonstrated in an earlier study [49], where failure of plant growth was found with large amounts of bentonite application in mining areas. In this study, 24 Mg BHA ha⁻¹ was identified to be the lower end of the optimum range, while the maximum rate of 30 Mg BHA ha⁻¹ appeared not to have reached the upper limit of the optimum range for some of the soil and plant growth parameters. Ideally, BHA rates in an experiment would be distributed equally on either side of the optimum. The failure of identifying the plateaus of soil health indicators in the dry year (2015) appeared to indicate that higher rates of BHA may be required, or it may simply reflect the reduced effectiveness of BHA due to the limited water under the adverse conditions. Therefore, future experiments including higher rates of BHA may help to more clearly identify the optimum application rate.

4.4. Longevity of BHA

Bentonite is a stable mineral which could be lost by erosion or downward migration through the soil profile, while the HA is organic and could be decomposed. Rollins and Dylla [22] estimated that the effectiveness of bentonite for sealing dams declines at a rate of ∼20% year⁻¹ due to the loss by wind and water erosion, which would indicate no residual effect after five years [22]. Although the rate of degradation/disappearance of BHA after its application was not measured in this study, we demonstrated a substantial residual effect at 4–5 years after the BHA application. This reflected that our test condition (incorporation in the soils) was suitable for the semi-arid region and is in contrast to the estimated rate of BHA broken down or lost under the wet conditions by Rollins and Dylla [22]. If the rate of loss of effectiveness was known, then appropriate recommendations could be made for farmers to reapply a portion of the BHA after several years to ensure its continued function in improving soil and agroecosystem services.

The rate at which the effectiveness of BHA declines is a function of the site’s climate, soil characteristics, and ecosystem and deserves further study. Other soil amendments, e.g., biochar and manure, can be used for similar purposes in improving crop productivity, but the availability of biochar is limited in our study area. An optimum biochar application rate of around 30 Mg ha⁻¹ (roughly 4761 USD ha⁻¹) is about 15 times the cost of a single 30 Mg ha⁻¹ (317 USD ha⁻¹) application of BHA [50]. On the other hand, the active component of biochar is stable and may only require a single application.

5. Conclusions

For the first time, we demonstrated that one-time BHA application improved and sustained better agricultural ecosystem balance and crop productivity. This was achieved through effective soil hydrologic regulation, soil enzyme activity, and nutrient turnover in topsoil and subsoils, thereby leading to the efficient use of soil water and nutrient by the oat crop and the corresponding increases in grain protein, grain yield, water use efficiency, and partial factor productivity of nitrogen. The improvements measured four to five years after the initial application were linear with the BHA application rates, and many, but not all, response parameters reached a plateau or saturation point at approximately 24 Mg ha⁻¹. Our study implies that one-time BHA application may serve as a new sustainable strategy to improve soil health and crop productivity in arid and semi-arid regions. However, it should be pointed out that a successful strategy for drought abatement in one crop, soil, and site-specific agroecosystem may have little effect in other soils and ecosystems; this aspect needs further study. Further research at higher application rates is also required to better define the optimum application rate, to measure the degradation and long-term stability of the BHA amendment, and to determine the economics under variable weather conditions.