Conservation Agriculture Improves Long-term Yield and Soil Quality in Irrigated Maize-oats Rotation

Abstract

Conservation agriculture, characterized by minimal tillage, permanent soil cover and crop diversification, has been widely adapted under rainfed conditions, but adoption under irrigated conditions is limited. An experiment initiated in 1997 to evaluate the level of tillage and the amount of soil cover needed for conservation agriculture to work under irrigated conditions in the highlands of San Luis Potosí, Mexico. The trial encompassed seven treatments: conventional tillage, two types of reduced tillage and permanent raised beds (PB), which are untilled, with four levels of soil cover in an irrigated rotation with maize (Zea mays L.) in the summer and fodder oats (Avena sativa L.) in the winter. Maize and oats yielded significantly more under PB than with tillage. Maize yielded on average 1.2 Mg ha⁻¹ more in PB with the hybrid used from 1997 to 2008 and 3.9 Mg ha⁻¹ more with the hybrid used from 2012 to 2017. Permanent raised beds generated on average $18,424 MXN ha⁻¹ more profit than the tilled treatments and increased soil organic carbon on average 63% at 0 to 5 cm depth and 32% at 5 to 30 cm depth. The trial shows that conservation agriculture, with PB and little residue cover, is a viable and sustainable option in similar production systems.

1. Introduction

Conservation agriculture is a crop management system based on minimal soil disturbance, keeping the soil covered with residues or crops, and growing diverse crops, to reduce production costs, minimize soil degradation and improve soil quality [1,2]. In Mexico, over 40% of the territory is affected by soil degradation, caused mainly by tillage and overgrazing [3]. Adoption of conservation agriculture could reduce soil degradation in Mexico but, given the tremendous diversity of agro-ecological conditions at different locations, the components of conservation agriculture have to be adapted for specific, individual production environments [4].

Conservation agriculture has been implemented mainly under rainfed conditions; adoption under irrigated conditions has been limited [5]. The benefits of conservation agriculture are more apparent under rainfed conditions where it helps capture and retain moisture, thereby increasing yield [6]. Notwithstanding, as under rainfed cropping, conservation agriculture under irrigated conditions can improve soil quality and save farmers money by reducing tillage costs [7,8]. Conservation agriculture under irrigated conditions is unlikely to improve yields, since water availability is generally not limiting in those conditions, but can lead to substantial water savings. For example, in the irrigated rice-wheat system of the Indo-Gangetic plains, conservation agriculture has been shown to reduce water use in flood irrigation compared to conventional tillage [9]. In Mexico, conservation agriculture has been evaluated under irrigation mostly for wheat-based production in the north, where it improved soil quality [7,8], but yield differences were small [10]. A lack of knowledge thus exists on the long-term effects of conservation agriculture under irrigated conditions, especially for maize.

The conditions in the Mexican highlands are representative of tropical highlands across the globe, which are also densely populated, intensively cropped, and prone to soil degradation and declining soil fertility [11]. Conservation agriculture could help reduce soil degradation and soil fertility problems in these regions, but research on how to implement conservation agriculture in the tropical highlands is limited, especially under irrigated conditions. In the highlands of the state of San Luis Potosí, Mexico, soils are typically prepared for sowing by plowing and disking. At the end of the cropping cycle, farmers remove all residues and cattle can graze freely on the leftovers in the fields, leaving the soil exposed to wind erosion during the dry season and to water erosion from increasingly intense rainfall events during the rainy season. Soils in the state are mostly xerosols and litosols, which are prone to rapid degradation if not managed well [12]. As a consequence of tillage and overgrazing, approximately 13.5% of San Luis Potosí state is now considered heavily eroded [13] contributing to the stagnation of maize yields since 1980 [14]. In the dry climate of the San Luis Potosi highlands, irrigation is necessary to produce maize, but water levels are declining in the aquifers used [15]. Maize irrigation in Mexico is generally done by flood irrigation, in which the entire field is covered with water, or by furrow irrigation, in which water is applied to furrows between raised beds. Both irrigation methods can cause erosion and soil degradation [16].

