Introduction
Southern Africa is heavily affected by declining soil fertility concurrent with negative impacts of climate change, posing severe risks to the sustainability of farming and food security (Tittonell and Giller, Reference Tittonell and Giller2013). Current yields of maize (Zea mays L.), the most important food crop of the region cultivated on 50–80% of the land area, average only 0.5–2.5 t ha−1, whereas the yield potential for maize is 10–15 t ha−1 under optimal rainfed and unlimited nutrient conditions (FAOSTAT, 2020). Changes in climate will manifest in more erratic rains and increased heat stress that will be further detrimental to maize yields (Cairns et al., Reference Cairns, Hellin, Sonder, Araus, MacRobert, Thierfelder and Prasanna2013). Therefore, efforts are required to identify maize crop management systems that will help ensure long-term sustainable production under these conditions.
Sustainable intensification of crop production under climate change can be achieved, for example, through the build-up of soil organic matter (SOM). SOM supports biological abundance and diversity, represents a store of plant nutrients and confers physical (e.g., soil structure, stability and water holding capacity), biological (e.g., earthworm populations and greater diversity of microbial communities and functions) and chemical (e.g., ion exchange capacity) benefits (Schnitzer and Khan, Reference Schnitzer and Khan1975), which define soil health or the continued capacity of soil to function (Lehman et al., Reference Lehman, Cambardella, Stott, Acosta-Martinez, Manter, Buyer, Maul, Smith, Collins and Halvorson2015). However, as a consequence of heterogeneity in SOM distribution, changes in organic matter stocks can take several years to be measurable (Chivenge et al., Reference Chivenge, Murwira, Giller, Mapfumo and Six2007) and under no-tillage (NT) conditions can also lead to stratification with soil depth (Luo et al., Reference Luo, Wang and Sun2010). Timescales for measurable changes in SOM content may also vary with soil type (as a function of change relative to the size of the existing SOM stock) (Chivenge et al., Reference Chivenge, Murwira, Giller, Mapfumo and Six2007), so informed selection and management of improved cropping systems requires early and quantifiable indicators of trajectories toward sustainability for beneficial impacts on soil health and nutrient provision. Despite the efforts to investigate soil biological properties in southern Africa (e.g., Bationo et al., Reference Bationo, Waswa, Kihara and Kimetu2007), the links between trajectory indicators (e.g., properties linked with SOM abundance) and maize productivity under different management practices, and how they vary between soil types, are largely unknown.
Poor soil health in southern Africa is often exacerbated by the prevalent soil types, including granitic sandy or ferralitic soils, which are characterized by low pH, lack of available phosphorous (P) and nitrogen (N), and low levels of soil organic carbon (C) (Nyamapfene, Reference Nyamapfene1991). SOM accrual promotes soil health through increasing structural stability, reducing nutrient losses associated with erosion, improved nutrient and water capture and reduced leaching losses (Lal, Reference Lal2016). Decline in soil water and nutrient holding capacity is closely related to reduced SOM-C stocks. This is exacerbated under sandy soils where SOM is the primary mechanism for water retention (Hudson, Reference Hudson1994), and under a tropical climate where there is limited biomass return to the soil, for example, due to crop residue removal. Assessments of soil quality in this region, such as in Zimbabwe, have typically focused on soil chemical and physical properties [e.g., soil texture, structure, bulk density, pH, electrical conductivity (EC), cation exchange capacity (CEC), SOM, and C, N and P concentrations] (Tittonell et al., Reference Tittonell, Zingore, van Wijk, Corbeels and Giller2007; Zingore et al., Reference Zingore, Murwira, Delve and Giller2007). The biological characteristics of these soils, particularly the microbial biomass that is a key regulator of rates and magnitude of SOM build-up and turnover, are often overlooked and rarely reported. This is despite the microbial-driven turnover of SOM being a key determinant of the availability of some major soil nutrients for crop growth.
