Agricultural Crop-Zone Temperatures in the CPE Framework: Heat Stress Thresholds, Growing Degree Days, and the Reversal of the Paper 6 Sugar Signal

Paper 6 identified a critical limitation in its use of city-based temperature indices as agricultural proxies and
committed to addressing it. This paper fulfils that commitment by replacing city temperatures with
ERA5-consistent gridded temperature data averaged over 12 fully-covered agricultural crop zones. Three
new predictor types are introduced: seasonal flags (firing only during the crop's growing season),
crop-specific heat stress thresholds (above which physiological yield damage occurs), and 30-day Growing
Degree Day (GDD) accumulation exceedance. Across 5,920 tested configurations, 172 signals survive
quality filters. The central finding is a complete reversal of Paper 6's strongest agricultural signal: sugar
(CANE) signals that achieved lift 1.42–1.63× under city-temperature proxies produce zero surviving signals
on actual sugar production zone temperatures. Conversely, wheat and corn signals emerge strongly: US
Great Plains heat stress ® wheat (lift 1.60×), Ukraine/Russia wheat heat stress ® gold (lift 1.72×), and US
Corn Belt GDD ® wheat (lift 1.49×). Heat stress thresholds are the strongest predictor type (mean lift
1.397×, max 1.948×). The strongest signal in the entire CPE series is EU Urban Energy heat stress (>30°C)
® UNG at 21 days (lift 1.95×). A sign-flip permutation test (N=10,000 per signal) confirms all 10 pre-specified
key signals pass at p£0.001, with nine at p<0.0001 — not a single random permutation produced CPE as
high as observed.