Seeding next to previous year’s crop row (near-row sowing) can increase grain yields on water repellent soils

Keywords: crop establishment, crop row placement, non-wetting soils, on-row sowing, preferential water infiltration, siliceous sand, soil water repellency, soil water repellence, strategic tillage, wax-degrading bacteria.

Soil water repellency associated with the surface layers of coarse sandy soils restricts crop and pasture production on 10 million ha of agricultural land in south-west Western Australia (WA) (van Gool 2016) and 1.4 million ha in South Australia (SA) (Unkovich et al. 2020). Hydrophobic materials of plant and microbial origin coat soil particles (Franco et al. 1995) and reduce agricultural production due to patchy wetting of surface soils after rainfall (Ritsema and Dekker 1994; Doerr et al. 2000), delayed germination and poor plant establishment, and increased risks from wind and water erosion (Bond 1964; Tate et al. 1989; Lowe et al. 2021).

A number of management strategies for water repellent soils have been developed and include: (1) amelioration where water repellency is decreased by changing the surface soil properties, e.g. by rotary spading; (2) mitigation where repellency is managed rather than reduced; and (3) avoidance where severely repellent soils are removed from agricultural production (Roper et al. 2015).

Roper et al. (2013) and Ward et al. (2013, 2015a) evaluated no-tillage as a ‘mitigation’ strategy for managing soil water repellency. They observed that crop roots from previous seasons were preserved under no-till seeding systems and acted as pathways for water infiltration into the soil, by-passing repellent surface soil layers and thus promoting plant growth. This led to the question: would seeding crops near the previous year’s crop row (near-row sowing) improve plant performance and grain yields on water repellent soils? Near-row sowing (sometimes referred to as on-row sowing) is usually aligned within 0.02 m of the previous year’s crop row to avoid disturbing existing root pathways; therefore, the term ‘near-row’ sowing has been adopted here.

Farmers have generally aimed to seed in between the previous crop rows (inter-row sowing) to reduce seeder blockages by residual stubbles (Scott et al. 2010) and to potentially reduce the risk of disease carry-over (Verrell et al. 2017). However, there is experimental evidence that compared to inter-row sowing, near-row sowing improves crop establishment in wheat and barley (Ward et al. 2015b; Desbiolles et al. 2019; McBeath et al. 2019) and reduces competition from weeds (McBeath et al. 2019). Near-row sowing can increase the risk of root and crown diseases compared to inter-row sowing (McCallum 2007; Verrell et al. 2017; Gupta et al. 2018) but this has not always led to increases in disease expression/severity (Gupta et al. 2018) or reductions in crop yields after consecutive wheat crops sown near-row (Verrell et al. 2017).

The purpose of this study was to quantify the benefits of near-row sowing in water repellent sands and explore some of the mechanisms that might be at play in four locations (Moora, Calingiri and Pingrup in WA; and Wanilla, Eyre Peninsula in SA). The hypotheses tested were: (1) near-row sowing under no-till improves plant performance (crop emergence/plant biomass) and grain yields compared with inter-row sowing on water repellent sands, but cultivation or stubble burning can reduce or eliminate these yield differences; (2) in near-row sown crops, soil water contents below crop or stubble rows exceed those in the inter-row; (3) in near-row sown crops, soil water repellency is less severe below crop or stubble rows than in the inter-row; and (4) over several years of continuous near-row sowing, wetter soil conditions under near-row sown crops increase the size of wax-degrading microbial communities which aid the remediation of soil water repellency.

Field experiments were conducted at four locations (three in WA and one in SA). Details of each of these sites are shown in Table 1, and main treatments and measurements are shown in Table 2.

Table 1.  Details of field sites for assessment of near-row sowing under different tillage and stubble managements: site locations, rainfall, soil properties and repellency status.

Table 2.  Details of field sites and experiments for assessment of near-row sowing under different tillage and stubble managements, including crops, main treatments and measurements.

At Moora (WA) and Wanilla (SA), opportunistic comparisons of crop growth and grain yields, for one season, between near-row and inter-row sown crops were enabled by changes in seeding spacing within individual tillage and/or stubble treatment plots, as outlined in the next section. At Wanilla, soil water contents were measured in the crop row and crop inter-row throughout the cropping season.

The benefits of longer-term near-row sowing were investigated at two other sites (Calingiri and Pingrup, WA) where near-row sowing had been practised for 4 and 5 years, respectively, and distinct soil characteristics in the row and inter-row had become well established. At these two sites, soil water contents, soil water repellency and communities of wax-degrading bacteria (at Pingrup only) were compared between the crop or stubble row and inter-row within near-row sown crops only.