Since the 1990s, crop management options designed to improve soil fertility and build SOM stocks have been widely promoted. An example is NT with crop residue retention, as utilized in different forms of conservation agriculture (Thierfelder et al., Reference Thierfelder, Baudron, Setimela, Nyagumbo, Mupangwa, Mhlanga, Lee and Gérard2018) that are currently being practiced on more than 180 M ha of arable land worldwide, and on over 1.5 M ha in Africa (Kassam et al., Reference Kassam, Friedrich and Derpsch2019). There have been attempts to relate the soil health impacts of NT with residue retention on crop yield (Thierfelder and Wall, Reference Thierfelder and Wall2009; Nyamangara et al., Reference Nyamangara, Nyengerai, Masvaya, Tirivavi, Mashingaidze, Mupangwa, Dimes, Hove and Twomlow2014), but the success of this varies with soil type and environmental constraints (Thierfelder and Wall, Reference Thierfelder and Wall2012). Direct measurement of statistically significant changes in SOM stocks is challenging in the short-term (<15 years), due to stock size and spatial heterogeneity (O'Dell et al., Reference O'Dell, Sauer, Hicks, Thierfelder, Lambert, Logan and Eash2015). As timescales for measurable change vary with soil type, as well as agro-ecology and on-site management (Chivenge et al., Reference Chivenge, Murwira, Giller, Mapfumo and Six2007), this may in part explain contradictory findings in southern Africa, with reports of both evidenced increases (Thierfelder and Wall, Reference Thierfelder and Wall2012) and no increases in SOM-C (Cheesman et al., Reference Cheesman, Thierfelder, Eash, Kassie and Frossard2016) in response to NT with residue retention, highlighting the need for additional indices to track trajectories of management impacts on soil health.
Here, we examined the relationships between key soil health attributing properties—chemical, physical and biological—and maize productivity under conventional tillage with crop residue removal and NT with crop residue retention at long-term on-station and on-farm sites of contrasting soil types in Zimbabwe. Our aim was to identify the soil characteristics, or combinations of characteristics, most closely related to maize grain yield and above-ground biomass (i.e., stover) that could serve as early indicators of the sustainability of the management practices.
We hypothesized that (i) relationships between key soil properties and maize productivity (grain yield and above-ground biomass) vary with soil type, for example, a stronger reliance on soil mineral N concentrations in sandy soils, and (ii) management impacts (conventional management with residue removal vs NT with residue retention) on these relationships also vary among locations with contrasting soil types, with the greatest impact of management in the sandy soils due to their low inherent nutrient status and low buffering capacities.
Materials and methods
Site description
Sites were selected from existing long-term on-station and on-farm trials in Zimbabwe to include contrasting soil types (sandy and clayey soils). Soil sampling was conducted in June and October 2017 at the on-station and on-farm sites, respectively, from paired plots under conventional tillage with crop residue removal and NT with crop residue retention. Two on-station trials with contrasting soil types were selected at the Domboshawa Research Centre (Domboshawa) (−17.603 S; 31.604 E; 1545 m.a.s.l.) and the University of Zimbabwe Farm (Harare) (−17.722 S; 31.021 E; 1494 m.a.s.l.). Additionally, on-farm trials were selected at Hereford Farm (Hereford) (−17.423 S; 31.445 E; 1106 m.a.s.l.), Chavakadzi (Shamva) (−17.189 S; 31.493 E; 1164 m.a.s.l.) and Madziwa (−16.991 S; 31.415 E; 1177 m.a.s.l.). All sites were characterized by moderate, unimodal rainfalls of 600–900 mm which usually fell between November and April (Table 1). The on-farm sites had been under different management practices since 2004/2005 (Shamva) and 2005/2006 (Madziwa and Hereford) cropping seasons, whereas the on-station trials (Harare and Domboshawa) were under treatment since the 2009/2010 cropping season.
Table 1. General characteristics of the experimental trial sites
Farmers in these areas mostly grow maize which is sometimes rotated or intercropped with groundnuts (Arachis hypogaea L.) and cowpeas (Vigna unguiculata Crantz), but not as part of a systematic rotation (Waddington et al., Reference Waddington, Karigwindi and Chifamba2007). Rotations on more fertile soils may also include soybean [Glycine max (L.) Merr] (Table 1). Conventional land management at the sampling sites is based on soil tillage with a single row mouldboard plough at shallow depth (0–15 cm), crop residue removal or burning and monocropping or full rotations with legumes.
Experimental design and crop management
The on-station trials had three replicates and two main treatments (5 m × 4.6 m plot size): (a) conventional management with soil tillage and crop residue removal (CT); and (b) no-tillage and crop residue retention (NT). Sub-treatments for these trials, but not used in this study, were eight maize varieties sown in two plant populations. Seven drought-tolerant and one non-drought-tolerant maize varieties were seeded in these trials. The maize was fertilized with 83 kg N ha−1, 28 kg P2O5 ha−1 and 14 kg K2O ha−1 supplied as basal dressing at seeding, and 69 kg N ha−1 as a split-applied topdressing 4 and 7 weeks after crop emergence (WACE).
