Effects of Rotating Rice with Upland Crops and Adding Organic Amendments, and of Related Soil Quality on Rice Yield in the Vietnamese Mekong Delta

Abstract

In the Vietnamese Mekong Delta, soil quality and crop yield are steadily declining under rice monocultures with three crops per year. The objective of this study was to evaluate the medium-term effects of rotating rice with upland crops and adding organic amendments on rice yield, and to relate this to soil quality. A field trial with split-plot design including two factors and three replicates was carried out from 2017 to 2020, over the course of nine consecutive cropping seasons. Crop rotations and organic amendments were applied as main-plot and subplot factors, respectively. The rotations were (1) rice–rice–rice (R–R–R), (2) soybean–rice–rice (So–R–R), and (3) sesame–rice–rice (Se–R–R), while organic amendment treatments included (i) no amendment (NO-AM), (ii) compost of rice straw and cow manure (RS+CM), and (iii) sugarcane compost (SGC); the composts were applied at a rate of 2.0 t ha⁻¹. The rotation cycle started with the so-called spring–summer (SS) season, followed by the summer–autumn (SA) season and ending with the winter–spring (WS) season. Rice yield significantly (p < 0.05) increased under organic amendments after nine growing seasons (2019–2020 WS), with an increment of 5.1% for RS+CM (7.07 ton/ha) and 6.1% for SGC (7.14 ton/ha). Contrary to our expectation, rotations with upland crops did not significantly increase rice yield. Rice yield significantly and positively correlated with an integrated soil quality index–SQI (r = 0.85) for the topsoil (0–15 cm), but not for the subsoil (15–30 cm). The increased availability of soil nutrients (Si and marginally also P) and improved soil physical properties probably induced by organic amendments, along with other soil properties under study, cumulatively attributed to enhanced rice yield. Repeated organic amendments thus becomes an effective management practice in improving soil quality under rice-based systems and could be applied to sustain rice yield in rice-producing regions with similar soil types and climatic conditions. Use of a SQI involving several soil quality indicators enables us to quantify the overall importance of soil fertility for rice yield versus other factors, and it provides an effective means of quantifying the integrated effect of improved management. Moreover, integrating a wide range of soil quality indicators in a SQI ensures its applicability across diverse settings, including different crop rotations and various soil types.

1. Introduction

Rice is a staple food in the Vietnamese Mekong Delta, the third largest delta in the world [1]. The delta contributes to approximately 50% of the national rice production and is the largest rice exporting region of the country [2]. Irrigated triple rice is the dominant farming system [3] and resulted from the intensification of rice production to meet domestic consumption and export demand [2]. However, the sustainability of this system has been threatened. Over the past decades, rice yield decline in the Mekong delta has been reported [3], and this could be due to soil quality deterioration [4,5,6,7]. Elsewhere, the yield of long-term rice monoculture systems was likewise found to decline [8,9,10,11,12,13] or stagnate [14,15]. Various reasons have been given to explain these declining or stagnating yield trends, and they are mainly soil related, although some have reported that the declines in rice yield are attributed to the lack of adaptation of cultivars to the changing environment [10,12,13]. Firstly, nutrient deficiency resulting from the cumulative imbalanced use of fertilizer has been identified as a reason for rice yield decline, including macro-nutrients such as P and S and micro-nutrients such as Zn and Cu [9]. Likewise, Hoa et al. [5] reported that the improper use of fertilizers resulted in a negative balance of K in the Mekong Delta. Moreover, burning residues or removing rice straw after harvest for various purposes, such as mushroom cultivation and for cattle fodder, are popular practices in the delta [16] that contribute to reduced nutrient recycling [17]. Provision of N to rice in paddy fields should be regarded somewhat separately from other nutrients as it is often sufficiently or even oversupplied by farmers but may still be deficient. A low nitrogen use efficiency (NUE) has been frequently reported for continuous lowland rice cropping systems, despite the overuse of N fertilizers [18]. Cassman et al. [8] suggested that the declining yield trends in long-term triple rice cultivation systems may mainly be attributed to a particularly low N supplying capacity of the soils in these systems. Similar problems for the intensive rice systems have been recognized in the Mekong Delta [19]. In addition to nutrient shortages, several other negative consequences of monoculture rice on soil quality have been documented by Fageria et al. [20], including Fe and Al toxicity. On the other hand, soil structural degradation in the form of soil compaction caused by intensive agricultural production hampers root access to deeper nutrients, therefore contributing to their deficiency [7]. In a recent study, Ladha et al. [13] suggested that yield decline can be reversed through improved management (e.g., N fertilization) and cultivars. However, these effects seem to be climate-dependent since a declining tendency in rice yield was still observed during the late wet seasons. The authors suggested that the observed continuous decline in rice yield would be attributed to biotic and abiotic stresses, including stresses linked with poor soil health.

In recent years, efforts have been made to improve or sustain soil health and rice yield, such as by the introduction of crop rotations and the use of organic amendments. Crop rotations have been applied and their benefit over monocultures has been widely reported in different agroecological areas [6,21]. For lowland rice production in particular, the inclusion of upland crops into the rotation could substantially alter soil into a more beneficial environment for the subsequent crop [6,7]. Variation in crop types may result in improved chemical, physical, and biological soil properties, and ultimately in improved crop yield [22] and reduced greenhouse gas emissions [21]; it can also bring about economic efficiency—compared to monocropping—by reducing water use and use of agrochemicals [4]. Positive effects of crop rotation on soil fertility may include increased total soil N and organic matter, and enhanced microbial activity [22]. Physically, crop rotations help improve aggregate stability and water-holding capacity, decrease bulk density, and reduce soil compaction; such improvements have been observed for the Mekong Delta as well, with associated beneficial effects on rice yield [3,7,23]. A farm household survey by Linh et al. [4] from 109 farmers suggested that crop rotations increased on-farm rice yield by 6% to 9%, depending on rotation sequence and cropping season. Use of organic amendments has been found to be an effective practice to improve soil fertility and crop yields, since these replenish soil organic matter and essential nutrients for plants [24,25]. In the Mekong Delta in particular, the studies of Watanabe et al. [26,27] revealed that addition of rice straw compost could improve rice yield and soil physical properties such as penetration resistance.