For irrigated maize-based cropping in the Mexican highlands, where tillage is commonplace and residues are in high demand, it is necessary to determine adequate levels of tillage and residue retention for the sustainability of conservation agriculture. This study evaluated the long-term effects of four soil preparation practices—including conventional tillage and various forms of reduced tillage, as well as use of permanent raised beds—and several levels of residue removal or retention on maize yields, system profitability and soil quality, in irrigated cropping in a tropical highland setting.

2. Materials and Methods

2.1. Study Area and Experiment

The experiment reported herein involved an irrigated, grain maize (Zea mays L.)-forage oats (Avena sativa L.) rotation and tested several soil tillage practices and residue retention levels. It was located at the “San Luis” Experiment Station of the Instituto Nacional de Investigaciones Forestales, Agricolas y Pecuarias (INIFAP) in Soledad de Graciano Sánchez, San Luis Potosí, Mexico (22°13′35.30″ N 100°50′56.67″ W, 1838 m above sea level).

At the start of the experiment in 1997, the soil was characterized as a xerosol with a sandy clay loam texture and 1.4% organic matter, pH of 8.1 and electric conductivity of 0.81 dS m⁻¹. The electric conductivity of irrigation water was 0.29 dS m⁻¹ and it had a sodium ratio of 1.26. The region has a dry temperate climate with an annual average temperature of 16.8 °C, a frost-free period from April to September, and an annual average rainfall of 332 mm (Figure 1). Monthly averages of precipitation and temperature from 1997 to 2018 were taken from the weather station closest to the trial (Supplementary Table S1).

The seven treatments evaluated consisted of four tillage methods: three conventional and one comprising four variants of permanent raised beds (PB) (

Table 1. Overview of treatments included in the trial in Soledad de Graciano Sánchez, San Luis Potosí, Mexico.

No.	Abbreviation	Tillage Practice	Residue Management
1	P+D	Conventional: plowing and disking	Remove
2	D	Minimal: disking only	Remove
3	M+D	Reduced: multiplow and disking	Remove
4	PB0%	Permanent raised beds	Remove
5	PB33%	Permanent raised beds	33% cover (1.3 Mg ha−1)
6	PB66%	Permanent raised beds	66% cover (2.6 Mg ha−1)
7	PB100%	Permanent raised beds	100% cover (4.0 Mg ha−1)

). The three conventional methods were (1) inversion tillage with a disc plow and disk harrowing to a depth of 30 cm (P+D), (2) disk harrowing to a depth of 30 cm (D), and (3) non-inversion tillage with a “multiplow”, a ripper equipped with sweeps to break up the soil at the operating depth (Supplementary Figures S1 and S2), followed by disk harrowing to a depth of 30 cm (M+D). In all three conventional tillage treatments, 1.65 m wide beds were shaped after disking; because all three treatments were disked, the soil was left in a similar condition, with clods of up to 5 cm on the surface. In treatments P+D, D, and M+D, soil preparation took place at the beginning of each maize and oats cycle. The PB were created in 1996 and were 1.65 m wide. Furrows in the PB have been reshaped every five years; the tops of the beds were not tilled during the study period. Furrows in all treatments were approximately 10 cm deep, with some variation in the periods between reshaping (every five years in PB) or remaking (every season in tilled treatments) of the furrows, when they gradually become shallower.

Four levels of soil cover with maize residues were tested on the PB: zero (PB0%), one-third (PB33%), two-thirds (PB66%) and full soil cover (PB100%). The amounts of residue to cover one-third, two-thirds and the entire soil surface correspond to 1.3, 2.6, and 4.0 Mg ha⁻¹, respectively. Residues were applied once a year after maize harvest. For oats, only a 5–10 cm tall stubble was retained. All maize residues were removed after harvest, chopped to 15–20 cm long pieces, and appropriate amounts were returned to the soil. No residue was returned to treatments P+D, D and M+D.