The on-farm locations had replicates across farms (six farmer replicates in each community) with three main treatments (each 1000 m2 in size; total 3000 m2 at each farmer's field): (a) conventionally ploughed maize-legume rotation with residue removal (CT); (b) NT-ripline seeding with a maize-legume rotation and crop residue retention (NT with ripping); (c) NT-animal traction direct seeding with a maize-legume rotation and crop residue retention (NT with direct seeding). The main plots were split into four subplots and three drought-tolerant and one non-drought-tolerant maize varieties were seeded in these at all on-farm sites. Fields were fertilized with 11 kg N ha−1, 21��kg P2O5 ha−1 and 11 kg K2O ha−1 as basal dressing at seeding, and 69 kg N ha−1 as topdressing split-applied at 4 and 7 WACE, reflecting low N inputs in smallholder plots in Zimbabwe and southern Africa. In the legume phase of the rotation, only basal dressing was applied at the same rate as maize. The choice of legumes planted at these on-farm sites (Table 1) depended on farmer preferences as well as their suitability for the local soils. However, for the purpose of this study, only maize yields were reported in the assessments. At all sites, weed control was achieved with an initial spray of a non-selective herbicide, glyphosate [N-(phosphono-methyl) glycine] at a rate of 2.5 l ha−1 (1.025 l ha−1 active ingredient) followed by manual hoe weeding (1–3 times per season according to weed pressure).
Soil sampling procedures
Soils were sampled from on-station and on-farm trial plots during off-season in June and October 2017, respectively. Ten sub-samples (0–10 cm soil depth) from each plot were thoroughly mixed into a composite sample and sieved through a 2 mm mesh on-site. These soil samples were then packed in cooler boxes and transported to Aberdeen, United Kingdom, where they were stored at 4°C until analyses for soil physico-chemical properties and microbial biomass C (MBC).
Characterization of soil physico-chemical properties and MBC
Soil texture, total nitrogen and carbon
Soil texture determination for on-station sites followed a particle-size soil fractionation procedure of Garcia-Pausas et al. (Reference Garcia-Pausas, Casals, Rovira, Vallecillo, Sebastià and Romanyà2012). In this procedure, the soil was separated into three size fractions of coarse sand (2000–250 μm), fine sand (250–53 μm) and silt plus clay (<53 μm) by wet sieving. Dry mass of each soil particle-size fraction was determined following oven drying to constant weight at 65°C. Physical particle-size soil fractionation was not performed for on-farm trial sites as recent soil texture data for these sites were obtained from previous work (CIMMYT, unpublished). Whole soil samples (i.e., non-fractionated soil) for on-station and on-farm sites were ball milled and analyzed for total C and N concentrations on a Flash EA 1112 Series Elemental Analyzer (Thermo Finnigan, Bremen, Germany). In addition, soil bulk density data for on-station sites were obtained from previous work (CIMMYT, unpublished) and used to estimate soil C and N stocks. Soil C and N stocks were not determined for on-farm sites in this study as soil bulk density data for these sites were not available.
Electrical conductivity, pH and cation exchange capacity
Soil pH and EC were measured using an electrode in a soil/water suspension (15 g soil and 45 ml H2O) (Thomas, Reference Thomas and Sparks1996). As soil pH values were <7, CEC was estimated as the sum of exchangeable cations and exchangeable acidity (Ross and Ketterings, Reference Ross and Ketterings1995). Exchangeable cations and exchangeable acidity were determined according to the methods described by Thomas et al. (Reference Thomas, Page, Miller and Keeney1982) and Parker (Reference Parker1929), respectively.
Ammonium, nitrate and MBC
Soil mineral N (NH4+ and NO3−) concentrations were determined using an autoanalyzer (Technicon Traaks 800, Saskatoon, Canada) following extraction of 10 g fresh soil with 50 ml of 1 M KCl solution. To determine soil MBC, chloroform fumigation-extraction was used according to Vance et al. (Reference Vance, Brookes and Jenkinson1987), where fresh fumigated and non-fumigated soil samples (equivalent 12.5 g dry soil) were extracted with 50 ml of 0.5 M K2SO4 solution. Organic C of the extracts was analyzed on a TOC Analyzer 700 (Corporation College Station, Texas, USA). MBC was calculated as the difference between organic C in the paired fumigated and non-fumigated extracts using a conversion factor k EC of 0.45 (Vance et al., Reference Vance, Brookes and Jenkinson1987).
Crop productivity
Maize was harvested at physiological maturity. Fresh cob and biomass samples were taken from 10% of the on-farm treatment areas and 60% of the on-station treatment areas and weighed in the field. A representative sub-sample was then collected, measured fresh, dried for 4 weeks, shelled and the grain and biomass sub-samples reweighed, and the grain moisture determined. The final grain yield was then averaged for all varieties and plant populations for 2018 and 2019 seasons, resulting in a single average yield value for statistical analysis, expressed as kg ha−1 at 12.5% moisture content.
