strong effect on the amount of P recovered from these five soils. The negative effect of the buffering capacity on the soil P levels extracted with MIP, M3P, BIP, and, to a certain extent, Resin-P indicates that these extractants are sensitive to the capacity factor: M3P was the most
sensitive and Resin-P was the least sensitive.
Regression equations between percentage of yield and each of the four extractants, using a quadratic model, were performed, and the results are given in Table 3. These results show that the least sensitive extractant (Resin-P) was the best correlated with yield (R2 = 0.60), while MIP was the worst (R2 = 0.38).
The percentage of P in leaves predicted as a function of extractable P (Table 4) showed the same tendency as the relationship between yield and MIP, M3P, and BIP. However, the relationship between leaf P and Resin-P was no better than those of the other extractants, as was the case with yield. This is apparently due to the small size of the data set used for leaf P.
Conclusions
Among the four soil P tests, the overall results indicate that both M3P and BIP have about the same potential for use in the Cerrado area of Brazil to assess soil P.
Resin-P produced the best correlation between percentage of yield and soil P alone. However, with any of the four soil P extractants, inclusion of soil properties that estimate the P-buffering capacity may enhance soil test interpretations for P.
298