Skip to main content
ARS Home » Research » Publications at this Location » Publication #82614
Title: MODELING SOYBEAN CULTIVAR DEVELOPMENT RATES USING FIELD DATA FROM THE MISSISSIPPI VALLEY

Author
item Acock, Basil
item PACHEPSKY, YAKOV - DUKE UNIVERSITY
item Acock, Mary
item Reddy, Vangimalla
item WHISLER, FRANK - MISSISSIPPI STATE UNIV.

Submitted to: Agronomy Journal
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 8/12/1997
Publication Date: N/A
Citation: N/A

Interpretive Summary: Some farmers use mathematical models of crop development and growth to predict how crop yield will respond to proposed management practices in various soil and weather scenarios. To make crop models effective for this purpose, we need a reliable understanding of how environmental factors affect their development. There are many different cultivars extant all with different development rates, and new cultivars appear each season. The only practical way of obtaining cultivar-specific parameters describing development rates is to use field data. However, these data are rarely collected more than weekly, giving about 15 observations of reproductive stage per season. Thus, the rates must be described with the fewest practical number of parameters. During farm testing of the soybean crop model GLYCIM, from 1993 to 1995, the development of five soybean cultivars was recorded on five or more sites. In these data were several relatively long periods when the crops remained in the same stage. We therefore fitted the data with a piecewise linear regression of reproductive stage on accumulated thermal units, with alternating horizontal and sloping segments. Stages up to the end of flowering were made functions of day of emergence, a surrogate for daylength. The length of the plateau at the end of seed fill was expressed as a linear function of soil water deficit (precipitation - potential evapotranspiration). Explicitly recognizing the developmental plateaus enabled us to more effectively separate the influence of daylength (day of emergence), temperature, and soil water deficit on development rates, and to model soybean reproductive stages using 10 cultivar-specific parameters.

Technical Abstract: To predict how soybean crop yield will respond to management practices in various soil and weather scenarios, we need a reliable understanding of how environmental factors affect soybean ontogeny. Soybean yields are highly dependent on the length and timing of the pod filling period. The many soybean cultivars extant all have different ontogenies, and new cultivars appear each season. The only practical way of obtaining cultivar-specific parameters describing development rates is to use field data. However, these data are rarely collected more than weekly, giving about 15 observations of reproductive stage per season. Thus, the ontogenies must be described with the fewest practical number of parameters. During farm testing of the soybean crop model GLYCIM, from 1993 to 1995, the development of five soybean cultivars was recorded on five or more sites using the development stages defined by Fehr and Caviness. In these data were several relatively long periods when the crops remained in the same stage. These developmental plateaus occurred near R2, R5, and R6. We therefore fitted the data with a piecewise linear regression of reproductive stage on accumulated thermal units, with alternating horizontal and sloping segments. Stages up to the end of R2 were made functions of day of emergence, a surrogate for daylength. The length of the R6 plateau was expressed as a linear function of soil water deficit (precipitation - potential evapotranspiration). Explicitly recognizing the developmental plateaus at R2, R5, and R6 enabled us to more effectively separate the influence of daylength (day of emergence), temperature, and soil water deficit on development rates, and to model soybean reproductive stages using 10 cultivar-specific parameters.