Comparison of soil test phosphorus methods in neutral to calcareous Manitoba soils

Increasing concern for the amount of P entering lakes in Manitoba may lead to regulation of P concentration in agricultural soils. A possible means for this regulation is the use of soil test P. This may require a means of comparing soil test P analyses as various laboratories in Manitoba employ different methods of soil test P determination. Thus, the objectives of this study were to (i) compare the methods of P determination in Manitoba soils, and (ii) develop equations for converting different soil test P methods from one to another. One hundred and fifteen archived surface soil samples representing major soils of Manitoba were used for the study. Soil test P was determined in these soils using the original Kelowna (K1) method and the two modified Kelowna methods (K2 and K3). Mehlich-3, Olsen- and water extractable-P were also determined for all soils. The results were analyzed statistically and were related using a simple regression analysis model. The amount of P extracted by the different extracting agents varied widely. Mehlich-3 extracted the largest amount of P (a range of 5.4-200 mg kg(-1)) while water extracted the smallest amount (a range of 0.2-70 mg kg(-1)). The P extracted by the Olsen and the three Kelowna methods were intermediate between Mehlich-3 and water-P. Our results also showed that the three Kelowna methods were not significantly different from one another. The Olsen method compared well with those of the modified Kelowna methods, but extracted less P than the original Kelowna and Mehlich-3 methods. Overall, the different agronomic soil test P methods were well correlated to each other with correlation coefficients (r) ranging between 0.95 and 0.98. However, the correlations between these soil test methods and water extractable P was not as high. Water extractable P and the agronomic soil test P methods were better related by a non-linear relationship than by a linear relationship. The coefficient of determination (R-2) for all the regression equations relating the different agronomic soil test P methods ranged from 0.91 to 0.97. As such, the equations generated in this study can be used to convert the result from one soil test P method to another.