Plots were 8.25 m wide and 30.0 m long. Treatments were evaluated in a randomized complete block design with two replications.

Maize was planted in the spring and oats in the fall each year. In the summer cycles of 2009, 2010 and 2011, and in the winter cycles of 2007-2008, 2008–2009, 2009–2010, 2010-2011 and 2011–2012, the trial lay fallow with no activities, due to budget constraints. In the 2002-2003 oats cycle the crop was lost because of frost damage. Different maize hybrids were used in two periods of the trial, each at their recommended planting density. From 1997 to 2008, white maize hybrid H-311 (INIFAP, Mexico) was used, with a planting density of 50,000 plants ha⁻¹. From 2011 to 2017, the white maize hybrid XR-45 (Ceres, Mexico) was planted at 69,000 plants ha⁻¹. Maize row spacing was 0.825 m. The oat cultivar used was “Chihuahua” with a seeding rate of 80 kg ha⁻¹. Oats were planted in two double rows per bed. The center of the double rows was 0.825 m apart, with 0.20 m between each single row within a double row. Planting was done with a zero tillage planter. The fertilization dose for maize and oats was 200-100-00 (kg ha⁻¹ N-P-K) and 80-40-00 respectively, applying half of N and all of P at sowing time and the other half of N at approximately 35 days after planting.

Before sowing maize and oats, weeds in the PB treatments were controlled with glyphosate (1.4 L a.i. ha⁻¹), whereas in the tilled treatments, weeds were controlled by the tillage operations. In the maize crop, atrazine (0.75 L a.i. ha⁻¹) was applied in the first two weeks after planting in all treatments. Additionally, weeds were controlled mechanically in the tilled treatments approximately 35 days after planting, whereas in the PB treatments manual control was used if necessary. In 2017, nicosulfuron (60 g a.i. ha⁻¹) was applied to control grass weeds in the PB treatments. In the oats season, weed pressure was generally low and only mechanical or manual control was used when necessary in all treatments.

The crops were irrigated according to the conventional local practice, approximately every 18 days by furrow irrigation. Maize was irrigated every 14 days during grain filling. When it rained, irrigation was postponed according to the recorded rainfall. All treatments were irrigated simultaneously. In general, six irrigations of an estimated quantity of 1,000,000 L ha⁻¹ (100 mm) of water were applied per cropping cycle for both crops. Water was applied to the furrows but in treatments with low infiltration rates could cover the top of the beds as well.

Maize and oat forage yields were determined in sampling areas of 10.2 m². Maize was harvested at 14% grain moisture content and oats were harvested when the grain reached milk or dough stage. Oats moisture content was approximately 70% at harvest, yields were reported as dry biomass. Oats outside the sampling area were harvested with a cutter bar, leaving 5 to 10 cm of stubble on the field. Plant height in maize was measured from the ground to the base of the flag leaf.

2.2. Soil Quality

Samples for chemical soil analysis were taken on 24 October 2017 at depths of 0–5 and 5–30 cm. Five samples were taken per plot and a composite sample was made. Soil analyses were carried out by Fertilab, Celaya, Mexico. Phosphorous content was determined by the Bray and Kurt method as described in [17] and by the Olsen method as described in [18]. Interchangeable bases (Ca, Mg, Na and K) were determined with the ammonium acetate method, micronutrients (B, Cu, Mn, Fe, Zn) were determined with the DTPA-Sorbitol method at pH 7, and organic matter content was determined with the Walkley and Black method [19]. Sulphur was determined with the turbidimetric method [17] and nitrates were determined with the colorimetric method as described in [20].