Statistical analyses
The software package GenStat (Eighteenth Edition, VSN International Ltd) was used to test for data normality and to conduct two-way analysis of variance (ANOVA) to assess, using models, the effects of management practices (called management where necessary), trial location, and their interaction on soil physico-chemical properties, MBC, and on maize grain yield and aboveground biomass as measures of crop performance (Equation 1). In these models, replicates were regarded as random effects.
where Yijk is response variable in the jth site and kth management, μ is the overall response variable average, Site × Management is the interaction effect of Sitej and Managementk and ξijk is the error associated with the jth site and the kth management and is normally and independently distributed. In addition, one-way ANOVA was used to test for differences in soil properties and maize crop performance between management practices. For these analyses, on-station sites (n = 3) were analyzed separately from on-farm sites (n = 6) due to the different trial designs. Where significant (P < 0.05) treatment effects were found, least significant difference tests were used to separate the individual means.
In addition, principal component analysis (PCA) was used to assess the relationships between maize productivity and soil characteristics under the different management practices and trial locations or soil types. Since the different variables were recorded at different scales, all data were first standardized by scaling using the PCA() function of the ‘FactoMineR’ package in R. The eigenvalues (variances retained) for each of the principal components were calculated using the get_eigenvalue() function of the ‘factoextra’ package in R. A correlation analysis was conducted using the ‘corrplot’ package (Wei and Simko, Reference Wei and Simko2017) in R and using the Pearson product moment correlation test, to measure the strength of the relationships.
Results
Soil texture differed between sites (Table 2), so we divided the sites into two broad textural groups based on their silt + clay content: ‘sandy’ (for Domboshawa and Madziwa; <20% silt + clay content) and ‘clayey’ (for Harare, Shamva and Hereford; >30% silt + clay content).
Table 2. Texture (0–10 cm depth), pH (H2O), electrical conductivity (EC) and cation exchange capacity (CEC) of soils at Domboshawa and Harare on-station sites and Madziva, Shamva and Hereford on-farm sites under conventional tillage (CT) with crop residue removal and no-tillage (NT) practices (ripping, direct seeding) with residue retention
Notes: Values are means ± one standard error of the mean. Lowercase letters indicate significant differences (P < 0.05) between treatments within columns for each site.
a Soil texture of on-farm sites was determined in previous work (CIMMYT, unpublished).
b Average value of NT systems (i.e., ripping and direct seeding).
Maize crop performance
Maize grain yield and above-ground biomass averaged across management practices were significantly lower (P < 0.05, Table S1) at the sandy soil sites than at the clayey sites (Figs. 1 and 2). Average grain yield and above-ground biomass across treatments at Domboshawa were 2.3 and 2.4 t ha−1, and 2.3 and 2.8 t ha−1 at Madziwa, respectively. The highest (P < 0.001) average grain yield and above-ground biomass of 5.9 and 6.1 t ha−1, respectively, were obtained at the clay soil Shamva site, whereas yields from the Harare and Hereford sites were intermediate (Figs. 1 and 2).
Fig. 1. Boxplots of maize grain yield and above ground biomass in sandy soil sites of Zimbabwe under conventional tillage with crop residue removal and different no-tillage practices (ripping, direct seeding) with residue retention across on-station and on-farm trials. Lowercase letters indicate significant (P < 0.05) differences between management practices at each site. Values are means ± one standard error of the mean (2018 and 2019 seasons).
Fig. 2. Boxplots of maize grain yield and above ground biomass in clay soil sites of Zimbabwe under conventional tillage with crop residue removal and different no-tillage practices (ripping, direct seeding) with residue retention across on-station and on-farm trials. Lowercase letters indicate significant (P < 0.05) differences between management practices at each site. Values are means ± one standard error of the mean (2018 and 2019 seasons).
Management affected grain yield and above-ground biomass at the on-station sites (Domboshawa and Harare), but not at the on-farm sites (Shamva, Hereford and Madziwa) (Figs. 1 and 2, Tables S1 and S2). Both maize grain yield and above-ground biomass were significantly increased (P < 0.05, Table S2) under NT with residue retention relative to CT with residue removal at Domboshawa, by 100 and 50% (1.5 t grain ha−1 and 1.0 t above-ground biomass ha−1), respectively. At Harare, above-ground biomass increased (P < 0.05, Table S2) by 23% (1 t above-ground biomass ha−1) under NT compared to CT.
Overall and across all sites and management practices, maize grain yield and above-ground biomass were positively correlated (P < 0.0001) with total N, total C, MBC, NH4+-N, CEC and negatively correlated (P < 0.0001) with NO3−-N whereas EC was not correlated with grain yield and above-ground biomass (Fig. 3).
Fig. 3. Combined Pearson correlation analysis of all sites and management practices on combined sandy and clayey soils of Zimbabwe. GY, grain yield; APB, aboveground biomass; MBC, microbial biomass C; TC, total carbon; TN, total nitrogen; CEC, cation exchange capacity; NH4.N, ammonium nitrogen; NO3.N, nitrate N; EC, electrical conductivity.