In 2016, a field experiment was established with the original aim of testing tillage treatments as an amelioration strategy for water repellency. Plots of 18.3 m × 15 m were arranged in a randomised complete block design and replicated three times. The entire experimental site was deep ripped (using an Ausplow Easitill deep ripper, Cockburn Central, WA) to a depth of approximately 0.4 m to remove compaction as a factor in the experiment immediately prior to the application of subsequent tillage treatments. One-off (strategic) tillage treatments, undertaken in March 2016, included a control (no-tillage), shallow (0.1 m) tillage using an off-set disc and deep (0.35 m) tillage using a rotary spader, designed to bury water repellent topsoil and lift wettable subsoil to the surface (Davies et al. 2013, 2020) and thereafter, the plots were managed under no-till. The entire site was sown and managed by the grower utilising a fully-matched 12 m controlled traffic farming system. Plots were located so that they received no machinery traffic following implementation of the treatments, apart from a small plot harvester at the end of the growing season when the soil was dry. The severity of water repellency (determined using the molarity of ethanol drop, MED, method) of the topsoil (to 0.1 m depth) was measured for the control and after the application of the tillage treatments.

A barley (Hordeum vulgare L. cv. La Trobe) crop was sown on 16 May 2016 using a John Deere Conserva Pak (Deere and Company, Illinois, United States of America), 2 months after the one-off tillage treatments were applied. The following year a lupin (Lupinus angustifolius L. cv. Barlock) crop was sown using a Cross Slot no-till seeder (CrossSlot IP Ltd., Fielding, New Zealand) on 15 May 2017. Due to a change in row spacing between the seeding implements used in 2016 and 2017 seasons, the 2017 lupin rows (row spacing of 0.25 m) partially overlapped the 2016 furrows (0.28 m) which contained standing barley stubble. As a result, a sowing pattern consisting of approximately 3–4 rows sown on or very near (<0.03 m) the 2016 barley rows, created sowing conditions similar to those achieved by near-row sowing. Another 3–4 furrows were at such a distance from the 2016 furrows that sowing conditions similar to inter-row sowing were created. Therefore, the sowing pattern created at Moora in 2017 allowed for an opportunistic comparison of lupin establishment and growth under near- and inter-row sowing conditions within the same paddock, under three different tillage treatments.

Differences in lupin establishment between the near- and inter-row conditions were assessed by estimating the green ground cover (GGC) along the lupin rows from aerial images taken on 25 July 2017 (70 days after sowing; DAS) and on 30 August 2017 (107 DAS). Grain yields of lupin were estimated by manual harvest of 3 × 1 m lengths of crop per plot on 22 November 2017. The number of samples per replicate was six (no-till) and two each (off-set disc and rotary spader).

A field experiment originally set up to test two tillage treatments (no-till and annual cultivation) and two stubble treatments (retained and burned) on a water repellent sandy soil was established in 2011 and continued until 2014. The trial was designed as a randomised block with four replicates. Annual burning on designated plots was done prior to cultivation (to 0.1 m), which was achieved by one pass of a machine with narrow points on the day of seeding every year.

In 2013, a barley crop (Hordeum vulgare cv. Scope) was sown following a canola (Brassica napus L. cv. Hyola 474 CL) crop. Plots (25 m × 1.56 m) were sown in two opposite runs using a plot seeder with narrow points (six edge-on tynes and Morris seeding boots). In 2013, this unintentionally created one run with plants sown on the previous year’s row (near-row) and a second run where plants were sown on the inter-row within individual treatment plots. Canola stubble in the ‘stubble burned’ treatments was burned on 2–4 April 2013, and cultivation was done immediately before sowing barley on 30 May 2013.

Data on crop performance from each seeding run was kept separate and this enabled comparisons to be made between near-row and inter-row sown crops. There were eight treatments analysed: [two tillage treatments (no-till (NT) vs cultivated (CT)] × [two stubble treatments (retained (R) vs burned (B)] × [two seeding positions (near-row vs inter-row)]. Because of the opportunistic nature of the data collection, the replication was unequal, with three replicates (NT-R near-row sown; CT-R inter-row sown; NT-B inter-row sown); five replicates (NT-R inter-row sown; CT-R near-row sown; NT-B near-row sown); and four replicates (CT-B near- and inter-row sown). Soil water repellency (MED) of the topsoil (0–0.1 m depth, in the crop row) was measured on 8 June 2013. Crop emergence (20 June 2013) and grain yields at harvest (25 November 2013) were determined for both near- and inter-row sown crops. Soil water contents in crop rows and crop inter-rows at 10 randomly located pairs in each plot were measured on 9 June, 12 August, 25 October and 9 December 2013. Patterns of water infiltration using blue dyes were observed at the beginning of the trial in 2011 within a wheat (Triticum aestivum cv. Justica Clearfield) crop sown on 25 May 2011.

