Influence of tillage depth of a cultivator on the incorporation of crop residues of winter barley in a chernozem soil

The experiment was conducted from July 7 to August 8, 2016, at the experimental farm Groß-Enzersdorf of the University of Natural Resources and Life Sciences Vienna. The annual average temperature for the area is 10.6°C with semi-arid climate conditions and an annual precipitation of 538 mm (1980–2009). During the field tests, a total of 62.2 mm of precipitation fell, with 54.6 mm being recorded from June 12 to 14, 2016 (own measurements with the rain sensor Rain Gauge IM523 of the weather station iMetos 3.3 (both Pessl Instruments GmbH, Weiz, Austria; position: latitude, 48.19653; longitude, 16.57015). The soil at the 3.5 ha experimental area is silty loam chernozem (20.2% sand, 59.6% silt, 20.3% clay) with a pH of 7.6 and 2.2–2.3% organic substance. Immediately before tillage, the average gravimetric water content at a depth of 0–0.30 m was 15.6% (measured by weighting of soil samples before and after drying in an oven at 105°C).

The winter barley, Hordeum vulgare var. distichon (L.) “Caribic”, was harvested using an Axial-Flow 2366 combine harvester (Case IH, Racine, Wisconsin, USA; working width 4.8 m with straw chopper) with a grain yield of 7.9 t/ha. After all measurements had been carried out on the untreated area, the soil was tilled with a CVT 6240 tractor (Steyr, St. Valentin, Austria) at a constant target working speed of 8 km/h. Tractor parameters were recorded via the CAN and the ISO bus system using a CAN-logger Series 5102 (RM MICHAELIDES SOFTWARE & ELEKTRONIK GMBH, Fulda, Germany). A Synkro 3030 three-row cultivator (PÖTTINGER Landtechnik GmbH, Grieskirchen, Austria) was equipped with 11 tines standard chisel shares with a tool distance of 0.27 m, leveling discs, and a cage roller. The TD was the performed at depths of 0.10, 0.20, and 0.30 m, corresponding to the area of application according to the operating instructions (Pöttinger, 2007). After cultivating, subsequent tests were carried out on the cultivated area and the samples were processed in the laboratory. For each of the 3 TDs, 15 measuring points were randomly distributed and the tests were carried out. To evaluate the straw distribution before tillage, 15 measuring points were randomly selected from the 45 measuring points. A R10 GNSS antenna in combination with a TSC3 field computer (both Trimble, Sunnyvale, California, USA) using a RTK correction signal (EPOSA, Vienna, Austria) was used to locate the measuring points.

Similar to many studies (Winnige et al., 1998; Raper, 2002; López et al., 2003; Hänsel et al., 2009; Walther, 2009; Mitchell et al., 2015), the line transect method was chosen to determine the soil coverage. In these experiments, the line length and number of measuring points were adapted to the 3 m working width of the cultivator. Along the 4.24 m long base of the isosceles triangle, a line with 25 measuring points at 0.10-m intervals was taut. All measuring points that completely cut a minimum 3-mm long straw particle were counted and multiplied by four, resulting in soil coverage factor in percent (%) of the total number of pearls.

As a second method, image analysis was used. Photos (resolution: 6,000 × 4,000 pixels; exposure time: 1/125 to 1/160 s; aperture: f/13 to f/11; focal length: 16 mm; ISO 200 B) were taken using a camera of the model Alpha SLT-A77V (Sony, Minato, Japan) at a distance of 1.5 m above the ground surface with an angle of view of 83° capturing an area of about 1.2 m × 1.8 m.

For the image analysis, a self-written program running in OpenCV 3.3.0 was used. The photos were first transformed from red, green, and blue (RGB) color space into the hue, saturation, and value (HSV) color space. With individual adjustment of the upper and lower thresholds of the three axes, HSV, the changing light conditions were taken into account and only those pixels were depicted, which represented straw. The ratio of straw pixels to the total number of pixels in the photo represented the surface residue coverage in percent (%) to the covered area. By individually adjusting the respective threshold values of color HSV, image analysis was able to achieve a satisfactory differentiation between straw and soil despite the spectral similarity and changing light and moisture conditions.

As in other studies (Staricka et al., 1991; Arvidsson, 1998; Raper, 2002; López et al., 2003; Mitchell et al., 2015), the surface amount of crop residues was determined by removing them within a defined area. As in the study by Arvidsson (1998), the area had an extension of 0.25 m². After manual collection and air drying, straw and soil residues were separated using a 2 mm sieve and manual sorting. The dry matter of the straw was determined by weighing after drying in an oven for 24 h at 105°C.