Soil properties
pH, CEC and EC
Significant effects of management practice on soil pH and CEC were found at the two on-station sites, but not at the on-farm sites (Table 2 and Table S2). Soil pH was higher (P < 0.05) under NT (5.11) compared with CT (4.79) at Domboshawa and was lower under NT (5.28) compared with CT (5.38) at Harare, whereas CEC was significantly higher (P < 0.05) under NT at both sites (CEC of 1.58 and 1.01 meq 100 g−1 soil under NT and CT, respectively, at Domboshawa, and CEC of 13.61 and 12.75 meq 100 g−1 soil under NT and CT respectively, at Harare). There were no significant differences in pH across the on-farm sites, but CEC was significantly lower (P < 0.05) at the sandy Madziwa site compared to the clayey Shamva and Hereford sites (Table 2). EC was significantly (P < 0.05) affected by management at both Harare and Domboshawa with a 163% increase of EC under NT as compared with CT in Domboshawa (Table 2).
Total C and N concentrations
On-station, NT significantly (P < 0.05) increased total soil C and N concentrations at Domboshawa (both 59%) and Harare (22 and 23%, respectively) as compared with CT. Comparable treatment effects were observed for soil C and N stocks at these sites (Table S3). On-farm, NT with ripline seeding significantly (P < 0.05) increased total C and N concentrations compared with CT at Madziwa (61 and 78%, respectively) and Hereford (21 and 18%, respectively), but differences at Shamva were not significant (Fig. 4a–d). Comparable increases were observed in total C and N concentrations in soil under NT with direct seeding at Madziwa (63 and 80%, respectively) and Hereford (28 and 23%, respectively). Overall, total soil C and N concentrations were significantly higher (P < 0.05) at the clayey sites of Harare, Shamva and Hereford, than the sandy sites of Domboshawa and Madziwa (Fig. 4a–d).
Fig. 4. Total C and N concentrations in soils at two on-station and three on-farm research sites in Zimbabwe under conventional tillage with residue removal and different no-tillage practices (ripping, direct seeding) with residue retention: total C and N at sandy soil sites (a) and (c) and clay soil sites (b) and (d). Lowercase letters indicate significant (P < 0.05) differences between management practices at each site. Values are means ± one standard error of the mean.
Soil mineral N concentrations
At the Domboshawa and Harare on-station sites, soil NH4+-N concentrations were greater (P < 0.05) under NT (relative to CT) by 106 and 79%, respectively, whereas NO3−-N concentrations were relatively greater (P < 0.05) under NT at Domboshawa and Harare by 18 and 53%, respectively (Fig. 5a–d and Table S1). At the on-farm sites, soil NH4+-N and NO3−-N concentrations significantly (P < 0.05) varied between management practices at Madziwa but not at Shamva and Hereford (Fig. 5a–d and Table S2). At Madziwa, soil NH4+-N concentration was 148 and 155% higher (P < 0.05) under NT with ripping and direct seeding, respectively, compared with CT. Similarly, at Madziwa, soil NO3−-N concentration was higher (P < 0.05) under NT with ripping and direct seeding (+104% and +131%, respectively), compared with CT (Fig. 5a–d).
Fig. 5. Mineral N and microbial biomass C concentrations in soils at two on-station and three on-farm sites in Zimbabwe under conventional tillage with residue removal and no-tillage practices (ripping, direct seeding) with residue retention: ammonium, nitrate and microbial biomass C levels are shown in sandy soils sites (a, c and e) and in clay sites (b, d and f). Letters above the columns indicate significant (P < 0.05) differences between management practices at each site. Values are means ± one standard error of the mean.
Overall, soil NH4+-N concentrations were highest (P < 0.05) at Shamva and Hereford clayey on-farm sites, but NO3−-N concentrations were lowest (P < 0.05) at these sites (Fig. 5a–b). The highest (P < 0.05) NO3−-N concentration was at the Harare clayey on-station site (Fig. 5c and d).
Soil MBC
When averaged across management practices, soil MBC concentrations were lowest at the sandy Madziwa site among the on-farm sites, and concentrations were higher at the clayey Harare on-station site than at the sandy Domboshawa on-station site (P < 0.05). Management significantly (P < 0.05, Table S2) impacted MBC concentrations at the Harare site only, where soil MBC was 61% greater under NT compared to CT (Fig. 5e and f).
