Optimizing simulated fertilizer additions using a genetic algorithm with a nutrient uptake model

Abstract

Intensive management of pine plantations in the southeastern coastal plain typically involves weed and pest control, and the addition of fertilizer to meet the high nutrient demand of rapidly growing pines. In this study we coupled a mechanistic nutrient uptake model (SSAND, soil supply and nutrient demand) with a genetic algorithm (GA) in order to estimate the minimum addition of phosphorus necessary to meet the initial four-year demand. Optimal P additions were estimated using two pine root length density inputs scenarios, a mycorrhizal scenario, and a grass competition scenario. The low root length scenario required 32.5–35.0 kg P ha −1 (depending on the fitness criteria) to best match the four-year loblolly pine P demand. With higher root length density, only 14.5 kg P ha −1 was required to meet demand. Adding a large fungal length density of mycorrhizae changed the simulated P uptake kinetics so that two separate fertilization amendments were optimal, although the total P added was similar to the higher root scenario. GA fitness functions were adjusted to eliminate SSAND uptake underestimates smaller than loblolly P demand, at the cost of an additional 9 kg ha −1 (higher root scenario) or 1–2 kg ha −1 (mycorrhizal scenario). Optimal GA estimates for P addition with grass competition were 1.5–3 kg ha −1 higher, with a sharp fitness peak associated with decline of grass in the first growing season.