Results from the research showed that the DGT soil P test can more accurately predict a response to applied P fertiliser that the commonly used soil P tests.

Of the fifteen replicated trials, only four were significantly responsive to applied P application (responsive sites); hence, eleven of the sites did not achieve significant increases in yield when P fertiliser was applied, owing to the high starting P status of the soil (non-responsive sites).

Analysis of the yield and soil chemical data indicated that the commonly used Colwell, Olsen and Bray2 soil P tests show a poor correlation between measured soil P concentration and relative yield* (Figures 1 – 3). This is particularly true for the P responsive sites, which is primarily due to the soils ability to tie up fertiliser phosphorous (high Phosphorous Buffering Index, PBI^). By way of example, yield penalties between 69 and 0% were incurred at measured Colwell-P concentrations between 40 and 60 mg/kg (Figure 1). Similar trends were observed for the Olsen and Bray tests.

In contrast, the DGT P test shows a strong relationship between DGT P values and relative yield (Figure 4) as confirmed by the high coefficient of determination (R2=0.74).

Trial results to date indicate that the critical DGT-P value required to obtain >90% relative yield is 171 µg/L; with a 95% confidence interval fitted to this value, the critical range is 75 – 310 µg/L.

  • This range is considered very conservative and will likely be refined with further research.
  • Many more trials are needed on moderate to high PBI soils to increase the certainty of critical values (likely to reduce).
  • DGT P values <75 µg/L indicate that moderate to substantial yield penalties would occur if no P fertiliser were applied.
  • DGT P values >310 µg/L indicate that >90% relative yield would likely be achieved with no P fertiliser addition (Fig. 4).

P responsive sites:

  • Of the four P responsive sites (blue data points on Figures 1 – 4 below), yield penalties of between 15 – 69%  were incurred when no P fertiliser was applied.
  • All four of these sites had starting Colwell-P and Bray-P concentrations (mg/kg) that exceeded the previously determined critical values of Maier et al (1989), yet significant yield penalties were observed when no P fertiliser was applied.
  • When PBI was taken into consideration, correction of the critical Colwell-P values found that the soil P concentration did not exceed the critical values of Moody (2007).
    • Results also found that only 50% of crop advisors were requesting the PBI test in addition to the Colwell P test.
  • Three of the four P responsive sites had starting DGT-P values <75 µg/L and PBI values ranging from 154 – 239 (moderate ability to tie up P fertiliser).
    • The P application rates required to achieve >90% relative yield at these sites ranged between 90 and 140 kg/ha, with maximum yields obtained with rates ranging from 100 to 210 kg/ha.
  • The one site that was P responsive that fell within the >310 µg/L zone was new ground that had recently been clay spread; hence, we suspect some applied P fertiliser was tied up by the clay, resulting in the observed P response. The P rate required in this case to achieve >90% relative yield was 40 kg/ha.
  • This data further confirms that the commonly used soil P tests are poor predictors of likely yield response to applied P fertiliser, particularly if the phosphorous buffering capacity of the soil is not taken into consideration.

Non-responsive sites:

  • There were a total of eleven project research sites that did not show a statistically significant response to applied P fertiliser (orange data points# on Figures 1 – 4 below) i.e. there was no significant increase in yield when P fertiliser was applied to the crop.
  • Many of these sites were dominated by shallow to deep sandy soils that possessed a low to very low PBI; hence, P in the soil is likely to be highly available for plant uptake.
  • For the eight non-responsive sites that fell within the >310 µg/L zone, the average starting soil P concentration (mg/kg) was Colwell-59, Olsen-34 and Bray-134, indicating that starting soil P reserves were well in excess of all previously determined critical values for these soil types.

Figure 1: Relationship between Colwell-P concentration and potato tuber response (R2= 0.0268).

Figure 2: Relationship between Olsen-P concentration and potato tuber response (R2= 0.1116).

Figure 3: Relationship between Bray2-P concentration and potato tuber response (R2= 0.0569).

Figure 4: Relationship between DGT values and potato tuber response  (R2= 0.74).


*Relative yield %: reflects the yield of tubers that was obtained when no P fertiliser was applied relative to the maximum tuber yield obtained ([yield with no P applied/max yield] x 100).

^Phosphorous Buffering Index (PBI): gives an indication of a soils ability to “tie-up” P; it is often used in conjunction with the Colwell phosphorus test. Soils with a high PBI will quickly bind P to exchange sites and make it unavailable for plant uptake. Conversely, soils with a low PBI will tie up only small amounts of P, leaving the majority of fertiliser applied P available for plant uptake. See for more information.

#There are only 10 data points shown for the non-responsive sites, owing to concerns about the accuracy of the measured Colwell, Olsen and Bray P concentrations at one of the sites.

NA Maier, KA Potocky-Pacay, JM Jacka and CMJ Williams (1989). Effects of phosphorous fertiliser on the yield of potato tubers (Solanum tuberosum L) and the prediction of tuber yield response by soil analysis. Australian Journal of Experimental Agriculture 29(3), p 419-32. CSIRO publishing.

PW Moody (2007). Interpretation of a single-point P buffering index for adjusting critical levels of the Colwell soil P test. Soil Research 45(1) p 55-62. CSIRO Publishing.

Project Partners

Potatoes South Australia PIRSA RSSA National Landcare Programme The University of Adelaide premiumgreen-logo