The incorporated crop residues were determined using soil cores (as done by Christian and Miller, 1986; Allmaras et al., 1988; Ball and Robertson, 1990; Staricka et al., 1991; Raper, 2002; Duiker and Beegle, 2006; Mitchell et al., 2015). The individual analyzed cores had a length of 0.05 m (cf., Raper, 2002). To avoid partial compaction and thus a distortion of the results by using only one long core, separate 0.05 m long soil cores were used. Core removal was performed with a device that was placed in an excavated profile pit next to the test location. After removing the surface straw, a sharpened 1.5 mm thick ST37 steel plate was driven in horizontally at a specified depth of 0.05 m, guided by the removal device. As a result, the 0.05-m segment was defined in advance and compression during sampling no longer had a falsifying effect on the result. The sampler (ST35 steel tube, 52 mm external diameter, 48 mm internal diameter, 2 mm wall thickness) was carefully hammered from above down to the plate. Before removing, the tube was rotated several times in order to cut off outstanding straw. After removing the core and the surrounding soil, the plate was pulled out and driven in again, but 0.05 m deeper. This process was repeated for all segments. Repeated sharpening of the tube and the plate prevented the straw being pushed into the soil instead of being cut. After air drying of samples, crop residues and soil were sorted using a 2 mm sieve. In addition, remaining small lumps and stones were removed by hand. After drying in an oven for 24 h at 105°C, the dry matter of the straw was determined by weighing.

Before soil tillage, 15 measuring points were analyzed with soil cores up to a depth of 0.35 m in order to determine any crop residues from the pre-crop. After soil tillage, the number of cores was selected depending on the TD, because new straw incorporation was not expected without tillage influence. Ball and Robertson (1990) reported that by loosening the soil, the depth of the working profile was greater than the original TD. Therefore, samples were always collected 0.05 m deeper than that corresponding to the actual TD. So, the 15 measuring points with 0.30 m TD were sampled at a depth up to 0.35 m (i.e., 7 soil cores), the 0.20 m TD up to 0.25 m (5 soil cores) and the 0.10 m TD up to 0.15 m (3 soil cores).

For the statistical analysis, SPSS Statistics 24 software (IBM, Armonk, New York, USA) was used. Owing to the high variance of the data, outliers were removed from the values of the surface and incorporated crop residues. As outliers, all values that differed by more than two standard deviations from the mean value were defined. Line transect data and image analysis data were analyzed using the Levene test, the one-way analysis of variance, and Tukey’s post hoc test. Data on surface and total incorporated residues were analyzed using the Levene test, the one-way analysis of variance, and the Tukey–Kramer post hoc test. In addition, the distribution in the depth segments (DSs) was analyzed using the multivariate variance analysis (MANOVA) with the Pillai–Spur test. Subsequently, univariate analyses and the Games-Howell post hoc test were conducted for each DS. The MANOVA was limited to the top three DS, as these were the only ones for which individual values were collected for all TD.

Data from the CAN and ISO bus were processed with MATLAB R2017a (The Mathworks Inc., Natick, USA) using a program specially written for the application. For the averaging of the process parameters, the turns were geometrically excluded from the calculation.

Before tillage, soil coverage of 100% was determined. After tillage, no significant differences (p = 0.136) could be detected by image analysis, whereas for the line transect method, a significantly higher (p = 0.001) soil coverage was observed for the TD of 0.30 m (Table 1). It was observed that soil coverage in these trials was higher with greater TD. In contrast, in the study of Walther (2009), both methods resulted in a higher degree of soil coverage with shallower tillage than with deeper tillage. Hänsel et al. (2009), on the other hand, observed no significant correlation between the TD of a cultivator (0.05, 0.10, and 0.15 m) and the degree of soil coverage. Figure 1 shows the marked differences in results with the two methods: The line transect method resulted in a high degree of coverage with all three TD, while the mean values (MV) of the image analysis were significantly lower for all three TD. In the studies of Walther (2009) and Winnige et al. (1998), the line transect method also resulted in a higher degree of soil coverage compared to image analysis. Pforte (2010) explained this systematic overestimation of the visual method by the fact that if a straw particle is located below a measuring point, the degree of soil cover is always valued equally irrespective of the actual amount of straw present.