Physical soil quality data and samples for mean weight diameter (MWD) analyses were resampled on 10 October 2018, in the sowing line. A composite sample of five samples per plot from 0–5 cm was analyzed for MWD by dry and wet sieving as described in [21,22]. Sand correction was applied. Bulk density was measured by sampling at 0–5 and 15–20 cm and determining volume and oven dry weight for three samples per plot. Penetration resistance was measured with a dynamic penetrometer at depths of 0–15, 15–30, 30 to 45 and 45–60 cm in 4 points per plot. Infiltration was measured as direct surface infiltration by the “time to pond” method as described in [23]. The time for water to infiltrate during an irrigation was measured for three irrigation events during the 2018 maize crop cycle and was defined as the time between opening the irrigation tube and the moment when no standing water was visible on the surface.

2.3. Economic Analysis

From 2015 onwards, production costs were documented. The common local costs to contract each operation were used to calculate production costs. Input prices were obtained from local suppliers. No rent or opportunity costs, no residue costs or income, no income from animals were taken into account, as the analysis only took into account the maize production system. All costs per hectare are expressed in Mexican Peso (MXN), the exchange rate of which fluctuated between 14.5 to 22 MXN per USD between January 2015 and December 2017. Grain prices used to calculate income were $4500, $5200 and $4000 MXN t⁻¹ in 2015, 2016 and 2017, respectively.

2.4. Data Set and Statistical Analysis

The complete dataset is available on Dataverse as “Long-term tillage and residue management experiment in San Luis Potosí, Mexico” at the following link: https://data.cimmyt.org/dataset.xhtml?persistentId=hdl:11529/10548248. Statistical analysis of the effect of the treatments on grain yield and soil quality parameters was performed in R 3.1.1 (R core team, Vienna, Austria) using the ‘glm’ and ‘aov’ functions from the ‘stats’ package. The growing cycle and the repetition were considered random factors. Post-hoc analysis was performed using the ‘HSD’ function from the ‘agricolae’ package which performs a Tukey test.

3. Results

3.1. Yield

From 1997 to 2008 the maize hybrid H311 was sown, and differences in grain yield between treatments were small, although in 1999 and 2002 yields in P+D and D treatments were significantly lower than in PB treatments (Figure 2). With the hybrid XR45 and a higher plant density from 2012 to 2017, differences between PB treatments and tillage treatments (P+D, D and M+D) were more pronounced, with all the PB treatments yielding significantly more. XR45 did not yield more compared to the 1997–2008 period under conventional tillage but yielded about 3.9 Mg ha⁻¹ more on PB than in conventional tillage. The low yields in 2001 in all treatments were due to low temperatures before the plants had reached physiological maturity. In 2004 and 2005, yields were similar in all treatments because of favorable rainfall distribution throughout the growing cycle.

Maize yields were similar with the different levels of soil cover in PB; the only residue effect observed was PB0% being not significantly different from P+D and M+D from 1997 to 2008 (

Table 2. Average maize grain yield in Mg ha⁻¹ per hybrid and treatment (1997–2008: H311, 2012–2017: XR45). Treatments indicated with same letter are not significantly different per hybrid and period.

Treatment	Maize Yield (Mg ha−1)
	1997–2008	2012–2017
P+D	6.4 ± 1.9 bc	6.7 ± 1.1 b
D	6.2 ± 1.7 c	6.3 ± 1.9 b
M+D	7.2 ± 1.2 abc	7.2 ± 0.6 b
PB0%	7.9 ± 1.9 ab	10.7 ± 0.7 a
PB33%	7.9 ± 1.8 abc	10.4 ± 0.5 a
PB66%	8.1 ± 1.5 ab	11.0 ± 0.6 a
PB100%	8.2 ± 1.8 a	10.4 ± 0.6 a
Treatment average	7.4 ± 1.8	8.9 ± 2.2

). The multiplow in M+D provided only a very slight yield advantage over the conventional plow (P+D); about 0.8 Mg ha⁻¹ with H311 and 0.5 Mg ha⁻¹ with XR45. The worst performing treatment over all years was the use of only disking for land preparation.