Although including upland crops and organic amendments have been widely realized as alternative cropping practices to rice monoculture, these systems can also present constraints in terms of sustainability. For instance, Witt et al. [28] reported that the sequestration of soil N and C declined when rice was rotated with upland crops such as maize. This was due to a higher rate of mineralization of C under crop rotation conditions than under continuous rice-cropping conditions. Linh et al. [6] reported that crop rotation significantly decreased the SOC content of the topsoil (0–10 cm) as compared to the continuous rice-cropping system in the Mekong Delta. According to Cass et al. [29], the long-term (21 years) rotation of rice with soybean resulted in a significant increase in soil bulk density and soil strength, while air-filled porosity was found to be significantly decreased. Meanwhile, Haynes and Naidu [30] reported that organic amendments at high rates in the long term might result in some detrimental effects on soil structure, caused by excessive accumulation of monovalent cations such as Na⁺ and K⁺. Likewise, sequestration of C brought forth by the use of organic amendments does not always result in improved N availability, as for instance Cassman et al. [31] found N supply to stagnate in locations where soil organic carbon (SOC) increased. Overall, rice-cropping systems that include upland crops and the use of organic amendments have shown beneficial effects on rice production as a result of changes in soil quality, but their effects in terms of sustainability remain questionable.

A new field experiment was thus set up to assess the potential of these cropping systems in the Vietnamese Mekong Delta, by evaluating the effects of the combined application of crop rotations and organic amendments on rice yield, while using a recommended inorganic fertilizer rate and a moderate dose of organic fertilizers. The specific objectives of this paper were to (i) test whether the introduction of one upland crop season in a triple rice rotation and the application of organic soil amendments could beneficially affect rice growth, and (ii) if so, to test whether yield increases could be linked to improved soil quality. As upland crops, soybean and sesame were tested, as these are economically valuable for and acceptable by Mekong Delta farmers. Inclusion of both crops has, moreover, been shown to potentially improve soil quality and rice yield in previous studies (e.g., Linh et al. [32]). As soil amendments, sugarcane compost and rice-straw mixed with cattle manure were used, as both are relatively readily available to local farmers. Rice straw with cattle manure is a by-product of farmers’ households that can be easily composted and can be made on-farm [33]. We hypothesize that already in the short term, after three years (or nine growing seasons), the application of soil amendments and the introduction of one upland crop will tend to positively affect rice growth, and that this relates to improved soil quality.

2. Materials and Methods

2.1. Description of the Study Site

The field experiment was set up in the My Loi Hamlet, Thien My Commune, Tra On District, Vinh Long Province (9°57′13.6″ N, 105°55′58.6″ E), in the Vietnamese Mekong Delta. The concerned field had been using rice for 40 years before initiation of the experiment in 2017. The area has a tropical monsoon climate with two distinct dry and rainy seasons. The dry season lasts from late October to May, while the rainy season lasts from June till late October. During the complete experimental period, the average annual precipitation was 1345 mm, and the monthly average air humidity, air temperature, and solar radiation were 87.9%, 27.1 °C, and 16.2 MJ m⁻², respectively. Monthly weather data, from January 2017 to January 2020, measured on the spot with a WS-GP2 automatic weather station (DeltaLINK software, version 3.8.2, Delta-T Devices Ltd., Cambridge, UK), are shown in Figure S1 (Supplementary Materials).

The soil in the experimental field is classified as Rhodi-Gleyic Luvisol [34] and is characterized by its high clay content (>45%) with silty clay texture [35]; it is strongly acidic (pH_KCl < 3.0 in the puddle layer). The plow layer had a low cation exchange capacity (CEC) (<14.0 meq(+) 100 g⁻¹) [36], in spite of its elevated organic matter content (55 g kg⁻¹) (which was however very much lower in the subsoil layers, at about 6 g kg⁻¹) [37]. Background values of the soil properties by horizon, measured to a depth of 180 cm before starting the field experiment, are shown in

Table 1. Physico-chemical characteristics of the soil profile before the commencement of the field experiment.

Soil Properties	Depth Interval
	0–15 cm	15–45 cm	45–100 cm	100–180 cm
pHH2O (1:2.5)	4.63	5.71	5.22	5.80
pHKCl (1:2.5)	3.0	4.23	4.21	4.03
H+ (meq 100 g−1)	0.24	0.01	0.01	0.02
K+ (meq 100 g−1)	0.23	0.25	0.32	0.47
Na+ (meq 100 g−1)	1.17	1.10	1.22	1.87
Ca2+ (meq 100 g−1)	7.46	9.06	8.46	6.80
Mg2+ (meq 100 g−1)	3.83	6.21	6.86	7.66
CEC (meq 100 g−1)	12.93	16.63	16.87	16.82
Organic matter (g kg−1)	55.2	6.0	6.0	9.4
Sand (g kg−1)	14	9	11	11
Silt (g kg−1)	535	484	457	453
Clay (g kg−1)	451	507	532	536
Texture	Silty clay	Silty clay	Silty clay	Silty clay
Bulk density (g cm−3)	1.04	1.43	1.36	1.27

. Detailed analysis methods are provided in Section 2.3.