Barley (Hordeum vulgare cv. Vlamingh) was near-row sown on 13 May 2013 using a Deep Blade System (DBS seeder: Ausplow, Cockburn Central, WA) in a paddock where near-row sowing had been practised for 4 years prior under no-till. The crop was planted by the farmer in one run of 48 m long × 8.9 m wide.

Soil water contents were measured four times during the growing season at 40 random paired locations in the crop rows and crop inter-rows to determine changes associated with repeated near-row sowing. Soil samples were collected from 36 paired row and inter-row locations, but in order to provide sufficient sample for soil water repellency analysis, samples were bulked, reducing sample numbers to four per sampling time for each sampling position (crop row or inter-row).

At this site, the iTill precision seeder system (http://www.itill.com) had been used (commencing 2010) to achieve near-row sowing every year with an accuracy of <0.02 m. Soil water repellency, soil water contents and numbers of wax-degrading microorganisms were measured in the stubble row and inter-row during summer fallow in February–March in 2015, 2016 and 2017. On each occasion, measurements of soil water contents were made at 60 random paired locations in the stubble rows and stubble inter-rows to determine changes associated with repeated near-row sowing. Soil samples for measurements of soil water repellency were collected from 120 paired stubble row and stubble inter-row locations and bulked, reducing sample numbers to 24 for each sampling position. A second set of 120 samples from approximately the same locations was collected and bulked in the same way for enumeration of wax-degrading microbial communities. The soil samples and soil water measurements were located over an area of approximately 1.0 ha. Oats (Avena sativa cv. Carolup) were sown on 10 May 2014 and barley (Hordeum vulgare cv. Scope) on 14 May 2015 and 15 June 2016.

Samples of surface soil (0–0.05 m or 0–0.1 m) were collected randomly in the crop rows and inter-rows of each individual plot using a custom-made soil corer (25 mm diameter). The numbers and locations of samples collected are specified in the text above for each individual site.

The MED test is a measure of the severity of repellency (King 1981) and has been used extensively by researchers due to its simplicity. The MED test correlates well with direct measures of contact angle and with the water drop penetration test (WDPT) that measures the persistence of soil water repellency (Smettem et al. 2021). For the MED test, sieved (<2.0 mm) soil samples were dried at 105°C for 48 h and cooled to room temperature (20°C) before measuring potential soil water repellency (Dekker and Ritsema 1994). In this test, droplets of aqueous ethanol solutions prepared in 0.2 M increments were placed on the soil surface and the lowest molarity of the ethanol droplet that entered the soil within ten seconds was recorded as the MED value. Soils have been classified as wettable (MED = 0), slightly repellent (MED 0.2–1.0), moderately repellent (MED 1.2–2.2), severely repellent (MED 2.4–3.0), or very severely repellent (MED > 3.0) (King 1981).

Volumetric soil water contents at 0–0.12 m depth at the Wanilla, Calingiri and Pingrup sites were measured using a hand-held time domain reflectometer (HHTDR) probe (HydroSense; Campbell Scientific, Logan, Utah) using the supplied calibration.

Patterns of water infiltration into water repellent soils at Wanilla were observed using a 1% solution in water of ‘Brilliant Blue’ dye (All Colour Supplies Pty Ltd., Sydney, Australia). Approximately 5 L of solution was applied to an area of 0.7 m × 0.7 m (equivalent to ∼10 mm rainfall) using a Hill’s Garden Sprayer (Bunnings Warehouse, Australia) at a rate equivalent to 30 mm h⁻¹. Vertical cuts across planting rows were made using a spade to observe and photographically record patterns of blue dye ∼2 h after application and 9 days later following 27 mm of rainfall during the intervening period.

At the Moora site, differences in lupin establishment and canopy development between the near- and inter-row positions were assessed by estimating the GGC from aerial images. Geo-referenced RGB ortho-mosaic images of the experimental site were produced using Pix4d software (pix4d.com) from high resolution aerial imagery taken by an unmanned aerial vehicle (UAV) at an altitude of 35 m. Colour thresholding of the RGB images, applying the Canopeo classification tool (Patrignani and Ochsner 2015) in the QGIS software (QGIS Development Team 2019; https://www.qgis.org/en/site/; accessed 20 January 2019), was used to create binary images representing the GGC. Mean values (n = 36, no-till; n = 12, offset-disc and rotary spader) of GGC (as a percentage) of the lupin crop sown near-row and inter-row were estimated from areas selected in the RGB ortho-mosaic (size 0.75 × 14 m) that included three adjacent lupin rows.

Crop emergence (seedlings m⁻²) at the Wanilla site was determined 3 weeks after sowing. Plants were counted in two adjacent crop rows over a length of 1.0 m at six random locations in each plot, and the distance between these rows was used to calculate the number of emerged plants m⁻².