Relationships between soil properties, sites, management regime and maize productivity
Specific relationships between all soil properties are presented in Figure 3 and the Supplementary Figures S1–S6, grouped by soil texture and management. PCA showed that the variables considered here explained most of the variation observed at the sites and in the various managements (Fig. 6a). Averaged across all sites and management, total soil C, total soil N and CEC were the strongest predictors of maize grain yield (R 2 = 0.73, 0.71 and 0.64, respectively; P < 0.0001) and above-ground biomass (R 2 = 0.81, 0.81 and 0.75, respectively; P < 0.0001). However, they exerted a weaker influence on maize grain yield and biomass at the sandy sites of Domboshawa and Madziwa (Fig. 6b). The influence of management on maize performance varied between the sandy and clayey sites (Fig. 6b and c). At the sandy sites, no individual soil parameters significantly predicted maize grain yield or biomass. There was clear isolation of the CT system with relatively weak associations with any of the soil parameters signifying that CT of the sandy soil had minimal effect on all soil parameters. To the contrary, NT practices associated more with different soil parameters under sandy soils (Fig. 6b). At the clayey sites, there was a wider distribution of system management across the PCA ordination space. Total soil C and total soil N were the strongest predictors of maize grain yield (R 2 = 0.53 and 0.44, respectively; P < 0.05) and biomass (R 2 = 0.40 and 0.41, respectively; P < 0.1) at the clayey sites (Fig. 6c). Soil texture had a strong effect on relationships as shown by the different responses of the soil and yield parameters when the sites separated into either sandy or clayey sites (Fig. 6b and c).
Fig. 6. Principal component analyses (PCA) of soil properties and maize productivity (grain yield and above-ground biomass) in (a) all sites under both conventional tillage with residue removal and no-tillage practices (ripping, direct seeding) with residue retention, (b) sandy soils and (c) clayey soils. GY, grain yield; APB, aboveground biomass; MBC, microbial biomass C; TC, total carbon; TN, total nitrogen; CEC, cation exchange capacity; NH4.N, ammonium nitrogen; NO3.N, nitrate N; EC, electrical conductivity.
Discussion
Impact of management on maize productivity
In recent decades, there has been a growing body of research on the links between soil characteristics, tillage and crop performance, although studies under the specific environmental conditions of southern Africa and also integrating soil biological and physical fertility indicators remain limited (De la Cruz-Barrón et al., Reference De la Cruz-Barrón, Cruz-Mendoza, Navarro–Noya, Ruiz-Valdiviezo, Ortíz-Gutiérrez, Ramírez-Villanueva, Luna-Guido, Thierfelder, Wall, Verhulst, Govaerts and Dendooven2017; Eze et al., Reference Eze, Dougill, Banwart, Hermans, Ligowe and Thierfelder2020). Our study was conducted under both controlled conditions (i.e., in on-station trials), and under smallholder farmer-managed conditions to acknowledge the diversity and heterogeneity of more real farm situations. Across the sites with more fertile, clayey soils (Harare, Shamva and Hereford), management had no significant effect on grain yield, and NT with residue retention increased maize above-ground biomass only at the Harare site. This lack of management impact on grain yield and on-farm maize above-ground biomass reflects the inherent nutritional status and buffering capacity of clay soils in Zimbabwe, which do not respond as dramatically to a decline in soil fertility resulting from CT management without residue retention. Clay soils are also more aggregated, which protects SOM through micro-aggregates from faster decomposition induced by tillage and reduces nutrient leaching (Chivenge et al., Reference Chivenge, Murwira, Giller, Mapfumo and Six2007).
However, total concentrations of C and N were strong predictors of maize grain yield and above-ground biomass in the clayey soils (Fig. 6). Management practices influenced these soil characteristics, and increased total C and N concentrations and stocks under NT with residue retention at Harare and Hereford. This suggests that residue retention and increased physical protection of organic matter in clay soils, with reduced soil disturbance and increased water storage, likely had benefits for crop productivity, although this did not result in a statistically significant increase in grain yield. Higher total C and N concentrations and stocks under the NT practices are in accordance with increased SOM stocks under minimum soil disturbance and crop residue retention (Lal, Reference Lal2009), and may provide possible trajectory indicators for sustainability through SOM build-up. Management practices influenced both total concentrations and stocks of C and N similarly. Soil NO3−-N and MBC concentrations were also higher under NT at the Harare on-station site, likely due to higher microbial activity at this more fertile site. Conversely, the stronger correlation between MBC and total N under CT may be indicative of increased coupling between native SOM and microbial abundance in this inherently fertile soil, as opposed to responses of microbial activity to inputs of maize aboveground biomass, with a wide C:N ratio, under NT with residue retention. However, previous research by Gentile et al. (Reference Gentile, Vanlauwe, Van Kessel and Six2009) from Kenya and Mupangwa et al. (Reference Mupangwa, Nyagumbo, Mutsamba, Mhlanga and Mataruse2018) from Zimbabwe highlight the challenge of N immobilisation under NT with cereal residue retention as opposed to CT, which affect maize yields and suggests that crop rotations with leguminous crops are essential in overcoming this challenge.
