Managing water repellent soils: What three sites showed in 2025

By Simon Kruger & Chanel Duggan, WMG

Soil water repellence is a longstanding problem across the Geraldton and Kwinana West port zones. The region’s sandy soils are low in clay and low in surface area, and waxy compounds from decomposing organic matter build up on the sand particles over time. Water then struggles to move into the soil. The result is patchy emergence, uneven crops, lower yields, and flow-on problems for weed control and fertiliser use.

Amelioration is the main response. Deep tillage that mixes, inverts or delves the soil breaks up the repellent surface layer and lets water back in. On the region’s deep sands, where few rocks get in the way, growers have used these methods for years with consistent yield gains. The harder question, and the one this project is working through, is what happens on the more variable soils, the shallow sands over gravel and the true gravels, where the return on a heavy, upfront investment is far less certain.

In 2025 the project ran across three sites and three soil types. The continuing site at Dandaragan (Site 1) entered its second year of monitoring on shallow sand over gravel. Two new sites were added, sand over gravel and true gravel near Gillingarra (Site 2), and deep sand near Moora (Site 3). All three use unreplicated demonstration strips rather than replicated plots, so the figures here are best read as indicative. First-year differences at the new sites in particular cannot be cleanly separated from where each strip sits in the paddock.

Two patterns recurred across all three sites. Strong biomass early in the season did not always translate into proportional grain, and weed management after amelioration turned out to be the clearest practical lesson of the year.

Dandaragan, shallow sand over gravel (Site 1)

Site 1 carries the longest record of the three. It was ameliorated in 2024 and carried a lupin crop in 2025, giving a second year of data on the same ground under a different crop.

The first thing the second year showed is that the 2024 amelioration held. Surface water repellence was absent in every treated strip when the soil was tested in March 2025, a full season on, while one of the two untreated controls still carried a low repellence reading. Penetrometer readings told a similar story below ground. The leading treatments, the Plozza Plow and the Nufab single pass, kept a workable root zone well past the depth at which the controls became limiting, with the controls reaching the root-limiting threshold by around 100 to 125 mm. The deeper root zone opened up by the 2024 work was still there a year later.

Yield followed the same ranking as the year before. The Plozza Plow returned the highest grain yield on the measured swath at 2.24t/ha, 54% above the average of the two controls, with the Nufab double pass next at 2.06t/ha. This is the second season running that the Plozza has come out on top at this site. It also produced the highest yield and harvest index in the 2024 oat crop, and in both years it did so off modest biomass rather than high early bulk. In 2024 that reflected a dry finish, where strips carrying less biomass into a dry spring converted the moisture on hand more efficiently to grain. Mid-season biomass that year pointed to other treatments, and the dry finish reshaped the order by harvest. A yield result that repeats across two seasons and two crops is far harder to put down to chance or paddock position than a single year would be.

Figure 1. Grain yield by treatment at Site 1, middle harvested swath. The Plozza Plow and Nufab double pass lead the two controls, while the Fanger sits below them.

Not every treatment beat the controls. The Nufab single pass and the Fanger Plow both fell below the control average on yield. The Nufab single pass shows the gap between early growth and final yield most clearly. It produced the highest peak biomass of any strip at Site 1, at 5,240kg/ha, yet its grain yield sat just under the control average. Strong early bulk did not carry through to grain. A high-biomass, low-yield result like this can come from a poor harvest index, late-season moisture stress during grain fill, or variability within the strip, and the 2024 season showed the same disconnect where moisture was tight through the reproductive stages.

Weeds were the other clear result. Weed density was lower in all four ameliorated strips than in the worst of the controls, ranging from 26 to 37 weeds/m² against 122 in Control 1. Ryegrass was the dominant species throughout. Burying or redistributing the surface seed bank through amelioration appears to have set the weed burden back, at least in this first post-amelioration season. The grower also noted that most of the ryegrass counted at sampling was already dead, and that chemical control looked more effective on the ameliorated ground, though that last point is a field observation rather than something the trial measured directly.

Figure 2. Weed counts by treatment at Site 1. All four ameliorated strips sit well below Control 1.

Moora, deep sand (Site 3)

Site 3 gave the clearest first-year response of the three. It is deep grey-yellow sand, uniform and free of the gravel and rock that complicate the other two sites, and it was ameliorated in May 2025 ahead of a wheat crop.

Baseline testing found surface water repellence on this sand, with a moderate reading at the top of the profile. Deep, uniform sand of this kind is the soil growers in the region already have most confidence treating, and the results here fit that longer record.

The below-ground response was marked. The untreated control became limiting to root growth by around 450mm, while every ameliorated strip carried the probe deeper, with the Plozza Plow and Horsch Tiger strips reaching the full 750mm measurement depth. On this clean sand the penetrometer reading is a fairly direct measure of soil strength, without the rock contacts that can throw off readings on gravel, so the depth gain is easier to trust here than at the other sites.

Establishment followed the same pattern. Every ameliorated strip established more plants than the control, and the crop was visibly denser and more even across the ameliorated ground.

Figure 3. The PDS treatment (left of centre) against the untreated control at Site 3, showing denser, more uniform establishment on the ameliorated ground.

Peak biomass was where the response showed most clearly. The Plozza and Horsch strips reached around 10,800 and 10,910kg/ha, both close to 48% above the control. This was the strongest treatment response at this timing anywhere in the trial, and it came in the first year after amelioration. Two features of that response are worth noting. Stacking treatments did not build more crop, with the two single deep treatments, the Plozza and the Horsch, producing more biomass than the Plozza-Delver-Horsch and Plozza-Delver-Spader combinations, not less. And the response arrived late, with no clear difference at GS30 earlier in the season, a reminder that early growth on deep sand can understate what the crop does later.