At Wanilla, crops were machine harvested by a plot harvester at maturity. At Moora, samples of crops were harvested by hand using electric shears to cut crop on either side of a 1 m length placed between crop rows in random locations in each plot. Samples were then threshed and the distance between rows was used to calculate yield (t ha⁻¹).

Wax-degrading microbial communities were enumerated using a miniaturised Most-Probable-Number (MPN) method described by Roper and Gupta (2005). The method selects specifically for the function of wax degradation by directly observing emulsification of coconut oil, which contains the same fatty acids known to cause repellency in sandy soils (Franco et al. 2000). Decimal dilutions (up to 10⁻⁶) of soil samples in sterile mineral salts solution were prepared and 0.1 mL of diluent was added to 2 mL of sterile mineral salts solution in 24-well plates (five replicate wells/dilution). Each well was then overlaid with a thin layer of sterilised liquid coconut oil combined with a beta-carotene colourant and incubated for 5 days at 30°C. Samples in which the oil was emulsified were scored as positive for surfactant production/wax-degradation and the MPN of wax-degrading microorganisms in the soil sample was calculated using a table of Most Probable Numbers (Alexander 1982).

Soil water repellency (MED) at 0–0.1 m depth, measured after applying the cultivation treatments, was 2.4 (severely repellent) in the no-till control, 1.8 (moderately repellent) for the off-set disc, and 0.0 (not repellent) for the rotary spader. No soil samples were collected to allow for a comparison between near-row and inter-row sowing positions.

Under no-till conditions, GGC was significantly (P < 0.001) greater for near-row sown lupins than for inter-row sown lupins, both early in the season (70 DAS) (Fig. 1a), and 5 weeks later (107 DAS) (Fig. 1b). For near-row sown lupins, there was no effect of tillage treatment, but for inter-row sown lupins, the off-set disc and spading treatments resulted in GGC values closer to those for near-row sowing, at both times of measurement.

Fig. 1.  Green Ground Cover (GGC) percentage of lupin (Lupinus angustifolius cv. Barlock; sown 15 May 2017) estimated from aerial images taken on (a) 25 July 2017 (70 DAS) and (b) 30 August 2017 (107 DAS) for plots sown near-row and inter-row following one-off tillage treatments (March 2016) and a preceding barley (Hordeum vulgare cv. La Trobe) crop (sown 16 May 2016) at Moora. Individual vertical bars represent l.s.d. for seeding position [1.7 (a) 5.5 (b); P = 0.05]. Capped bars represent s.e.m. [n = 36 (no-till); 12 (off-set disc and spader)].

Grain yields (t ha⁻¹) for near-row sown lupins [2.58 (no-till), and 2.04 (off-set disc)] were significantly (P < 0.001) greater than for inter-row sown lupins [1.09 (no-till), and 1.17 (off-set disc)] (Fig. 2). However, in the rotary spaded treatment, there was no difference in yield between near-row sown (2.40 t ha⁻¹) and inter-row sown (2.23 t ha⁻¹) crops both of which were statistically similar to the yields for near-row sown crops in the no-till and off-set disc treatments.

Fig. 2.  Grain yields (manually harvested) of lupin (Lupinus angustifolius cv. Barlock; sown 15 May 2017) on 22 November 2017 for plots sown near-row and inter-row following one-off tillage treatments (March 2016) and a preceding barley (Hordeum vulgare cv. La Trobe) crop (sown 16 May 2016) at Moora. Individual vertical bar represents l.s.d. for seeding position (0.34; P = 0.05). Capped bars represent s.e.m. [n = 18 (no-till); 6 (off-set disc and spader)].

Soil water repellency MED value (0–0.1 m depth, in the crop row), measured 9 days after cultivation and seeding, was 2.8 under NT and 2.6 under CT. Crop residue treatment resulted in MED values (0–0.1 m) of 2.7 (stubble retained) and 2.5 (stubble burned). All MED values at the Wanilla site were in the ‘severely repellent’ category.

Crop emergence (seedlings m⁻²) of near-row sown plants numerically exceeded those sown on the inter-row in all four tillage and stubble combinations (Fig. 3), but differences overall were just outside statistical significance (P = 0.059). There were no significant effects of tillage, crop residue management or treatment interactions.

Fig. 3.  Crop emergence of barley (Hordeum vulgare cv. Scope) sown either near-row or inter-row under four tillage and stubble treatments: no-till, stubble retained (NT-R); cultivated, stubble retained (CT-R); no-till, stubble burned (NT-B); and cultivated, stubble burned (CT-B) on 30 May 2013 at Wanilla. Individual vertical bar represents l.s.d. for seeding position (25; P = 0.05). Capped bars represent s.e.m. (n = 3 (NT-R near-row sown; CT-R inter-row sown; NT-B inter-row sown); 5 (NT-R inter-row sown; CT-R near-row sown; NT-B near-row sown); and 4 (CT-B near- and inter-row sown)].

