Predicting germination response to temperature: II. Three-dimensional regression, statistical gridding and iterative-probit optimization using measured and interpolated-subpopulation data.

Authors: Hardegree, Stuart; Winstral, Adam

Submitted to: Annals of Botany
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 6/30/2006
Publication Date: 8/20/2006
Citation: Hardegree, S.P., and Winstral, A.H. 2006. Predicting germination response to temperature: II. Three-dimensional regression, statistical gridding and iterative-probit optimization using measured and interpolated-subpopulation data. Annals of Botany 98 (2) pp. 403.410.

Interpretive Summary: Seed germination response to temperature is an important factor that determines whether native range grasses are able to compete with introduced annual weeds during the early spring. In order to develop strategies to successfully establish these species, we need the ability to predict how their seeds will germinate under expected temperature conditions in the field. Previous models for predicting germination response to temperature are relatively inaccurate, and/or require a large number of computational steps. In this study, we developed more accurate methods for estimating germination response to temperature that are also relatively easy to derive. Use of these models will help us design planting and weed control strategies to optimize establishment success in rangeland seedings in the Intermountain western United States. Similar models can also be adapted and applied to crop and horticultural plants.

Technical Abstract: Cardinal-temperature and 2-dimensional regression models are commonly used for predicting thermal-germination response. These models are parameterized with simulated-data, interpolated from cumulative-germination-response curves. The purpose of this study was to evaluate the accuracy of three-dimensional models for predicting cumulative germination response to temperature. Three-dimensional models are relatively more efficient to implement than two-dimensional models and can be parameterized directly with measured data. We germinated seeds of four rangeland grass species over the constant-temperature range of 3 to 38 'C and monitored subpopulation variability in germination-rate response. We estimated subpopulation-specific germination rate as a function of temperature and residual error for an optimized-regression model and a statistical-gridding technique using both measured and interpolated subpopulation data. These models were also compared to 2-dimensional model formulations in a previous study. Three-dimensional regression analysis provided relatively poor estimates of both germination rate and germination time compared to all other models. Statistical-gridding eliminated several data transformation procedures and provided relatively accurate estimates of both germination rate and germination time over the tested-temperature range. Use of measured data for model evaluation provided a more realistic estimate of predictive error did evaluation of the larger set of interpolated-subpopulation data. Statistical-gridding techniques may provide a relatively efficient method for estimating germination response in situations where the primary objective is to estimate germination time. This methodology allows for direct use of germination data for model parameterization and automates the significant computational requirements of the 2-dimensional piece-wise-linear model, previously shown to produce the most accurate estimates of germination time.