Despite soil CEC only being significantly impacted by management at the on-station trials (Table 2), the positive correlations between CEC and total N, total C, NH4+-N concentrations and MBC under both managements (Fig. 3) are indicative of CEC being linked to SOM abundance. These correlations were stronger under CT, possibly reflecting that native SOM (as opposed to recent residue inputs of high C:N ratio) dominates in the provision of ion exchange sites in soil (Oorts et al., Reference Oorts, Merckx, Gréhan, Labreuche and Nicolar2007). As these impacts on CEC were in the absence of clear impacts on SOM stocks, this suggests that increased CEC may be an early indicator of accrual of stabilized SOM. As expected for predominantly negatively charged clays, NO3−-N concentrations were negatively correlated with CEC under both managements. Residue retention and reduced tillage had no significant effect on NO3−-N concentrations at Shamva and Hereford where concentrations were very low, suggesting a strong influence of the constituent 1:1 clays (mainly kaolinite) on the high CEC at these sites, and the potential for leaching of NO3−-N.
Responses to management varied between on-station and on-farm sites, most likely due to differences in soil mineralogy, cropping history, site heterogeneity and different timescales under NT, but also possibly reflecting biases arising from researcher vs farmer management. Our on-station and on-farm clayey soil sites had been 7 and 14/15 years under NT, respectively, with the effects of NT with residue retention on total C, total N and NH4+-N greater at the shorter duration on-station site (Harare). This may have been due to differences in management adherence and/or more consistent retention of crop residues on-station as compared with on-farm sites.
Management impact varies with soil texture
Given the strong influence soil texture has on soil biogeochemistry and microbiology, we grouped the sites according to soil texture—sandy (Domboshawa, Madziwa) vs clayey (Harare, Shamva, Hereford) soils—and compared management impacts between these two textural groups. Higher total C and N and NH4+-N concentrations under NT with residue retention may represent the indications of benefits of residue management and minimum soil disturbance in both soils. However, the concentrations of total C and N were lower in the sandy soils than in the clayey soils and were only correlated to maize grain yield and biomass at the on-station site (Domboshawa), where NO3−-N concentration was also increased under NT. Under CT, maize grain yield and biomass at the sandy sites were also lower (1811–2744 and 2191–3057 kg ha−1) than in the clay soils (4369–4907 and 4930–5555 kg ha−1). PCA and regression analyses showed the influences of total C and total N on grain yield and biomass to be significantly weaker in the sandy soils compared with the clay soils. This may reflect the greater buffering capacity, higher CEC and ability to protect C and N more efficiently in clay soils (Chivenge et al., Reference Chivenge, Murwira, Giller, Mapfumo and Six2007). Even under NT with residue retention, SOM build-up in the sandy soils did not provide the crop nutrient supply benefit or increase in microbial biomass required to enhance maize above-ground biomass or grain yield, possibly because sandy soils are intrinsically less protective of SOM (physical occlusion and organo-mineral interactions) and available soil water.
The responses of our sandy soils to NT appear to have been due to different drivers than responses in the clayey soils. Other studies have reported the benefits of NT systems with residue retention being larger on sandy soils than clay soils (Nyamangara et al., Reference Nyamangara, Nyengerai, Masvaya, Tirivavi, Mashingaidze, Mupangwa, Dimes, Hove and Twomlow2014; Steward et al., Reference Steward, Dougill, Thierfelder, Pittelkow, Stringer, Kudzala and Shackelford2018), including maize yield at sandy loam on-farm sites across Malawi, Mozambique, Zambia and Zimbabwe (Thierfelder et al., Reference Thierfelder, Matemba-Mutasa and Rusinamhodzi2015). Our results highlight that NT with residue retention will likely have a greater impact on sandy soils due to lower buffering capacity and improvements in soil fertility. In contrast, clayey soils, already endowed with greater buffering capacity due to nutrient storage in the clay minerals and micro-aggregates, will respond less to an increase in soil fertility in the short term. The influence of soil texture on the response to management in general may in part account for previous variability in response to NT practices with residue retention in southern Africa, meaning that practices may need to be fine-tuned to local conditions, and for which early indicators of trajectory toward sustainability will be invaluable.
Sustainable management of degraded soils
It is well recognized that good quality, healthy soils are essential for sustainable agriculture and food production. In Zimbabwe, ~70% of farmers, and particularly smallholder farmers, cultivate sandy soils (Nyamapfene, Reference Nyamapfene1991). Our results show the importance of management approaches for sustainable production, and how relationships between early indicators of sustainability as predictors of maize productivity differ between two broad soil textural groups. Given the limited impact of the NT practices on maize productivity on the sandy soils, especially in the on-farm plots, we highlight the importance of integrating NT and residue retention with other sustainable management practices for these highly degraded soils.