Grain yields were significantly (P = 0.006) affected by seeding position (Fig. 4), and the yield of near-row sown crops (1.80 t ha⁻¹) was greater than that of inter-row sown crops (1.43 t ha⁻¹). There were no significant effects of tillage, stubble treatment or interactions for any treatment combination.

Fig. 4.  Grain yields (machine harvested) of barley (Hordeum vulgare cv. Scope) sown either near-row or inter-row under four tillage and stubble treatments: no-till, stubble retained (NT-R); cultivated, stubble retained (CT-R); no-till, stubble burned (NT-B); and cultivated, stubble burned (CT-B) on 30 May 2013 at Wanilla. Individual vertical bar represents l.s.d. for seeding position (0.25; P = 0.05). Capped bars represent s.e.m. [n = 3 (NT-R near-row sown; CT-R inter-row sown; NT-B inter-row sown); 5 (NT-R inter-row sown; CT-R near-row sown; NT-B near-row sown); and 4 (CT-B near- and inter-row sown)].

Volumetric soil water contents were measured for each plot (within both seeding runs) and there were no significant differences between stubble or tillage treatments [no-till, stubble retained (NT-R); cultivated, stubble retained (CT-R); no-till, stubble burned (NT-B); and cultivated, stubble burned (CT-B)] (data not presented). Therefore, values are presented as averages of all treatments for each sampling position at each date measured (Fig. 5). Soil water contents were significantly (P < 0.001) greater in the crop row than in the crop inter-row for all but the August measurement, and because of this, there was a significant (P < 0.001) interaction between sampling position and the date of measurement (Fig. 5).

Fig. 5.  Soil water contents (% v/v) in the 0–0.12 m layer in the crop row and crop inter-row of barley (Hordeum vulgare cv. Scope) sown on 30 May 2013 at Wanilla. Each value represents the average of four tillage and stubble treatment combinations: no-till, stubble retained (NT-R); cultivated, stubble retained (CT-R); no-till, stubble burned (NT-B); and cultivated, stubble burned (CT-B) for each sampling position, measured on 9 June; 12 August; 25 October and 9 December 2013. Individual vertical bar represents l.s.d. for sampling position × date (0.22; P = 0.05). Capped bars represent s.e.m. (n = 16).

Patterns of infiltration of aqueous blue dye solution in late September 2011 indicated that water flow in water repellent soils predominantly occurred down root channels in the surface repellent layer (Fig. 6a). Nine days later and after 27 mm of rainfall, the blue dye solution was washed further down the soil profile into the more wettable soil below, where it spread horizontally wetting this layer more uniformly (Fig. 6b).

Fig. 6.  Infiltration into water repellent soil of a 1% blue dye solution in water ∼2 h after application of the dye solution to the soil surface on 22 September 2011 (a) and 9 days later following 27 mm of rainfall on 1 October 2011 (b) in a wheat crop (Triticum aestivum cv. Justica Clearfield) sown on 25 May 2011 at Wanilla. Blue dye solution entered the soil via root pathways in the surface repellent layer (a), but below this layer water spread horizontally to more evenly wet the non-repellent soil (b).

Soil water contents were higher (P < 0.001) by up to 4% v/v in the crop row than in the inter-row during the growing season of a near-row sown barley crop (Fig. 7a). Over the same period, there was a significant (P < 0.001) decrease in repellency in the crop row (from a MED value of 3.3–2.9) but not in the inter-row where there was a small but significant (P < 0.05) increase during the growing season from a MED value of 3.5 in April to 3.8 in June and August (Fig. 7b).

Fig. 7.  Soil water content (% v/v) in the 0–0.12 m layer (a) and soil water repellency (MED value) in the 0–0.05 m surface soil layer (b) at four times during the growing season in the row and inter-row of a crop of barley (Hordeum vulgare cv. Vlamingh) sown near-row at Calingiri on 13 May 2013. Individual vertical bars represent l.s.d. for sampling position [0.4 (a), 0.2 (b); P = 0.05]. Capped bars represent s.e.m. [n = 40 (a); 4 (b)].

Soil water contents in the stubble rows at Pingrup were higher (by up to 1% v/v) than in the inter-rows in February–March of each year, although these differences were only statistically significant (P < 0.05) in 2017 (Fig. 8a). These values of soil water content corresponded with significantly (P < 0.001) lower levels of soil water repellency (by up to one MED unit) in the stubble rows compared with the inter-rows in all years (Fig. 8b).