2.2. Treatments and Experimental Design

In this study, a 3-year field trial was carried out from 2017 to 2020, spanning nine growing seasons (Figure 1). The trial was laid out in a split-plot design including two factors, with crop rotations and organic amendments as main-plot and subplot factors, respectively. The crop rotations were (1) rice–rice–rice (R–R–R), (2) sesame–rice–rice (Se–R–R), and (3) soybean–rice–rice (So–R–R). The organic amendment treatments included (i) no amendment (NO-AM), (ii) compost of rice straw and cow manure (RS+CM), and (iii) sugarcane compost (SGC)—a commercial organic product. The treatments were replicated three times, resulting in a total of 27 experimental plots (Figure 1). A split-plot design was chosen here, with crop rotations as the main factor, to overcome a potential border effect resulting from differences in water management between rice and upland plots. It should, however, be noted that by taking crop rotation as the main factor, the power to detect the effects of the main factor will be diminished compared to that of the subfactor [38].

The rotation cycle started with the so-called spring–summer season (SS, February–May), followed by the summer–autumn season (SA, June–September) and ending with the winter–spring season (WS, October–January). For the crop rotations, sesame (Sesamum indicum L.) and soybean (Glycine max L.) were grown in the SS season, followed by two rice crops (Oryza sativa L.) in the SA and WS seasons. The main plots of 17.5 m × 5.5 m (96.25 m²) were separated by bunds of 60 cm × 40 cm (width × height). Subplots of 5.5 m × 5.5 m (30.25 m²) were separated by bunds of 40 cm × 30 cm. Plastic sheets were inserted to a depth of 40 cm along the bunds to prevent lateral movement of water between adjacent plots.

For the rice crop, land preparation involved turning up the topsoil layer (0–15 cm) and breaking up the soil clods with a hoe. Next, puddling was performed under wet conditions to create a seedbed. Just before sowing, field water was drained, and the soil was levelled with a banana trunk. Pregerminated seeds were sown by broadcasting at a rate of 200 kg ha⁻¹, using a short-duration rice variety (IR50404) of 85–90 days. The upland crops, sesame (variety Me Den, 75–80 days) and soybean (variety MTD-748, 85–95 days), were sown in rows at a depth of about 4.0 cm and a planting density of 333.333 plants ha⁻¹ (20 cm x 30 cm spacing). No land preparation was performed in the upland crops. After rice and upland crops were harvested, stubbles and plant residues were manually removed from the field.

Compost was applied at a rate of 2.0 t ha⁻¹ crop⁻¹. Chemical fertilizers were applied to rice, sesame, and soybean at rates of 100/80N-20P-25K, 60N-26P-37K, and 30N-26P-25K kg ha⁻¹, respectively. For rice, N was applied at a rate of 100 kg ha⁻¹ in winter–spring, while a lower rate (80 kg ha⁻¹) was applied in spring–summer and summer–autumn. N, P, and K were applied as urea (46%N), superphosphate (16% P₂O₅), and potassium chloride (60% K₂O), respectively. Compost and P fertilizer were applied immediately before the sowing of rice, sesame, and soybean. For rice, N fertilizer was split into three doses and 20% was applied 7–10 days after sowing (DAS), 40% at 20–25 DAS, and 40% at 40–45 DAS. K fertilizer was split into two equal doses and applied at 20–25 DAS and 40–45 DAS. For sesame and soybean, N and K fertilizer were split into two equal doses, applied at 15–20 DAS and 30–35 DAS.

Rice plots were flooded over almost the entire cropping season, with field water being drained ~10 days before harvest. The water level was maintained at 1 to 5 cm above surface. Soybean and sesame plots were irrigated periodically, so that the soil was under moist conditions and the irrigation amount varied with the plant’s age. Weeds were removed manually, while pests and diseases were controlled by pesticides when necessary.

2.3. Sampling and Analysis of Compost and Soil during the Experiment

2.3.1. Compost Analysis

Chemical compost properties, including pH; EC; organic matter (OM); and total N, P, K, Ca, and Mg (

Table 2. Chemical properties of rice straw + cow manure (RS+CM) and sugarcane (SGC) composts used in the field experiment (2017–2020).

Parameters	RS+CM j	SGC
Organic matter (g kg−1) a	546	307
pHH2O (1:2.5) b	8.0	5.0
ECH2O (1:2.5) (mS cm−1) c	3.1	–
N (g kg−1) d	24.1	22.9
P (g kg−1) e	10.0	15.6
K (g kg−1) f	23.9	18.2
Ca (g kg−1) g	17.5	76.1
Mg (g kg−1) h	12.1	0.8
C:N ratio i	11.8	6.7

^a Walkley and Black [39]; ^b,c 1:2.5 ratio of soil:water extracts, measured with a glass electrode pH meter (model HI 8314, Hanna Instruments, Woonsocket, RL, USA) and EC meter (model Lab 960, Schott Instruments, Mainz, Germany), respectively; ^d Kjeldahl apparatus [40]; ^e–h acidic digestion [41]: ^e determined with a spectrophotometer (UV-1800, Shimazu Corporation, Kyoto, Japan), ^f–h determined with an atomic absorption spectrometer (iCE 3000 Series, Thermo Fisher Scientific Inc., Waltham, MA, USA); ⁱ the C-content, estimated as 50% of the OM content [42]; ^j averages from nine cropping seasons.

) were determined before each application for composts of rice straw and cow manure (RS+CM), while sugarcane compost (SGC), a commercial product, was only analyzed once.