Forage oat dry matter yields were significantly higher in PB treatments over all years (Figure 3). The difference was the largest in the 2012–2013 crop cycle (4.2 Mg ha⁻¹), when planting was done in October instead of late November–early December (the practice in the other years). There were no significant differences in yields between tillage treatments or between PB treatments. Similar to maize yields, the differences between tillage and PB treatments were more pronounced after 2012; from 1997 to 2007 the average difference between tillage and PB treatments was 0.4 Mg ha⁻¹, whereas from 2012 to 2018 the average difference was 2.0 Mg ha⁻¹.

The conventional tillage treatments resulted in reduced plant growth and height. From 2014 to 2018, plant height was on average 1.64 m in conventional tillage treatments, whereas plant height was an average 2.28 m in all PB treatments (

Table 3. Average plant height and standard error in m of the maize plants per treatment for the 2014, 2015, 2016 and 2017 summer cycles.

Treatment	2014	2015	2016	2017
P+D	1.74 ± 0.01	1.78 ± 0.03	1.78 ± 0.03	1.81 ± 0.05
D	1.72 ± 0.03	1.88 ± 0.08	1.88 ± 0.08	0.75 ± 0.61
M+D	1.84 ± 0.09	1.83 ± 0.02	1.83 ± 0.02	1.80 ± 0.08
PB0%	2.41 ± 0.03	2.40 ± 0.00	2.40 ± 0.00	2.32 ± 0.03
PB33%	2.23 ± 0.02	2.45 ± 0.05	2.45 ± 0.05	2.32 ± 0.08
PB66%	2.41 ± 0.09	2.45 ± 0.05	2.45 ± 0.05	2.39 ± 0.09
PB100%	2.16 ± 0.03	2.45 ± 0.05	2.45 ± 0.05	2.29 ± 0.05

).

3.2. Production Costs

Maize production costs were lower for PB treatments because of the elimination of tillage. The difference was small however, only $1,211 MXN ha⁻¹ on average. Soil preparation accounted for about 11–13% of the total cost in the conventional tillage treatments, while it only accounted for 1% of the total production costs in PB treatments (

Table 4. Average proportion (%) of total maize production cost per category, averaged over three summer cycles (2015, 2016 and 2017).

Cost Category	P+D	D	M+D	PB0%	PB33%	PB66%	PB100%
Land preparation	13	11	11	1	1	1	1
Sowing	11	12	12	12	12	12	12
Fertilization	22	22	22	23	23	23	23
Weed and insect control	11	11	11	17	17	17	17
Irrigation	29	30	29	31	31	31	31
Harvest	14	14	14	17	17	17	17

). In all treatments, fertilization was the major cost, accounting on average for 22–24% of production costs, while irrigation accounted for 29–31% of the cost. The difference between the highest production cost and the lowest was on average $1,533 MXN ha⁻¹_, ranging between $20,189 MXN ha⁻¹ for PB treatments in 2015 and $26,670 MXN ha⁻¹ for P+D in 2016 (

Table 5. Production costs, total income and profits in $MXN ha⁻¹ per treatment for the summer cycles of 2015, 2016 and 2017.