Fig. 8.  Soil water content (% v/v) in the 0–0.12 m layer (a), soil water repellency (MED value) in the 0–0.05 m surface soil layer (b) and numbers of wax-degrading microorganisms (MPN) in the 0–0.05 m surface soil layer (c) in the row and inter-row of stubble from oats (Avena sativa cv. Carolup; sown 10 May 2014) and barley (Hordeum vulgare cv. Scope; sown 14 May 2015 and 15 June 2016) at Pingrup in February/March 2015–2017. Individual vertical bars represent l.s.d. for sampling position [0.71 (a), 0.26 (b), 17033 (c); P = 0.05). Capped bars represent s.e.m. [n = 60 (a); 24 (b, c)].

MPN estimates of the numbers of wax-degrading bacteria revealed larger community sizes in the stubble rows compared with the inter-rows (Fig. 8c). In March 2015 and 2017 these differences were significant (P < 0.01), but not in February 2016 when numbers overall were lower than in the other 2 years.

In a 10-year field trial at Munglinup in WA, Roper et al. (2013, 2021) and Ward et al. (2015a) measured soil water contents under no-till and cultivated systems and observed that crop roots from previous seasons behaved as pathways for water entry resulting in higher soil water contents under no-till than cultivation, despite soils being more repellent under no-till. These findings are supported by Blackwell (2000) who speculated that old intact crop root systems might behave as preferred pathways for water entry to soil. Furthermore, farmers were noticing that crop plants sown on or near stubble rows from previous seasons often performed better than those sown in the inter-row and were devising methods to achieve this either by (1) sowing at a slight angle from the previous year to intersect old rows or (2) developing equipment to accurately seed near old rows. Near-row sowing has become feasible with recent advances in GPS technology combined with innovations in seeding equipment (for example, the iTill system (http://www.itill.com)). The amalgamation of studies, reported here, aimed to quantify the benefits of on- or near-row sowing at four different sites, and at one site (Moora) to compare near-row sowing with a higher cost amelioration treatment (rotary spading).

Data from the Moora experiment support the first hypothesis that near-row sowing under no-till improves plant performance (plant biomass) and grain yields compared with inter-row sowing in water repellent sands, but cultivation can reduce or eliminate these differences.

At Moora, three one-off tillage treatments had been applied in 2016: (1) rotary spading; (2) off-set disc; and (3) no-till as a control. Intense strategic deep tillage by rotary spading has been used to mix and bury water repellent surface soil layers resulting in more wettable surface soils through which rainfall can infiltrate (Davies et al. 2013, 2020; Hall et al. 2020). Cultivation with an off-set disc is a less intense and shallower tillage treatment but can mix and dilute repellency in surface soil layers (Roper et al. 2013; Scanlan and Davies 2019), albeit to much less an extent than a rotary spader or inversion plough (Scanlan and Davies 2019). These effects were demonstrated by reductions in soil water repellency (MED value) in the top 0.1 m due to cultivation, from severely repellent (no-till) to moderately repellent (offset-disc) and fully wettable (rotary spading). Enhanced crop performance (as measured by GGC) of near-row compared with inter-row sown lupins was evident early in the season (70 DAS) and became more pronounced as the season progressed (107 DAS) (Fig. 1). At this latter stage, under no-till, near-row sown lupins had a GGC of 47% compared with inter-row sown plants (21% GGC), suggesting that the absence of root pathways for water flow was the primary reason for reduced cover in the inter-row sown crops, since there had been no reduction of repellency due to cultivation in this treatment. However, in both the previously cultivated treatments, GGC was similar between near- and inter-row sown lupins, suggesting that the cultivation treatments increased GGC through reduced repellency overall and not near-row sowing per se.

Grain yields at Moora were greater for near-row than inter-row sown crops except for the previously rotary-spaded treatment, where yields were the same regardless of sowing position (Fig. 2). This again implies that burial of water repellent surface soil coupled with lifting of wettable sub-surface soil was the primary driver of improved yields in the spaded treatment. Grain yields of near-row sown crops under previous no-till and off-set disc treatments were better than inter-row sown treatments (Fig. 2), indicative of improved water infiltration down root pathways that were either established in the severely repellent no-till treatment or newly formed by the barley crop in 2016 in the moderately repellent off-set disc treatment. There were no differences among near-row sown treatments regardless of previous one-off tillage treatment. Crops sown near-row under no-till yielded similarly to those sown near-row with off-set disc tillage or with rotary spader tillage (Fig. 2) without the additional costs of the tillage treatments (Davies et al. 2013) and without the associated loss soil carbon (Six et al. 2000) and risk of losses of soil water (Jones et al. 1969) and surface soil through erosion (Harper et al. 2010; Lowe et al. 2021), in the year of tillage. Nevertheless, these results are in the context of an already deep ripped paddock in 2016 prior to imposition of the tillage treatments and may not apply to all soil types, crop types and environments. Soils with extreme soil water repellency may require intense amelioration management, in the form of strategic (one-off) deep tillage or clay application (Hall et al. 2020), to achieve adequate in-row plant and root system densities capable of channelling sufficient quantities of water down the soil profile to benefit near-row sown crops in the following season. Furthermore, strategic deep tillage can overcome multiple soil constraints such as compaction, subsoil acidity through lime incorporation and improved subsoil fertility in addition to overcoming topsoil water repellency. This can result in larger and more sustained productivity increases (McDonald et al. 2019; Davies et al. 2020; Hall et al. 2020) than might be expected when using strategies such as near-row sowing to mitigate water repellency alone.