Figure 4. Peak biomass by treatment at Site 3. The Plozza and Horsch strips sit around 48% above the control.

That biomass lead did not carry through to yield. The Plozza and Horsch strips, close to 48% up on biomass, converted it to a yield gain of only 2 to 3% over the control. The strongest strip yield came instead from the PDH combination at 5.37t/ha, 12% above the control, off a much more modest biomass gain. The rest sat within a few percentage points of each other.

Figure 5. Grain yield by treatment at Site 3. The biomass ranking above does not carry through to yield.

The whole trial sat on a better than average part of the paddock, with every strip, control included, yielding above the paddock average of 4.10t/ha, and a north-south yield gradient ran across the ground as well. In a single unreplicated year those factors cannot be separated cleanly from treatment. What the site shows plainly is the gap itself. On this deep sand, early crop bulk overstated the first-year payoff, and a decision made on biomass alone would have picked the wrong strips. Yield, measured across more than one season, is the figure that matters for the investment, and the full economic analysis planned for the end of the project will rest on it, not on early growth.

Gillingarra, sand over gravel and true gravel (Site 2)

Site 2 was the most difficult ground in the trial, which is what makes it useful. It carries two soil types, sand over gravel and true gravel, and 2025 was the host grower’s first amelioration on this ground, done with his own process. It was also the first outing at this site for the Fanger Plow. The results here are best read as early learning on a new tool and a first attempt on hard ground, not as a verdict on either.

The difficulty was evident from the outset. Large rock sat below the surface across both soil types, and some treatments pulled it up during amelioration, the Fanger most of all. A rock roller was needed to flatten the ground enough to seed. Seeding depth suffered as a result, with some barley plants found still under the surface weeks after sowing, and establishment across the site was uneven.

Water repellence at this site was not straightforward either. Laboratory testing returned no surface repellence, yet in the paddock the grower and the field team saw water still moving only down the previous year’s furrows, with volunteer wheat germinating along those lines while bare ground between them stayed dry. Areas holding stubble supported more even germination. The lab number and the paddock behaviour did not match. This is an observation across the site rather than a measured result, but it points to why paddock-scale assessment matters, and why a single lab test can miss what the soil is doing in the field.

Figure 6. Site 2 before seeding, with the top layer uncovered. Water movement was limited to the previous year’s furrows, visible as the darker lines.

Establishment and biomass showed no clear treatment differences at either sub-site, which is not surprising given the variability. Yield was harder to read still. In the sand sub-site the two untreated control strips yielded 0.89t/ha apart on their own, a wider gap than separated most of the treatments from each other. When two strips that received no amelioration differ by that much, position in the paddock is clearly driving a large part of the result, and treatment effects cannot be pulled out from underneath it in a single year.

The Fanger result is where the weed lesson comes through most plainly. Heavy ryegrass came up in the Fanger strips at both sub-sites. The two implements move soil differently, and it shows in the weed response. The Plozza inverts the profile and buries weed seed, while the Fanger mixes and lifts it, bringing dormant ryegrass seed up into the germination zone where moisture and light favour it. The flush is in part a sign of a soil that has been loosened and made more active by the disturbance, but on the true gravel it came at a cost. The Fanger strip carried the heaviest weed burden at the site and returned the lowest yield of any strip, at 2.09t/ha against 3.35 for the control and 3.17 for the Plozza, close to 40% down. On the same gravel the Plozza held level with the untreated control on both biomass and yield.

Figure 7. Heavy ryegrass in the Fanger strip at Site 2. The implement’s soil mixing action does not distinguish crop germination from weed germination.

Figure 8. Grain yield on the true gravel sub-site at Site 2. The Fanger strip, carrying the heaviest weed burden, sits close to 40% below the control.

None of this is a mark against the tool on its own terms. It is a first year, on the hardest ground in the trial, with a new implement and a grower ameliorating here for the first time. Whether the Fanger gap closes as the site settles will be one of the more useful things a second year tells us.

What the season points to

The clearest practical lesson from 2025 was about weeds. Aggressive soil mixing moves weed seed into the germination zone, and the implements do not tell crop from weed. Where amelioration buried the surface seed bank, as at Sites 1 and 3, weed density fell. Where it brought seed up, as in the Fanger strips at Site 2, ryegrass came away with the crop. Weed control belongs in the amelioration plan from the outset, not as an afterthought once the crop is up, particularly for implements that bring more soil to the surface.

Across the three sites the picture is uneven, which is the honest position after this much data. Site 1 shows a benefit holding into a second year on shallow sand over gravel. Site 3 shows a clear first-year response on deep sand, the soil type growers already treat with most confidence. Site 2 shows how much the outcome depends on the soil, the equipment and the operator’s experience once the ground gets difficult.

These are demonstration strips, not replicated plots, and one season cannot separate a treatment effect from where a strip happens to sit. Site 1 enters its third year in 2026 and the two new sites their second. A response that holds across seasons and crops is far harder to explain away as paddock position than one seen once, and the economic picture will firm up as yield builds across seasons, ahead of the full analysis at the project’s end.

The Soil Water Repellence Project (WMG2404-001SAX) is delivered by West Midlands Group with investment from the Grains Research and Development Corporation.