Using soil testing and petiole analysis to determine phosphorus fertiliser requirements of potatoes (Solanum tuberosum L. cv. Delaware) in the Manjimup-Pemberton region of Western Australia

Abstract

Field experiments were conducted over 3 years at 21 sites of varying phosphorus (P) fertiliser histories (Colwell P range: 9–170 g/g) in the Manjimup–Pemberton region of Western Australia to examine the effects of freshly applied (current) and previously applied (residual or soil test ) P on the yield of potatoes (Solanum tuberosum L. cv. Delaware). Phosphorus was placed (banded) at planting, 5 cm either side of and below seed planted at 20 cm depth, at levels up to 800 kg P/ha. Exponential [y = a – b exp (–cx)] regressions were fitted to the relationship between tuber yield and level of applied P at all sites. Weighted (according to the variance) exponential regressions were fitted to the relationship between yield responsiveness (b/a, from the yield versus level of applied P relationship) and Colwell P, and two P sorption indices—phosphate adsorption (P-adsorb) and a modified phosphate retention index (PRI(100)). A weighted exponential regression was also fitted to the relationship between the level of applied P required for 95% of maximum yield (P_opt; also from yield versus level of applied P) and P-adsorb and PRI(100). A weighted linear regression best described the relationship between P_opt and Colwell P.

Phosphorus application significantly (P<0.10; from the regression analysis) increased total tuber yield at all but 4 sites. Marketable tuber yield response paralleled total tuber yield response at all sites and averaged 85% of total yields (range 63–94%). Colwell P gave a good prediction of the likely yield response of potatoes across all sites. For example, the yield responsiveness (b/a) of potatoes in relation to Colwell P decreased exponentially from 1.07 at 0 g/g to 0, or no yield response, at 157 g/g Colwell P (R² = 0.96) i.e. the critical Colwell P for 95% of maximum yield of potatoes on soils in the Manjimup–Pemberton region. Similarly, no yield response (b/a = 0) would be expected at a P-adsorb of 180 g/g (R² = 0.69) or a PRI(100) of 46 (R² = 0.61).

The level of applied P required for 95% of maximum yield (P_opt) decreased linearly from 124 kg/ha on infertile sites (<5 g/g Colwell P) to 0 kg P/ha at 160 g/g Colwell P (R² = 0.66). However, a more accurate prediction of P_opt was possible using either P-adsorb or PRI(100). For example, P_opt increased exponentially from 0 kg/ha at <181 g/g P-adsorb (high P soils) to 153 kg/ha at a P-adsorb of 950 g/g (low P soils) (R² = 0.75) and exponentially from 0 kg/ha at a PRI(100) of <48 (high P soils) to 147 kg/ha at a PRI(100) of 750 (low P soils) (R² = 0.80). PRI(100) is preferred as a soil test to predict P_opt for potatoes in the Manjimup–Pemberton region because of its superior accuracy to the Colwell test. It is also preferred to P-adsorb because of both superior accuracy and lower cost as it is a simpler and less time consuming procedure — features which are important for adoption by commercial soil testing services. A multiple regression including Colwell P, P-adsorb and PRI(100) only improved the prediction of P_opt slightly (R² = 0.89) over PRI(100) alone.

When tubers were 10 mm long, the total P in petioles of youngest fully expanded leaves which corresponded with 95% of maximum yield was 0.41% (dry weight basis).

These results show that, while the Colwell soil P test is a useful predictor of the responsiveness of potato yield to applied P across a range of soils in the Manjimup–Pemberton region, consideration of both the soil test P value and the P sorption capacity of the soil, as determined here by PRI(100), is required for accurate predictions of the level of P fertiliser required to achieve maximum yields on individual sites.