At Wanilla, the soils were severely repellent (MED 2.5–2.8). Crop emergence of near-row sown plants numerically exceeded those sown on the inter-row in all four tillage and stubble combinations (Fig. 3), but these differences overall were just outside statistical significance, likely due to (1) considerable variability and patchiness in seedling emergence, which is typical of severely water repellent soils and (2) the non-orthogonal treatment design, which resulted in limited degrees of freedom. Despite these limitations, there was a significant (P = 0.006) 26% yield gain (Fig. 4) associated with near-row sowing compared with inter-row sown barley in 2013. At this site, cultivation was less intense than at Moora and had a minimal effect on soil water repellency (MED value 2.8 under no-till; 2.6 under cultivation), indicating very little mixing and dilution of repellent surface soils, and little disturbance of root systems. As a result, there was no significant effect of cultivation on yield at this site (Fig. 4). Nor were there any significant effects of stubble burning on yield. In 2012 at Wanilla, the average canola yield was less than 0.5 t ha⁻¹ (M. M. Roper,P. R. Ward, N. Wilhelm, R. Kerr, S. F. Micin and T. Blacker, unpubl. data) and consequently there was very little stubble remaining in April 2013 when stubble burning was done, resulting in little difference in stubble load between the burned and retained stubble treatments in 2013. Hence, the results from the Wanilla site support the first hypothesis that near-row sowing under no-till improves grain yields compared with inter-row sowing in water repellent sands. However, because of the low-intensity tillage and minimal effects of burning, the data did not support the second part of the hypothesis; that cultivation or stubble burning can reduce or eliminate these yield differences. This contrasts with the findings at a 10-year trial on the south coast of WA where it was shown that cultivation and burning significantly reduced crop performance compared with no-till and stubble retention (Roper et al. 2013). However, at this south coast site, cultivation and stubble burning were applied in a paddock where no-till and stubble retention had been in place for >20 years, and soil organic carbon contents were 1.5–2% and coverage of surface crop residues was around 70%. The imposition of stubble burning and cultivation resulted in significant and rapid losses of soil organic carbon, surface crop residue cover and soil water contents, negatively impacting grain yields in just 1 year after the trial was implemented (Roper et al. 2013). At Wanilla, the soils had no history of long-term no-till and stubble retention, and soil organic matter averaged just 1.3% (Table 1) with surface crop residues almost non-existent (Fig. 6); thus, these Wanilla soils resembled those in the south coast trial that were run-down from annual cultivation and burning of stubble. Rebuilding soil organic carbon and surface crop residue cover at the south coast trial site, following the return to no-till and stubble retention, was extremely slow (>6 years; Roper et al. 2021) and suggests that at the Wanilla site, little improvement could be expected in no-till stubble retention treatments compared with cultivated and/or stubble burned treatments in the short-term. The slow build-up of soil organic carbon (Six et al. 2000; Sanderman and Baldock 2010) and associated water holding capacity (Lal and Kimble 1997; Libohova et al. 2018; Werner et al. 2020) that buffers crop yields against variable weather conditions (Williams et al. 2016), together with minimal surface crop residues to protect soil water (Ward et al. 2013) were likely significant factors contributing to the lack of tillage and stubble treatment differences observed at Wanilla.

There is a possibility that the improved growth of near-row sown crops compared with inter-row sown crops was due to accessibility of new plants to an existing arbuscular mycorrhizal fungi network. However, at the Moora site, the grain yields of near-row sown crops were statistically similar for all tillage treatments, and this would be unlikely given that intense cultivation can significantly disrupt fungal networks including mycorrhizal networks (Kabir et al. 1997). Furthermore, at the Wanilla site, the barley crop in 2013 was preceded by a canola crop in 2012. Canola is a non-mycorrhizal crop (Gavito and Miller 1998) and therefore, it is unlikely that enhanced grain production in near-row sown barley in 2013 was due to better access to any mycorrhizal fungi networks either.