2.3.2. Soil Sampling and Analysis

Soil sampling was performed in the 2019–2020 WS, i.e., after nine cropping seasons. After the harvest, both undisturbed and disturbed soil samples were taken within two depth intervals (0–15 cm, 15–30 cm) to determine physical and chemical properties. The undisturbed soil cores were taken in standard sharpened steel 100-cm³ cylinders with a dedicated auger in three replicates per plot, while the disturbed samples were collected from a mixture of three replicates per plot.

Disturbed soil samples were first air-dried, ground, and passed through a 2 mm sieve. The soil particle size distribution (sand, silt, and clay content) was analyzed on the disturbed samples using the pipette-sieve method [43], and soil texture class was then determined based on the USDA texture triangle [35].

Bulk density (ρ_b) was determined as dry soil mass per bulk volume of soil core [44], with dry mass being obtained at 105 °C for 24 h. Particle density (ρ_p) was determined as dry soil mass per volume of soil particles using the pycnometer apparatus [45]. Then, porosity (ϕ) was determined from the bulk density and particle density [46].

Saturated hydraulic conductivity (K_S) was measured on pre-saturated intact soil cores with a KSAT meter (METER Group, Pullman, WA, USA) in falling head mode, based on Darcy’s law [47]. The SWRC was determined with a HYPROP system (METER Group, USA) using pre-saturated soil cores and a METER Group ring adapter [48]. The measured data points were fitted with the PDI variant of the unconstrained van Genuchten model [48]. Volumetric water content at field capacity (−10 kPa) and permanent wilting point (−1500 kPa) [46] were extracted from the SWRC. Plant available water capacity (PAWC) was determined as the amount of water held between field capacity and permanent wilting point, air capacity (AC) as the pore volume corresponding to the volumetric water content between saturation and field capacity [46], and relative water capacity (RWC) as the ratio of field capacity to saturation water content [49]. The pore size distribution was determined according to the classification by Brewer [50]. Pores with equivalent diameters of >75 µm were classified as macropores, those with diameters of 30–75 µm as mesopores, and those with diameters of 5–30 µm as micropores. Macro-, meso-, and microporosity were then determined from the water contents at matric potentials between 0 and −4 kPa, −4 and −10 kPa, and −10 and −60 kPa, respectively, based on the capillary equation [49].

Soil penetration resistance (PR) was measured to a depth of 80 cm with 1.0 cm depth intervals using a hand-held electronic penetrologger (Royal Eijkelkamp, Giesbeek, The Netherlands). A cone of 30° top angle and 2.0 cm² base area was used. Measurements were replicated five times per plot. The measured data were then averaged for two 15 cm depth intervals to a depth of 30 cm. At the measurement of penetration resistance, soil moisture was determined at 10 cm depth intervals to a depth of 80 cm, to assure that the PR was measured near field capacity conditions. The percentage of water stable aggregates (WSAs) was measured by wet sieving apparatus using a single sieve (250 µm) (Royal Eijkelkamp) according to Kemper and Rosenau [51].

Soil pH was determined on soil:distilled water and 1M KCl suspensions (1:2.5 ratio). Electric conductivity was determined on a saturated paste extract (ECe) with distilled water. The pH and EC were measured with pH and EC meters, respectively. Soil organic matter (OM) content was obtained by the Walley-Black method [39]. Soil cation contents in exchangeable form (Ca²⁺, Mg²⁺, K⁺, and Na⁺) were measured by the BaCl₂-compulsive exchange method [52], and the extracts were analyzed by an atomic absorption spectrometer (Thermo Scientific, iCE 3000 Series). The content of H⁺ on the exchange complex was estimated from the difference between the total acidity (pH_KCl) and actual acidity (pH_H2O). The cation exchange capacity (CEC) was then estimated as the sum of all measured exchangeable cations. For the samples taken after the harvest, total C and total N were measured by a Leco CN analyzer. Since the soil was acidic with pH_H20 values ranging from 5.0 to 6.8 and free ofinorganic carbonates, total C was assumed to be equal to total OC (TOC). Available P and exchangeable cations (K, Ca, Mg, Fe, Mn, Cu, Zn, and Si) were extracted with NH₄-Acetate EDTA (pH 4.65) [53] and measured with an ICP-OES Spectrometer (iCAP 6000 Series, Thermo Fisher Scientific Inc., Waltham, MA, USA).

2.3.3. Grain Yield, Above-Ground Biomass, and Yield Components

Final rice grain yield and above-ground biomass were determined from a 5 m² area in the centers of the plots [54]. The rice plants were harvested, and grains were separated manually. The grains and straw were weighed, and sub-samples (100 g) were taken. The sub-samples were oven-dried at 70 °C [54] for 48 h to determine moisture content. The grain yield was calculated based on a standard moisture content of 14% [55]. The total above-ground biomass was calculated as the sum of the dry weights of grain and straw.

To determine the yield components of rice, plants were collected from an area of 0.25 m² [56] with two replicates per plot. The samples were then treated to determine yield components including panicle number, total grain number, filled and unfilled grain number, 1000-grain weight, and filled grain percentage. The 1000-grain weight was reported at a moisture content of 14% [55], as determined using a grain moisture meter (GMK-303).