Cycle	Category	P+D	D	M+D	PB0%	PB33%	PB66%	PB100%
2015	Soil preparation	$2880	$2480	$2480	$240	$240	$240	$240
	Sowing	$2600	$2600	$2600	$2600	$2600	$2600	$2600
	Fertilization	$4089	$4089	$4089	$4089	$4089	$4089	$4089
	Weed, disease and pest management	$2020	$2020	$2020	$3060	$3060	$3060	$3060
	Irrigation	$6200	$6200	$6200	$6200	$6200	$6200	$6200
	Harvest	$4000	$4000	$4000	$4000	$4000	$4000	$4000
	Total production cost	$21,789	$21,389	$21,389	$20,189	$20,189	$20,189	$20,189
	Total income	$29,392	$34,087	$33,073	$49,370	$49,488	$52,519	$50,014
	Net income	$7603	$12,698	$11,685	$29,181	$29,299	$32,331	$29,825
2016	Soil preparation	$3600	$3150	$3150	$450	$450	$450	$450
	Sowing	$2750	$2750	$2750	$2750	$2750	$2750	$2750
	Fertilization	$6800	$6800	$6800	$6800	$6800	$6800	$6800
	Weed, disease and pest management	$2020	$2020	$2020	$2870	$2870	$2870	$2870
	Irrigation	$9100	$9100	$9100	$9100	$9100	$9100	$9100
	Harvest	$2400	$2400	$2400	$3600	$3600	$3600	$3600
	Total production cost	$26,670	$26,220	$26,220	$25,570	$25,570	$25,570	$25,570
	Total income	$36,400	$38,480	$36,920	$55,120	$53,560	$56,680	$53,560
	Net income	$9730	$12,260	$10,700	$29,550	$27,990	$31,110	$27,990
2017	Soil preparation	$3450	$2550	$3000	$0	$0	$0	$0
	Sowing	$3430	$3430	$3430	$3430	$3430	$3430	$3430
	Fertilization	$5710	$5710	$5710	$5710	$5710	$5710	$5710
	Weed, disease and pest management	$4580	$4580	$4580	$6080	$6080	$6080	$6080
	Irrigation	$6900	$6900	$6900	$6900	$6900	$6900	$6900
	Harvest	$4400	$4400	$4400	$4400	$4400	$4400	$4400
	Total production cost	$28,470	$27,570	$28,020	$26,520	$26,520	$26,520	$26,520
	Total income	$22,567	$12,060	$28,978	$31,434	$35,256	$36,346	$34,854
	Net income	−$5903	−$15,510	$958	$4914	$8736	$9826	$8334
Average	Total production cost	$25,643	$25,060	$25,210	$24,093	$24,093	$24,093	$24,093
	Total income	$29,453	$28,209	$32,991	$45,308	$46,101	$48,515	$46,143
	Net income	$3810	$3149	$7781	$21,215	$22,008	$24,422	$22,050

). In contrast, the difference between the highest and lowest total income was on average $22,564 MXN ha⁻¹, ranging from -$12,060 MXN ha⁻¹ for D in 2017 to $56,680 MXN ha⁻¹ for PB66% in 2016. The major impact of conservation agriculture on net income was therefore due to the higher yields, which generated higher total income for these treatments, rather than to the cost reduction due to reduced tillage (Table 5).

3.3. Soil Quality

Organic carbon content and P Bray were significantly higher in the PB treatments in the 0–5 cm layer (Figure 4). Contents of highly soluble nutrients such as nitrates, Na, K, and S tended to be higher in conventional tillage in the 0–5 cm layer than under PB, leading to higher electric conductivity in this layer under conventional tillage (Supplementary Table S2). On the other hand, Bray-P, organic matter, Fe and Mn were higher in the 0–5 cm and 5–30 cm layers under PB. No clear tendency was observed for P-Olsen, Cu, Zn, Ca, Mg or B. Significant differences between tillage and PB treatments were found in organic matter and Bray phosphorous content in the 0 to 5 cm layer and potassium and sodium in the 5 to 30 cm layer, whereas the differences in other chemical parameters were not significant.

No significant differences were observed in soil bulk density or penetration resistance at any depth, indicating that there was no soil compaction under PB and without tillage for more than 20 years. Direct infiltration was not significantly different between treatments, mainly due to the moment of measuring. The soil was dry at the moment of measuring; water infiltrated readily into the soil pores in the PB treatments but mainly through cracks under conventional tillage. A different moment or method of measuring infiltration might have indicated large differences; for example, irrigation water infiltrates rapidly (3 h 5 min on average) in the PB, whereas in conventional tillage treatments infiltration takes 4 full days due to surface sealing. However, differences in MWD under wet or dry sieving, which could be indicative of differences in soil structural integrity, were not statistically significant, even though the data indicate that aggregate distribution and stability is larger in the PB treatments (Figure 5). No differences in soil quality were observed between the treatments with different levels of soil cover.

