Premium
Predicting Flowering Dates in Wheat with a New Statistical Phenology Model
Author(s) -
Sharma Darshan L.,
D'Antuono Mario F.
Publication year - 2011
Publication title -
agronomy journal
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.752
H-Index - 131
eISSN - 1435-0645
pISSN - 0002-1962
DOI - 10.2134/agronj2010.0346
Subject(s) - vernalization , sowing , phenology , growing degree day , cultivar , growing season , frost (temperature) , agronomy , latitude , photoperiodism , biology , winter wheat , mathematics , horticulture , geography , meteorology , geodesy
Wheat growers in Western Australia (WA) value flowering date predictions for risk aversion from frost and terminal stresses. Currently used statistical models such as the “flowering calculator” (referred to as photothermal model as programmed by Tennant and Tennant [2000] [TT]) are less robust in colder regions and/or higher latitudes of WA. We developed a new statistical model (named DM) that includes a vernalization component in addition to photoperiod and temperature components, the three biological drivers of reproductive growth and development in wheat ( Triticum spp.). The model assumes that heatsums required for flowering comprise a variety specific minimum, plus some more that are linearly contingent on sowing day, daylength, and the statistical summaries of daily temperature profiles. The vernalization opportunity was quantified using a term named “coldsum” calculated as equal to the sum of daily periods under 5°C in hourly temperature profile, while “heatsum” referred to the periods above 0°C. Environments were characterized using a log‐log transformation on heatsums and a segmented‐lines approach on coldsums. The model coefficients were estimated using data from four sowing dates spanning prevalent sowing times in WA over 4 yr at three diverse locations. The DM predictions were statistically far superior to TT ( F probability < 0.001; RMSD 6.1 and 17.4, respectively) in all respect of location, season, sowing date, and cultivar factors when compared in independent datasets from six cultivar × sowing date trials across WA wheatbelt. Since DM relies on a structured application of seasonal profiles, it should be applicable to other geographic areas, cultivars, and crops.