Our data point to a hydrological explanation for the yield benefits seen under near-row sowing. Similar to observations made by Roper et al. (2013), blue dye solutions, applied to the surface of soil at Wanilla, flowed down the soil profile through root channels, by-passing the surface repellent soil layer (Fig. 6a). Once the solution reached the more wettable subsoil below, water moved horizontally, wetting up this layer more uniformly (Fig. 6b). These root pathways have been shown to persist well into the following season (Roper et al. 2013) and hence, have the potential to improve water availability to new crops sown in their vicinity. Measurements of soil water contents in the crop rows compared with crop inter-rows supported this visual evidence. Soil water contents at Wanilla could not be accurately assigned to near- and inter-row sown treatments, but mean measurements calculated on a plot basis (which included both near-row and inter-row seeding runs) indicated that soil water contents were frequently significantly larger in crop rows than in crop inter-rows (Fig. 5) and this was evident both during the cropping season and after harvest. The difference in soil water content between crop rows and crop inter-rows was only non-significant for the August measurement, when wet seasonal conditions during July were likely to have removed the impact of patchy infiltration associated with existing root pathways.

Measurements of soil properties in the row compared with the inter-row at two sites (during the cropping season at Calingiri, WA; and the off-season at Pingrup, WA) where near-row sowing had been practised for ≥4 years, support the second hypothesis that in near-row sown crops, soil water contents below crop or stubble rows exceed those in the inter-row; the third hypothesis that in near-row sown crops, soil water repellency is less severe below crop or stubble rows than in the inter-row; and at one site, the fourth hypothesis that wetter soil conditions under near-row sown crops are associated with an increase in the size of wax-degrading microbial communities which aid the remediation of soil water repellency.

Soils contain natural populations of bacteria capable of decomposing waxes responsible for soil water repellency (Roper 2004) but their activities in dry repellent sandy soils can be limited. Results at Calingiri showed that during the growing season (April–September inclusive), soil water contents were higher in the crop row than the inter-row by as much as 4% v/v (Fig. 7a) and this was associated with a progressive decline in the severity of repellency in dried soil samples from the rows compared with the inter-rows (Fig. 7b). Similarly at Pingrup, compared to the inter-row, wetter conditions in stubble rows (Fig. 8a) were associated with reduced soil water repellency in dried soil samples (Fig. 8b) and larger populations of wax-degrading microorganisms (Fig. 8c), at the end of summer when the measurements were made. This suggests that reductions in repellency in the row could be related to microbial degradation of waxes that cause repellency. The only exception to this was in 2016 when soils were significantly drier (1% v/v or less) than in the other two seasons (Fig. 8a) due to dry seasonal conditions, and there were no significant differences between water repellency in the rows compared with the inter-rows (Fig. 8b) or numbers of wax-degrading bacteria (Fig. 8c). These observations add further weight to the suggestion that reductions in potential soil water repellency associated with increased soil water contents in the crop or stubble row are due, at least in part, to microbial decomposition of waxes responsible for water repellency.

These findings are supported by Gupta et al. (2018) who showed that microbial biomass and microbial diversity were significantly greater in the crop and stubble rows than in the inter-rows in sandy soils at two locations in South Australia. However, these findings also raise the question: are new crop plants sown on or near old crop rows at greater risk of root diseases transmitted from old crop roots? Data from Gupta et al. (2018) suggest this may be true, but they found that increased risk of disease did not translate into disease expression in the new crop. Furthermore, they demonstrated, at two different field sites, that crops sown near-row had significantly less Rhizoctonia disease than those sown on the inter-row and suggested that this was due to higher microbial activity, microbial biomass and catabolic diversity of microbial communities that included disease suppressive organisms in the crop row. Similarly, there was no evidence of increased disease pressure in near-row sown crops in the trials at Moora and Wanilla. However, there is evidence of increased disease pressure in near-row sown crops, particularly with above-ground diseases such as Fusarium Crown Rot (McCallum 2007; Verrell et al. 2017). Clearly further research on the effects of near-row sowing on crop diseases is needed.

The results of our field trials support anecdotal evidence that the benefits for crop productivity on water repellent soils under the combination of no-till and stubble retention can be enhanced by seeding close to the previous season’s crop rows (near-row sowing) to take advantage of water infiltration down old root pathways. Compared with inter-rows, consistently wetter crop rows were associated with reductions in the severity of repellency, and at one site, with larger populations of wax-degrading bacteria. Hence, our findings show that there is a dual benefit of near-row sowing: directly through enhanced water infiltration down root pathways, and indirectly through reduced severity of repellency in crop/stubble rows.

The data that support this study will be shared upon reasonable request to the corresponding author.

The authors declare no conflicts of interest.

This research was funded by the Grains Research and Development Corporation (GRDC: Project Nos. CSP 00139 and DAW 00244) and supported by CSIRO, DPIRD (Western Australia) and SARDI-PIRSA (South Australia).