4. Discussion

The results demonstrate the sustainability of conservation agriculture practices. Yields were higher on PB than with tillage, especially in the later years of the trial, showing a positive long-term effect from conservation agriculture in the San Luis Potosi Highlands and, presumably, for similar regions. Typically, under irrigation, where water is non-limiting, yields for conservation agriculture and conventional tillage are similar [24]. However, there are relatively few studies on this topic; in the meta-analysis of Pittelkow et al. [24], only five out of the 610 studies included involved irrigated trials lasting more than 10 years. More long-term studies on conservation agriculture are needed, because positive effects can take years to manifest themselves and will depend on local soils. For example, in 10 years of trials in the Yaqui Valley in northwestern Mexico, wheat yields under conventional tillage and conservation agriculture did not differ [10], whereas in an irrigated conservation agriculture trial in the state of Bihar, India, wheat yields were higher after two years and rice yields after six years of conservation agriculture [25].

In the trial reported here, two effects might explain the higher yields for maize under conservation agriculture during 2012-2017. First, the irrigation method. When water is applied to the furrows in conventionally tilled plots, the degraded soil structure causes the soil to collapse and it takes several days for the water to infiltrate completely. Due to the amount of water applied, the water can also temporarily cover the top of the beds. Conversely, on conservation agriculture plots the same quantity of water infiltrates in a couple of hours. Five-to-six irrigations are needed per cycle, so spread over the different irrigation events, plant roots in tilled plots may spend up to a month per growing cycle in water-saturated conditions, unable to respire which reduces the effective growing cycle [26]. Using drip or sprinkler irrigation could probably reduce the problem of water-logging in tilled treatments, but these irrigation methods would not be profitable for grain production under current conditions. Conservation agriculture on PB can substantially reduce the water needed for maize production under flood irrigation [27]. Adapting irrigation to the treatment instead of applying the same irrigation as done in our experiment could demonstrate the water-saving and reduced water-logging benefits of conservation agriculture, which would be highly relevant for the study region [16]. Second, cultivar × management interaction contributes significantly to the differences in results between treatments, something that became apparent when the maize hybrid was changed to the higher-yielding XR45. On the one hand, XR45 needs the optimal growing conditions created under conservation agriculture to reach its full yield potential; on the other hand, a lower-yielding genotype is unable to take advantage of the improved conditions under conservation agriculture and there were no yield benefits with H311. This illustrates the need both to breed high-yielding varieties and to provide good growing conditions for those varieties to express their yield potential.

Soil cover is one of the three components of conservation agriculture and is necessary to maintain soil quality and associated crop yield benefits [28]. Retention of crop residues is a common way to ensure a suitable cover. In the present study, different levels of residue cover, ranging from zero to full soil cover with maize residues, did not affect yields or soil quality, most likely because the soil is almost permanently covered by a crop-maize in the spring-summer cycle and oats in the autumn-winter cycle. Crop residues are highly valued in Mexican maize cropping systems for fodder and other uses, especially in many of Mexico’s intensive cropping systems, such as the irrigated maize-wheat systems of the Bajio region [29], holding back the adoption of conservation agriculture [30]. In the conservation agriculture cropping systems evaluated in this study, farmers could remove crop residues after harvest and use them for animal fodder without compromising long-term sustainability.