2.3.4. Soil Quality Index

The soil quality index (SQI) is an overall indicator of sustainable land management, integrating various separate soil indicators that represent soil functions, including water infiltration; storage and supply; and nutrient storage, supply and cycling, as well as supporting biological activities [57]. In this study, a SQI was established using the 23 abovementioned soil properties, measured at both sampling depths, according to the procedure outlined by Andrews et al. [58]. Saturated hydraulic conductivity Ks was not included in the index given the high variation in the data, which could be attributed to the limited number of replicated within-plot samples in our study. We used the entire dataset to derive a SQI as this is suggested to be the better approach [57,59] compared with using a minimum dataset derived from statistical techniques such as principal component analysis. Furthermore, Vasu et al. [60] reported that an SQI based on expert opinion showed a better correlation with crop yield than one derived from an unsupervised variable selection and weighing method. Next, the soil quality indicator values were transformed to unitless scores and integrated into an index. For scoring the soil quality indicators, three scoring functions—“more is better”, “less is better”, or “optimum”—were applied with reference to maximizing rice yield. “More is better” is applied to soil quality indicators which are considered “good” when their value increases (e.g., available K). Conversely, “less is better” is applied to soil quality indicators that are considered “bad” when their value increases (e.g., penetration resistance). “Optimum” is assigned to those considered “best” at a certain value or range (e.g., pH), beyond which “more is better” or “less is better” is applied.

A non-linear sigmoid function was applied to transform soil quality indicator values into a unitless soil quality indicator score [61,62] S, ranging from 0 to 1:

$$S=\frac{a}{\left(1+{\left(\frac{x}{x_0}\right)}^b\right)}$$

where a is the maximum score (=1.0); x is the soil quality indicator value; x_o is the mean value of the soil quality indicator, as observed in the collected dataset; and b is the slope which holds at −2.5 for “more is better” and +2.5 for “less is better” [61]. The functions of the soil indicators are shown in Table S1 (Supplementary Materials).

The scores were then integrated into a non-weighted additive SQI [61,63]:

$$SQI={\displaystyle \sum }_{i=1}^n{S}_i/n$$

where S_i is the indicator score and n is the number of soil quality indicators integrated into SQI. A preliminary study using our data and various transformation functions showed that this non-weighted additive SQI combined with a non-linear sigmoid transformation showed the strongest correlation to rice yield and was retained as the best-suited approach.

2.4. Statistical Analysis

Statistical analyses of the data were performed using RStudio software (version 2022.12.0). The splitplot function from the Doebioresearch package (Analysis of Design of Experiments for Biological Research) [64] was used to perform analysis of variance (ANOVA). In case of significant treatment effects, an LSD test was used to compare treatment means at a significance level of 5%. Pearson’s correlation coefficients were calculated between rice yield and soil properties, as well as rice yield and soil quality index (SQI).

3. Results

3.1. Yield Components

Yield components of rice (Figure 2) displayed some variation after the nine growing seasons, covering three rotational cycles, with panicle numbers of 515–603 panicles/m², total grain numbers of 54.7–63.4 grains/panicle, filled grain numbers of 48.2–55.7 grains/panicle, unfilled grain numbers of 6.0–7.7 grains/panicle, and filled grain percentages of 88.1–90.7%. However, neither the factors crop rotation nor organic amendment alone (p > 0.05) affected any of these yield components, except for the marginal (p = 0.096) effect of organic amendments on the filled grain number. Although insignificant, inclusion of an upland crop in the rotation tended to increase panicle number/m² (4.7–7.0%), total grain number/panicle (0.9–5.1%), and filled grain number/panicle (0.8–4.2%) as compared to the rice monoculture cropping system. Likewise, use of organic amendments tended to increase total grain number (4.1–6.7%) and filled grain number (4.1–8.3%), but only a marginal (p < 0.1) increase in filled grain number was observed with the addition of SGC over the no amendment treatment.

Combined inclusion of upland crops in the rotation and organic amendments did (p < 0.001) affect the 1000-grain weight of rice, even though differences between treatments were small. Moreover, use of organic amendments likewise did, across the cropping systems, significantly (p = 0.022) affect the 1000-grain weight (SGC 28.1 and RS+CM 27.9 g vs. 27.8 g of 1000 grains), while the introduction of upland crops in the rotations did not (Figure 2). The 1000-grain weight was highest under Se–R–R and NO-AM (28.3 g), followed by Se–R–R and SGC (28.1 g), while lowest under R–R–R and NO-AM (27.6 g).

3.2. Grain Yield and Above-Ground Biomass

Rice grain yield ranged from 6.55 to 7.30 t ha⁻¹, with the highest yield observed for Se–R–R combined with SGC and the lowest for R–R–R without any amendment (NO-AM) (

Table 3. Rice yield and above-ground biomass after nine growing seasons (three rotation cycles; 2019–2020 WS season).

Cropping System	Amendment	Grain Yield (t ha−1)	Above-Ground Biomass (t ha−1)
R–R–R	NO-AM	6.55 ± 0.11	10.40 ± 0.35
	RS+CM	7.25 ± 0.04	11.48 ± 0.18
	SGC	7.00 ± 0.26	11.51 ± 0.42
Se–R–R	NO-AM	6.86 ± 0.15	11.71 ± 0.82
	RS+CM	6.87 ± 0.93	11.33 ± 1.54
	SGC	7.30 ± 0.46	12.38 ± 1.15
So–R–R	NO-AM	6.77 ± 0.40	11.65 ± 0.11
	RS+CM	7.08 ± 0.14	11.50 ± 0.31
	SGC	7.11 ± 0.41	11.76 ± 0.46
Cropping system
R–R–R		6.93	11.13
Se–R–R		7.01	11.81
So–R–R		6.99	11.64
Amendment
NO-AM		6.73 b	11.25 b
RS+CM		7.07 a	11.44 ab
SGC		7.14 a	11.88 a
LSD0.05
Cropping system		ns	ns
Amendment		0.29 *	0.49 *
Cropping system × Amendment		ns	ns

* indicates significance at p < 0.05; ns: non-significant. Means followed by a common letter are not significantly different at 5% level (LSD test). The figures following ± signs are standard deviation (n = 3). R–R–R, Se–R–R, and So–R–R refer to rice–rice–rice, sesame–rice–rice, and soybean–rice–rice, respectively; NO-AM, RS+CM, and SGC refer to no amendment, rice straw + cow manure compost, and sugarcane compost, respectively.

