Identification of Fertility Regimes that Enhance Long‐Term Productivity of Alfalfa Using Cluster Analysis

Recommendations regarding P and K fertilization and critical soil and tissue concentrations for alfalfa ( Medicago sativa L.) are inconsistent. Our objective was to use cluster analysis and binary logistic regression to determine if soil and herbage P and K concentrations can be used as diagnostics of P‐ and K‐limited alfalfa yields across soils with variable P and K soil test values. Samples were obtained from 28 consecutive harvests (four per year) of plots fertilized with four rates of P and five rates of K. Moderate to high rates of both nutrients were required to maintain high yields. Cluster analysis using May 2004 data for mass shoot −1 and shoots m −2 identified six clusters that also differed in yields. Averaged over the seven May harvests the two lowest yielding clusters had low herbage K (10.3–15.9 g kg −1 ), while intermediate yields were associated with low herbage P (1.6 g kg −1 ). Herbage P and K of the highest yielding cluster were 2.5 and 21.5 g kg −1 , respectively. Plant populations in 2004 were the greatest with intermediate yields (61 m −2 ) and lowest for the poorest yielding plots (6 m −2 ). High taproot N reserves were associated with high yields. Prediction of acceptable (high‐yielding clusters) vs. unacceptable (low yielding clusters) performance was best achieved using the combination of herbage K and P concentrations and their ratio as independent variables. Results illustrate a varying, interdependent nature for alfalfa critical P and K tissue concentrations and suggest the need to account for both nutrients simultaneously.