No significant differences in MWD, soil penetration resistance or direct infiltration were detected, though previous studies found differences in infiltration rate and bulk density [26]. This may be related to the moment of sampling or due to the place of sampling, as the first samples in the current study were taken in the middle of the beds and the samples in the Martínez Gamiño and Jasso Chaverría study were taken in the crop row. The measurement method could also have an effect. During sample preparation, aggregates larger than 8 mm were crumbled, and it is possible that the difference in soil structure, as evidenced by the marked differences in infiltration time of irrigation water, are at a larger scale. Furthermore, the trial consisted of only two repetitions and this, combined with the relatively large variation, makes it difficult to obtain statistical significance. Other studies of irrigated, conservation agriculture systems have documented effects on soil quality [8,30], but soil type, cropping system and climate were different. No evidence was found of higher compaction in the PB treatments, which are untilled, compared to P+D. Soil compaction is sometimes considered a negative effect of reduced tillage, but in this trial, that was not the case. Even after 22 years of PB, there is no need for tillage due to compaction.

Organic carbon contents were higher than previously reported in [26] in the 0–5 cm layer, but the overall increase since the last measurements in 2001 was minimal, indicating the soil may have reached an equilibrium organic matter content in the conservation agriculture treatments in the 0-5 cm layer. The data indicate a higher carbon content in the 5-30 cm layer in conservation agriculture than under conventional tillage; future studies are needed to determine whether organic matter can keep increasing in this layer. Organic matter content increased in all PB treatments, even where all residues were removed. Similarly, in another trial in a similar production system in Mexico, no significant differences in soil organic matter content were observed after five years of conservation agriculture, even though residue levels were significantly different between treatments [31]. In contrast, in a five-year trial in Northwestern India, soil organic carbon did increase by 12 to 19% in PB with residue retained, compared with residue removal [32]. In our study, organic matter content increased in all treatments from 1.4% at the start of the trial to over 2%, even with residue removal, which indicates an important role for the roots in increasing organic matter content. About 21% of assimilated carbon is allocated underground in cereals [33], which implies that a significant amount of organic matter is produced as roots in intensive systems such as the one described in this study.

The PB treatments, which were untilled, gave higher yields than the conventional tillage treatment of plowing and disking and they had slightly lower production costs. Although in intensive, irrigated cropping systems, the difference in tillage costs is small relative to total production costs, the combination of lower production costs and higher grain yield makes maize production on PB an economically attractive option for farmers.

Soil quality was not different between the treatments with tillage (P+D, D, M+D) and yield differences were small. Farmers who want to improve their profitability but are not able to change their system to PB could therefore reduce tillage with systems like the D and M+D treatments and thus slightly reduce production costs without reducing yields or soil quality compared to P+D. Since the M+D treatment did not differ much from conventional plowing in yield or soil quality, this treatment was changed to PB33% in the summer of 2018. This will allow researchers to determine the time it takes for the degraded soil to recover and for maize yields to reach the same levels as the treatments that have been under PB since 1996.

5. Conclusions

Conservation agriculture with permanent raised beds (PB) led to improvements in soil quality that sustained higher maize and oat yields than treatments with tillage. Maize yielded on average 1.2 Mg ha⁻¹ more in PB with the hybrid used from 1997 to 2008 and 3.9 Mg ha⁻¹ more with the hybrid used from 2012 to 2017. Two factors potentially contributed to the larger yield difference for maize in the second period. First, due to soil degradation in the tilled treatments, the irrigation water took several days to infiltrate, reducing the effective growing cycle of the crop, whereas with PB the same quantity of water infiltrated in a couple of hours. Second, the cultivar used in the second period had a higher yield potential and thus was able to take advantage of the improved conditions on PB. Permanent raised beds generated on average $18,424 MXN ha⁻¹ more profit than the tilled treatments, mainly due to the increased yield in those treatments, and increased soil organic carbon on average 63% at 0 to 5 cm depth and 32% at 5 to 30 cm depth. The level of residue cover did not cause significant differences between PB treatments, probably because the soil was almost continuously covered by either the maize or oats crop. The trial showed that conservation agriculture, with PB and residue cover ranging from little to full residue cover, is a viable and sustainable option in similar production systems.