Figure 1

Before tillage, surface crop residues had an MV of 10.49 t/ha (standard deviation, σ ± 2.38 t/ha) (Table 2). After tillage, there were significant differences (p < 0.001) in the amount of surface crop residues among the three TD. With 2.55 t/ha (σ ± 0.68 t/ha), at the TD of 0.30 m, significantly more crop residues were present on the surface than for 0.10 m (MV = 0.92 t/ha; σ ± 0.49 t/ha) and 0.20 m (MV = 1.39 t/ha; σ ± 0.54 t/ha). Although greater TD could be expected to leave less crop residues on the surface, in these experiments, the results (as well as those of the line transect method) showed the inverse effect. This could be due to the changed tool geometry resulting in a greater cutting angle at the surface for deeper tillage and thus a reduced mixing effect. In addition, the reduced speed over ground because of wheel slip and engine power limits at greater TD can also be a reason for a reduced mixing and thus incorporation of harvest residues.

Before soil tillage, there was an average of 2.08 t/ha of crop residues in the 0–0.35 m DS (Table 2). After tillage, no significant differences (p = 0.619) were observed in the 0–0.35 m DS between the TD in terms of the total incorporated crop residues, although the values showed wide scattering. However, in the experiments of Raper (2002), the two TD of a cultivator (0.076 and 0.152 m) also buried similar amounts of crop residues. In the study of Riley and Ekeberg (1998), also no significant difference was observed in the total organic matter content between three TD of a plough (0.10, 0.20, and 0.30 m). In contrast to plowing, where the depth of incorporation of residues increases with the TD (Eltom et al., 2015), this trend was not observed with reduced tillage.

The following results of the depth distribution of incorporated crop residues, also presented in Figure 2 and Table 2, are given in t/ha contained in one DS (each 0.05 m core and 0.35 m in total). After soil tillage, more than 90% of the crop residues were located in the uppermost two DS, independent of the respective TD; below the depth of 0.10 m, only a small proportion was present (Figure 3). These results are consistent with findings of several studies (Ball and Robertson, 1990; Grube, 2003; Duiker and Beegle, 2006; Dimassi et al., 2014; Eltom et al., 2015; Mitchell et al., 2015). For example, in the long-term experiments on reduced soil tillage of Tebrügge and Düring (1999), 60% of the crop residues were observed at a depth of 0–0.05 m and 30% at 0.05–0.10 m. Also Estler and Knittel (1996) reported that, depending on the type of cultivator, 50–60% of the crop residues were located in the top 0.05 m and 25–30% from 0.05 to 0.01 m. The predominant incorporation of crop residues into the uppermost 0.10 m of soil and the pronounced stratification over the soil profile seems to be a general occurrence in reduced tillage systems (Dimassi et al., 2014). However, in many studies, only shallow TD of a cultivator was used. Even though deeper tillage was used in this study, at a depth of 0.20 and 0.30 m, incorporation of crop residues was mainly in the near-surface layer and a pronounced stratification of crop residues was noticeable. This was also observed by Schulz et al. (2014) with a TD of 0.18 m. In the study of Christian and Miller (1986), only the practices involving plowing could incorporate significant amounts of straw at a depth of 0.10–0.15 m. When only a cultivator was used, most of the residues were observed in the near-surface layer (0–0.10 m) and 20–30% were observed also on the surface. The results of these experiments and those reported in the literature indicate that, with a cultivator and irrespective of TD, it is not possible to incorporate significant quantities of crop residues deeper than 0.10 m.

Figure 2

Figure 3

The MANOVA showed significant differences (p < 0.001) between the three TD for the three uppermost DS in relation to the crop residues incorporated. At a TD of 0.20 m, in the DS of 0.05–0.10 m, there were significantly more crop residues than at the other two TD (p < 0.001). No significant differences were observed in the DS of 0–0.05 m and 0.10–0.15 m (p = 0.081 and 0.105, respectively).

This indicates that greater mixing was achieved with the TD of 0.20 m, whereas this effect decreased again for the TD of 0.30 m. It seems that the mixing in of crop residues depends on the cutting geometry and the cutting angle at the surface. With an angle of less than 90° between ground surface and tines (seen in the direction of movement), the mixing in of crop residues is limited at a TD of 0.30 m. However, with all three TD, more than 90% of the incorporated crop residues was located in the upper 0.10 m and, in addition, the total amount of incorporated crop residues showed no significant differences.

The tractor data of these experiments (Table 3) showed a linear increase in the horizontal force acting on the rear hitch by increasing the TD. The change in TD from 0.10 to 0.30 m leads to an increase in draft force by a factor of 2.17. This caused a nonlinear enhancement in wheel slip too, resulting in increased fuel consumption by a factor of 2.67 (TD: from 0.10 to 0.30 m). In addition, the area efficiency decreased by 26% (TD from 0.10 to 0.30 m). These two parameters lead ultimately to a significant increase in tillage costs.