Modeling repeated fertilizer response and one-time midrotation fertilizer response in loblolly pine plantations using FR in the 3-PG process model

Abstract

Productivity of loblolly pine in the southeastern US is frequently limited by soil nutrient availability. Therefore, fertilization is commonly used to increase nutrient availability and subsequent growth. This study used the soil fertility rating (FR) in the 3-PG model (Landsberg and Waring, 1997) to predict fertilizer growth response of loblolly pine stands. We used the growth response following fertilization in a regional study installed at 11 locations to determine how FR is affected by repeated fertilization with N and P starting at a young age. FR values in the stands that received no fertilization treatment, baseline FR, were determined using the previously developed relationship between site index and FR. Two locations that did not respond to biannual fertilization of N and P were considered non-nutrient deficient. The baseline FR values on the control plots at these sites averaged 0.9. We assumed that this was the maximum FR value for loblolly pine in the southern US in an environment where N and P are non-deficient. FR was then adjusted to this value in the fertilized plots at the other sites and used in 3-PG to predict growth. Using this adjusted FR value 92% of the variation in observed aboveground biomass in the fertilized plots was described by 3-PG simulated aboveground biomass. We then derived the relationship between response to repeated fertilization and baseline FR in the control plots. An inverse relationship was observed between baseline FR and fertilizer response. Baseline FR described 56% of the variation in the fertilizer response.We also developed a dynamic function to model the FR to accurately predict growth in loblolly pine stands fertilized one time with N and P during the middle of the rotation. We used data from a series of midrotation loblolly pine plantations across the southeastern United States to model change in FR (ΔFR). This was done using an optimization process where the temporal distribution of the ΔFR was modeled using the Weibull function. Baseline FR, intensity of N fertilization, and year since treatment were used as input regressors to model dynamics of FR following midrotation fertilization. The model accurately described temporal changes in FR that enabled 3-PG to accurately predict response to midrotation fertilization with N and P. When FR values generated using this function were input into the 3-PG model, 80% of the variation in the observed aboveground biomass in fertilized loblolly pine plantations was described by the 3-PG simulated aboveground biomass.