Forecasting quality losses of standing forage may aid producers when m
aking harvest decisions. The objectives of this research were to compa
re morphological, climatological, and chronological models as predicto
rs of forage quality in 'Kentucky-31' and 'AU-Triumph' tall fescue (Fe
stuca arundinacea Shreb.) receiving three N treatments; and test those
models for their ability to predict forage quality. Pure stands of Ke
ntucky-31 acid AU-Triumph tall fescue were fertilized to meet Universi
ty of Georgia (UGA) Cooperative Extension Service recommendations with
all nutrients except N. Stands were divided into individual plots whi
ch received either 50, 100, or 200 kg N ha(-1), and harvested weekly a
fter initiation of spring growth. In 1990 and 1991, tillers were count
ed, staged for morphological development, and weighed. Samples were gr
ound, scanned with a near infrared reflectance spectrophotometer (NIRS
), and a subset was analyzed chemically to develop calibrations for fo
rage quality. Near infrared spectrophotometric quality values were reg
ressed with mean stage count, mean stage weight, growing degree days (
MSC, MSW, and GDD, respectively) and calendar day, and regression equa
tions were validated with samples harvested from plots in 1992 and 199
3. The MSC and MSW equations had poor fit in AU-Triumph (range of R(2)
from 0.28-0.35 and 0.58-0.61, respectively) and were not validated. C
alendar day and GDD had predicted quality better (range of R(2) from 0
.76-0.90 and 0.74-0.87, respectively); validation statistics of those
equations had intercepts equal to 0.0 and linear coefficients equal to
1.0 for in vitro dry matter disappearance (IVDMD) and occasionally fo
r crude protein (CP), but not for neutral (NDF) or acid (ADF) detergen
t fiber.