Sole reliance on inorganic fertilizer inputs in these sandy soils is not a sustainable option for most smallholder farmers, due to low availability and high cost. In addition, the low buffering capacity and lower CEC of these soils mean that nutrient retention is low, which may lead to non-responsiveness of soils to current fertilization strategies. We suggest that a combination of organic inputs (crop residues, manures) and mineral fertilizers will make a difference to maize productivity on these soils. These principles are also central for Integrated Soil Fertility Management (ISFM) (Vanlauwe et al., Reference Vanlauwe, Bationo, Chianu, Giller, Merckx, Mokwunye, Ohiokpehai, Pypers, Tabo, Shepherd, Smaling, Woomer and Sanginga2010), and may be further supported in combination with climate-smart agriculture practices such as NT. For sustainable management, an integrated approach, for example, combining NT with the incorporation of high biomass-producing grain legumes, green manures, even tree-based components, will be necessary to foster the return of soils to sustainable production (Thierfelder et al., Reference Thierfelder, Baudron, Setimela, Nyagumbo, Mupangwa, Mhlanga, Lee and Gérard2018). Diversification of current maize-based farming systems could ensure increased and diverse C input into the cropping system, for example, not simply a reliance on maize residues. Increased biomass input will raise soil C levels, potentially increasing SOM-associated physico-chemical and biological benefits such that resilience of these soils to sustainably support crop yields is significantly increased. There is a need to develop measures to reduce livestock grazing of crop residues on cropland during the long dry season in this region (e.g., Valbuena et al., Reference Valbuena, Erenstein, Tui, Abdoulaye, Claessens, Duncan, Gérard, Rufino, Teufel, van Rooyen and van Wijk2012).
Current cropping systems yielding on average 0.5–1 t ha−1 of biomass will be insufficient to close the large yield gap between our Madziwa sandy soil and Shamva clayey soil and would necessitate an additional annual return of at least 3.5–5 t ha−1 residue biomass input per year. This would need to be provided from other sources, such as integrating large biomass-producing cover crops and tree components. Previous research showed that for arid-tropical regions, a combination of 1.6 t ha−1 residue input with 2 t ha−1 animal manure with no N application may provide at least an additional 0.3 t ha−1 grain yield of pearl millet (Aggarwal and Power, Reference Aggarwal and Power1997), which shows the magnitude of investment needed. For smallholder farmers who cannot afford or access large quantities of mineral fertilizer, improving the soil through combined management of different inputs to enhance SOM levels and in turn soil fertility and water storage (Fageria, Reference Fageria2012) may prove to be an essential climate-smart approach enabling continued maize productivity under the threats of climate change. SOM management for sustainability will depend on soil characteristics, and our results show that consideration of textural differences is key to this.
Conclusions
Our findings indicate the trajectories of soil improvements with no-tillage and crop residue retention, that, in accordance with our hypotheses, were related to maize productivity and varied with soil type: (1) texture had a much stronger influence on maize productivity than soil management; (2) to influence maize productivity, there is a need for an increase in total C and N and NH4+-N which is particularly relevant for the sandy soils; (3) both total C and N were strong predictors of maize productivity in the clayey soils, but not in the sandy soils; (4) NT and residue retention strengthened the relationship between soil parameters and maize productivity in the sandy soils. Future research should focus on more sites, soil sampling points and harvest seasons.
We propose that sandy soils should be the priority target for wider promotion of NT and crop residue retention. Investment in soil C and gradually increasing the productivity of maize-based farming systems, especially under sandy soils, will further provide greater benefits as multiple soil indicators (physical, chemical and biological) will be positively influenced, including the nutrient and water-holding capacities, which will confer greater sustainability and climate resilience.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S1742170521000442.
Data
The data used for this study are held in the University of Edinburgh and CIMMYT repositories and can be made available on request.
Acknowledgements
This work was funded through the UK Global Challenges Research Fund administered by the Biotechnology and Biological Sciences Research Council [BB/P022936/1]. We acknowledge the MAIZE CGIAR Research Program (www.maize.org) who supported this research with staff time and logistical support. The CGIAR Research Program MAIZE receives W1&W2 support from the Governments of Australia, Belgium, Canada, China, France, India, Japan, Korea, Mexico, the Netherlands, New Zealand, Norway, Sweden, Switzerland, UK, USA and the World Bank.
Author contributions
All authors contributed to the writing of this manuscript. CT, EB, EP, LM, TD, and JC conceptualized the project. LM undertook the soil and data analyses, and BM supported the data analysis and development of Figures.
Open access