). Use of organic amendments did (p < 0.05) raise grain yield by 5–6%, with 7.07 t ha⁻¹ for RS+CM and 7.14 t ha⁻¹ for SGC, which was significantly higher than 6.73 t ha⁻¹ in the case where no amendments were applied. Averaged across amendment treatments, the rotations tended to have higher rice yields than the R–R–R monoculture, with the average values in the order of Se–R–R (7.01 t ha⁻¹) and So–R–R (6.99 t ha⁻¹), compared to 6.93 t ha⁻¹ for R–R–R. Rotations of rice with sesame and soybean therefore increased average rice yield by 0.8–1.1% as compared to the R–R–R monoculture, respectively (Table 3), but these differences were not significant.

Above-ground biomass dry matter ranged from 10.4 (R–R–R without amendment) to 12.38 t ha⁻¹ (Se–R–R plus SGC) (Table 3). As for rice yield, crop rotation alone or in interaction with the use of organic amendments did not significantly (p > 0.05) affect the above-ground biomass, while organic amendments did. Application of SGC significantly (p < 0.05) increased the above-ground biomass by 5.6% compared to the NO-AM treatment (Table 3).

3.3. Yield vs. Soil Properties

Rice yield correlated (p < 0.05) negatively with the 0–15 cm soil bulk density and penetration resistance, and positively with porosity, air capacity, macroporosity, mesoporosity, and microporosity (Figure 3). Additionally, rice yield was found to marginally (p < 0.1) positively and negatively correlate with the plant available water capacity and EC, respectively. Regarding soil chemical properties, rice yield marginally (p < 0.1) and positively correlated with available P content and significantly (p < 0.05) and positively with Mn and Si contents (Figure 3). Similar relations were not found between rice yield and properties of the 15–30 cm depth layer.

3.4. Rice Yield vs. SQI

The relationship between rice yield after nine growing seasons and SQI (three rotation cycles) is presented in Figure 4. Rice yield was significantly (p < 0.01) and positively related with the SQI of the topsoil (0–15 cm) (r = 0.85). No such relation existed for the subsoil (15–30 cm).

4. Discussion

4.1. Soil Amendment and Crop Rotation Effects on Rice Growth

Rice yield may be determined by yield components including panicle number per m², grain number per panicle, filled grain number, filled grain percentage, and 1000-grain weight [55]. In the present study, the inclusion of a crop rotation and the use of organic amendments had a significant combined positive effect on the 1000-grain weight of rice after three rotation cycles. In addition, use of organic amendments alone also favored the filled grain number per panicle, though only marginally (p = 0.096). Inclusion of an upland crop in the rotations on its own had no significant effect on crop growth after these three rotation cycles. Our findings are thus not entirely in accordance with those by Linh [23], who reported that crop rotation after 10 cycles (30 cropping seasons) significantly increased panicle numbers per m² and grain number per panicle of rice. This contrast is probably due to the shorter duration of our experiment, as both yield components did tend to improve with the introduction of crop rotations. In addition, the nature of our field trial reduced the power to detect the effect of the main factor (i.e., cropping systems), as documented by Gomez and Gomez [38].

In this study, organic amendments significantly increased the 1000-grain weight of rice both in the So–R–R and R–R–R systems, while a reverse trend was observed under the Se–R–R system. The beneficial effects of organic amendments on the 1000-grain weight of rice were reported by previous studies [65,66,67]. This can be attributable to the higher N availability at the heading stage, as reported by De Datta [68], or the all-around improvement of soil N supply [67,69]. Furthermore, Litardo et al. [70] saw beneficial effects of organic amendments on other rice yield components, with the combined application of chemical fertilizer and compost increasing panicle length, grain number per panicle, and filled grain percentage as compared to the chemical fertilizer application alone. In line with the improved yield components, organic amendments also increased rice yield and above-ground dry biomass after three rotation cycles with upland crops (Table 3). Though rotations as such did not significantly affect rice yield and biomass, there was an increasing tendency in their increase when rotating rice with one upland crop. It is likely that a significant effect of rotation on rice yield may be expected with a long-term introduction. In a 10-year rotation experiment of rice with various upland crops, Chen et al. [22] reported that rice yield was significantly increased in all rotation systems, which was attributed to the improvement in soil organic matter and total N content by rotations. Likewise, it was confirmed by previous studies in the Mekong Delta that inclusion of upland crops helped significantly improve rice yield as of the 10th cropping season onwards [56]. Analysis of soil physical and chemical properties 10 years after the introduction of uplands in these studies revealed that they improved, especially in the subsoil (within 10–30 cm), also leading to an expansion of the rooting depth [23]. It must be noted that in these studies, upland crops were grown on temporarily raised beds. This practice could facilitate deeper soil mixing and enhanced root growth in the heavy-textured soil. The absence of such a land preparation practice in our study might have contributed to the delayed effect of crop rotation on soil properties and thus rice growth. More research would be needed to verify whether the trend of increased rice yield after nine cropping seasons observed in our study without raised beds substantiates on the longer term. This would be of particular interest given that the inclusion of temporary raised beds in the upland cropping seasons is more labor intensive [23]. Wang et al. [71] reported that the rotation of rice with upland crops increased rice yield and above-ground biomass by improving C and N stocks and the N availability of the subsoil (20–30 cm). To further interpret these effects, we investigate the overall effect of crop rotation and the use of organic amendments on soil quality, as well as its relation with rice productivity, in Section 4.2 below.

