Environment Characterization in Sorghum (Sorghum bicolor L.) by Modeling Water-Deficit and Heat Patterns in the Great Plains Region, United States

Abstract

Environmental characterization for defining the target population of environments (TPE) is critical to improve the efficiency of breeding programs in crops, such as sorghum (Sorghum bicolor L.). The aim of this study was to characterize the spatial and temporal variation for a TPE for sorghum within the United States. APSIM-sorghum, included in the Agricultural Production Systems sIMulator software platform, was used to quantify water-deficit and heat patterns for 15 sites in the sorghum belt. Historical weather data (∼35 years) was used to identify water (WSP) and heat (HSP) stress patterns to develop water-heat clusters. Four WSPs were identified with large differences in the timing of onset, intensity, and duration of the stress. In the western region of Kansas, Oklahoma, and Texas, the most frequent WSP (∼35%) was stress during grain filling with late recovery. For northeast Kansas, WSP frequencies were more evenly distributed, suggesting large temporal variation. Three HSPs were defined, with the low HSP being most frequent (∼68%). Field data from Kansas State University sorghum hybrid yield performance trials (2006-2013 period, 6 hybrids, 10 sites, 46 site × year combinations) were classified into the previously defined WSP and HSP clusters. As the intensity of the environmental stress increased, there was a clear reduction on grain yield. Both simulated and observed yield data showed similar yield trends when the level of heat or water stressed increased. Field yield data clearly separated contrasting clusters for both water and heat patterns (with vs. without stress). Thus, the patterns were regrouped into four categories, which account for the observed genotype by environment interaction (GxE) and can be applied in a breeding program. A better definition of TPE to improve predictability of GxE could accelerate genetic gains and help bridge the gap between breeders, agronomists, and farmers.

Keywords: adaptation; climate; drought; simulation; stress.

FIGURE 1

Simulated vs. observed days to anthesis (A) and grain yield (B) for the hybrids Hybrid 1 (triangles) and Hybrid 2 (circles) for the evaluation testing trials. The slashed line represents the 1:1 line.

FIGURE 2

(A,D) Relative transpiration (RT) index (A) and grain temperature (GT) index (D) throughout the crop life cycle expressed in thermal time units for the identified patterns. Water stress patterns (WSP): WSP1, in green is a low water deficit condition; WSP2, in red is a preflowering water deficit condition; WSP3 in blue is a grain filling water deficit condition with no recovery at the ends of the crop life cycle; WSP4 in purple is a grain filling water deficit condition with recovery at the ends of the crop life cycle. Heat stress patterns (HSP): HSP1, in yellow is a low heat stress condition; HSP2, in orange is a moderate heat stress condition; HSP3 in dark red is a severe heat stress condition. Each line represents the index average for the seasons clustered under the same group. The RT and GT values go from 1 to 0, with 1 representing a no stress condition and 0 a complete stress condition. (B,E) United States Great plains map with the WSP (B) and HSP (E) frequencies are represented as pie charts. The pie charts are placed in the trial sites for the model evaluation. (C,F) Frequency of the occurrence of the different WSP (C) and HSP (F) over 36 years of climatic data for the whole studied region.

FIGURE 3

Simulated (A,C) and observed (B,D) yield distribution of the hybrids Hybrid 1 and Hybrid 2, within each (A,B) WSP and (C,D) HSP. Different letters inside each graph stand for significant differences in Student’s t-test.

FIGURE 4

(A) Observed yield distribution for the four environmental classification groups. (A, inset) Variance components for observed grain yield in field experiments are expressed as percentage of the total variance explained by each effect. (B) Principal component analysis for observed yield. The points represent the six evaluated hybrids and the vector the four environmental categories. (C) United States Great plains map with the environment category group (ECG) frequencies represented as pie charts. The pie charts are placed in the trial sites for the model evaluation.