Forage yield gap analysis for tall fescue pastures in Argentina: A modelling approach

Abstract

AbstractA large gap between actual and potential herbage production of tall fescue is a major limitation for livestock production systems in Argentina. The objectives of this work were to (a) calibrate and test the ability of a published pasture–soil water model to represent herbage growth dynamics of tall fescue [Lolium arundinaceum (Schreb.) Darbysh.] under different growing conditions, using data from controlled field experiments; (a) use the evaluated model to predict the magnitude and time of the year that N or soil water would constrain tall fescue from attaining its potential growth rate in the south‐eastern Pampas of Argentina; and (c) quantify herbage production gains and temporal variability for a proposed improved management practice (IMP) of N fertilizer under different meteorological and soils scenarios. After tall fescue‐specific calibration, the model accurately represented the response of tall fescue herbage mass to irrigation and N fertilization (root mean square error of prediction <550 kg DM ha−1, R2 > .70) observed in thirteen single‐season, controlled, field experiments. Results from a 40‐year simulation show that the gap between fertilized and unfertilized tall fescue pastures in Argentina can exceed 14 t DM ha−1 year−1. The proposed IMP maximized responses to N in autumn and early spring, reducing the annual herbage production gaps by 38%. The effect of IMP on annual herbage production was larger (4–5 t DM ha−1 year−1) than the effect due to variability in soils’ ability to store water (2–4 t DM ha−1 year−1) or annual variability in meteorological conditions (±1–2 t DM ha−1 year−1). This work provides a sound modelling approach to identify, for a particular forage species and site, most of the main gaps of herbage production; and to quantify, without the need of long‐term, expensive field studies, some of the potential gains (and risks associated with temporal variability), that could be achieved by producers. In addition, our results clearly indicate that site‐specific calibration is required for the model to provide accurate predictions and sound management advice to farmers.