4.2. Soil Quality Index as a Predictor of Rice Yield

Sustainable management practices to improve rice production, such as the use of crop rotations and organic amendments, need to be assessed in terms of their overall effect on soil quality [72]. To provide an integrated assessment of such effects, a soil quality index (SQI) that integrates various soil properties used as indicators of soil quality has been forwarded as an effective means for evaluation [62]. In the current study, the rice yield after nine experimental cropping seasons (2019–2020 WS) was found to significantly and positively relate (r = 0.85) with our topsoil SQI (0–15 cm) (Figure 4). Liu et al. [73] reported a significant correlation between rice yield and SQI, which explained 69% of the variation in rice yield. Similar results have been reported in several studies, where other SQIs could predict 71–80% [74], 47–57% [75], and 50% [76] of the variation in rice yield across soils. In our study, the SQI (0–15 cm) explained 72% of the variation in rice yield (Figure 4), confirming SQI to be an effective tool to predict variation in rice production across different rice-based crop rotations with or without use of organic amendments. However, a SQI does not necessarily positively correlate with rice yield, as several previous studies found no [57] or even a negative [77] correlation between a derived SQI and rice yield. Lima et al. [57] suggested that such poor relations could be due to the fact that rice yield not only depends on soil quality but also on other factors such as seed quality and irrigation. Obviously, variation in such factors was ruled out in our field experiment. Another explanation for the variable prediction of rice yield by SQIs reported in literature is the large existent diversity in indexing methods applied [78].

In the current study, rice yield did not correlate to our subsoil SQI (15–30 cm). This suggests that the effects of the applied cropping practices on subsoil quality were either not pronounced enough to impact crop productivity or of lesser importance to rice growth than their effects on topsoil quality. Previous studies have reported a better correlation between rice yield and SQI in the topsoil (0–15 cm) than that averaged from top and subsoil (over 0–30 cm) (e.g., Lenka et al. [75]).

4.3. Rice Yield as Affected by the Cumulative Effects of Soil Properties

We suppose that the improvement in rice yield results from the enhancement in soil quality which can be attributed to improvement in the studied soil properties induced primarily by organic amendments and to a much lesser extent by crop rotations. These improvements are indicated by their positive and negative relationships between various soil properties with rice yield (Figure 3), as discussed further below.

Organic amendments have been widely reported to improve soil fertility and rice yield [27,67,69]. In the current study, it is accordingly plausible that the stimulatory effect of the organic amendments on rice yield and above-ground dry biomass resulted at least partly from the increased availability of plant nutrients. Such was indeed suggested by significant positive relations between rice yield and Mn (r = 0.79) and Si (r = 0.68), and P to a lesser extent as well (r = 0.63 but only at p = 0.067). Morevoer, there was a marginally significant negative relation with EC (r = −0.6), indicating that lowered salt stress could have also played a role (Figure 3). The marginal relation with EC may have been coincidental or indirect as it was within the optimal range for rice growth (<3 dS/cm) [79].

Mn is usually not considered a vital plant nutrient but it does play an important role in photosynthesis in rice plants [80]. The results suggested that increased Mn availability by organic amendments and rotations stimulated rice growth. This seems a particularly plausible explanation as Fairhurst et al. [80] identified 12 mg Mn kg⁻¹ as a critical threshold for Mn deficiency, which was surpassed by the use of RS+CM or SGC (Figure 3). The observed positive correlation between rice yield and exchangeable Si content (0–15 cm) (Figure 3) may likewise indicate that Si was limiting plant growth. Rice is well-known to be a Si-demanding crop [80,81], and Linh et al. [56] reported that in the Mekong Delta the soils’ available Si was below the deficiency threshold (40 mg kg⁻¹) for rice, as a consequence of the repeated removal of rice straw in rice monoculture systems. The use of OM amendments improved soil Si availability here, in line with Watanabe et al. [27], who reported an increase in available Si and rice yield under the application of rice straw compost (6 ton ha⁻¹) after 25 rice cropping seasons. There may also be a link between Si and P availability, as recently Schaller et al. [81] demonstrated that improved Si availability also results in enhanced P availability in paddy soils via the replacement of P with Si at the surface of amorphous and crystalline Fe complexes. An increase in P availability resulting from organic amendments has been often associated with increased P uptake in plants and grains [82] and therefore increased rice yields [67,69,70]. However, the limited strength of the relation observed here between yield and P suggests that other soil properties were more important. This seems a likely explanation as doubled or even tripled P availability after use of SGC did not seem to drastically improve rice yield.

