Modeling nitrogen and phosphorus interactions in intensively managed nursery soil-plant systems

Modeling simultaneous response to multinutrient fertilization to take advantage of element interactions is a promising approach in fertilizer prescription, but the practice is relatively unexplored in forestry applications. This paper demonstrates that simultaneous prescription of nitrogen (N) and phosphorus (P) fertilizer for tree seedling culture can be facilitated by response surface models generated from a multi-level factorial experiment designed to quantify growth, nutritional response and N x P interactions in white spruce [Picea glauca (Moench) Voss]. Seedlings grown from seeds in potted soil were tested in a 6 (N) x 4 (P) x 3 (replicates) factorial experiment in a greenhouse. Analysis of variance indicated significant quadratic trends in treatment responses and significant N x P interactions, hence quadratic response surface models were generated from the total biomass database. However, model accuracy (R(2) = 0.56) was low and the N x P interaction was underestimated when luxury to toxic N levels were included. Since the positive N x P interaction was enhanced by N additions at lower (deficient) levels but diminished at higher (luxury to toxic) levels, a model constructed from a data subset which excluded high N treatments markedly improved accuracy (R(2) = 0.88) and significantly increased the regression-coefficient estimate for the interaction term. Verification in a nursery field trial for a 3-yr rotation confirmed that this refined model improved fertilizer prescription by increasing acceptable seedling numbers and fertilizer recovery by plants in bare-root stock production.