Rice yield was found to correlate to several soil physical properties, aligning with several previous reports of soil physical quality control on rice yields [83,84]. More specifically, Nwite et al. [84] reported a strong correlation between rice yield and bulk density, aggregate stability, saturated hydraulic conductivity, and aggregate size fraction, and all these properties were affected by organic amendments. After the nine seasons in our study, several of these soil physical properties were altered by the treatments although they appeared predominantly changed by the use of organic amendments rather than by implementing crop rotations (Figure 3). Reynolds et al. [85] also found that the long-term application of composts helped to improve the physical quality of fine textured soils, as evidenced by the increased macro- and microporosity. Positive correlations of yield with topsoil macro-, meso-, and microporosity indicated that the overall enhanced porosity was beneficial for rice growth in our experiment. With soil pores playing an important role in providing living space to soil biota [86], and in storing and transporting air, water, and nutrients [87], it is logical that pore space architecture should also in turn affect crop yield. According to Kar and Ghildyal [88], rice root tips can penetrate macropores as small as 75 µm, corresponding with the macropore size of >75 µm considered in this study. Thus, enhanced macroporosity would potentially favor root growth, thereby improving nutrient and water uptake [89]. We also found a highly positive correlation (p < 0.01) between air capacity and rice yield (Figure 3). Although air capacity does not have much significance when rice fields are flooded and soils are saturated, farmers in the Vietnamese Mekong Delta often implement surface drainage 2–3 times during the cropping season, thereby making air capacity a relevant factor. The practice aims at enhancing root anchoring and stimulating root growth, and reducing risks of toxicity (e.g., Fe²⁺). Increased air capacity might thus favor the diffusion of oxygen into soil and thus potentially bring about several benefits, such as increased soil microbial activity and nutrient availability [90], reduced iron toxicity, and—finally—enhanced rice yield [91]. Indeed, several studies have evidenced the important role of soil aeration to rice growth, even though rice plants have the capacity to conduct oxygen from the atmosphere to their roots through their stem for respiration [55,92]. For instance, Zhu et al. [92] found that the provision of oxygen to soil via aerated irrigation resulted in a significant increase in effective panicles, seed setting rate, and grain yield, and this improved the root function of rice. On the other hand, rice yield decreased with increasing bulk density of the topsoil, and this decrease appeared linear between the lowest (0.91 g cm⁻³) and highest (1.12 g cm⁻³) observed bulk densities (Figure 3). Nwite et al. [84] also reported that rice yield decreased (r² = 0.763) as bulk density started to increase above 1.05 g cm⁻³, though quadratically in their case. Likewise, Linh et al. [93] observed—in the Mekong Delta—negative effects on rice yield from increased soil bulk density and decreased macroporosity of the puddled layer during later growth stages, which might be attributed to the settlement of soil particles during the cropping season [94]. It should be noted that in conventional rice cultivation, puddling, which results in increased bulk density and microporosity [59], is a common practice that facilitates crop establishment and growth [7].

In addition, there was a positive correlation between rice yield and penetration resistance (r = −0.77) (Figure 3)—in line with Mohanty et al. [77], who reported a similar relationship—and a remarkable decrease in rice yield when it reached 0.75 MPa. As topsoil penetration resistance was in all treatments above this threshold, it is not entirely clear if the relation to rice yield we found was direct or just via mutually correlating variables.

In sum, the SQI derived in the current study integrates 23 soil quality indicators and was more strongly related to rice yield than any of these individual indicators. Thus, as could be expected, the higher rice yield appeared to result from the cumulative effect of improved soil fertility overall, rather than from the enhancement of one or a few single soil quality indicators. In assessing the effectiveness of soil quality at improving management, it thus seems advisable to integrate the effects on individual soil properties into a wide SQI that encompasses both chemical as well as physical soil properties. While the observed improvement in SQI after nine growing seasons is encouraging, care should be taken before extrapolating these short-term gains to the longer term. For example, the efficacy of applying composts at a rate of 2 t ha⁻¹ (as investigated here) on offsetting the anticipated loss of SOC resulting from the inclusion of upland crops in the rice-based rotation, as reported by previous studies (e.g., Linh et al. [4], Witt et al. [28], and Cass et al. [29]), warrants careful consideration. Some effects of implementing adjusted farming management may only significantly affect soil quality and rice yield on the longer term. On the other hand, the use of organic amendments could be labor intensive. Therefore, labor resources should be considered in sites where organic amendments are applied. Additionally, organic amendments may pose environmental risks such as CH₄ emissions, although the emission potential varies with the type and quantity of organic amendments [17]. To minimize these risks, organic amendments should be used as compost with moderate rates when applied under similar conditions as in our study, in combination with other management strategies such as mid-season drainage [17].

5. Conclusions

After nine cropping seasons with three rotation cycles, rice yield was significantly increased (5–6%) by use of organic amendments, but not by implementing crop rotations, although an increasing tendency in yield (0.8–1.1%) was observed. Repeated organic amendments thus become an effective management practice to enhance rice yield already in the short term in rice-producing regions with similar soil types and climatic conditions, irrespective of the introduction of crop rotations with upland crops. Incorporating upland crops in rotations has the potential to further enhance yield (even without the use of temporary raised beds), but only on the longer term.

Rice yield was significantly and positively correlated with a derived SQI of the topsoil (0–15 cm), which predicted 72% of overall rice yield. Conspicuously, there was no relation between a SQI of the 15–30 cm layer and rice yield, although possibly the duration of our experiment may have been too short to obtain sufficient variation in the subsoil properties critical for rice growth. The relation between the topsoil SQI and yield was suggested to result from the increased availability of nutrients (Si and possibly also P), alleviated soil compaction, and improved aeration and porosity, primarily achieved via organic amendments. It is likely that not only these but also other soil properties as well cumulatively determined rice yield. Even though the SQI used in this study proved a good predictor of rice yield, some difficult-to-quantify yet relevant soil functions could still be included in the future to improve its predictive power. For instance, actual crop availability of N cannot be readily assessed from a simple soil test. Use of a SQI encompassing many soil properties indeed then precisely enables us to quantify the overall importance of soil fertility for rice yield versus other factors, and it provides an effective means of quantifying the integrated effect of new management. Moreover, including a wide range of properties (here 23) in a SQI should secure its applicability across diverse settings, including different crop rotations and various soil types.