Obtaining climate data and phenotypic predictions in the past and future
To predict the adaptation potential of haplotype materials for past and future planting, meteorological data were acquired with the R package “raster” and the function “getData” (Supplementary Table S15). The precipitation and temperature data from 1961 to 2018 were retrieved from the historical monthly weather data section in WorldClim (https://worldclim.org/data/monthlywth.html). The predicted precipitation and temperature data from 2021 to 2080 were retrieved from the Future climate data section in WorldClim (https://worldclim.org/data/cmip6/cmip6climate.html). A spatial resolution of 10 minutes was selected, with the data aggregated to align with the scales of NOAA and USNO data.
September 20th was set as the simulated planting date for future predictions. The 166–195 interval for the environmental index PR was chosen, corresponding to March 5th to April 3rd. Precipitation data were extracted from this interval. Similarly, the 183–192 interval for the environmental index DTR was chosen, corresponding to March 22nd to March 31st. From this interval, minimum (tn) and maximum (tx) temperature data were extracted, resulting in DTR being defined as: DTR = tx − tn.
However, original models were constructed using data from NOAA and USNO. WorldClim data are on another scale, which was converted to the scales of NOAA and USNO. Subsequently, precipitation data in March from both databases between 1999 and 2018 in the seven locations were collected separately and fitted using a general linear model, revealing a correlation of 0.97 between the two datasets. A model,\(y\ =\ 1.107x\ +\ 0.052\), was established, where yrepresents the precipitation data from NOAA and USNO, while \(x\)represents the precipitation data from WorldClim. Using this equation, WorldClim data were calibrated to match the scales of NOAA and USNO for subsequent predictions.
The means for climate data spanning 20-year intervals in historical records were computed, resulting in three periods: 1961–1980, 1981–2000 and 2001–2018. Incorporating the three future periods 2021–2040, 2041–2060 and 2061–2080 resulted in six data periods for each location (Fig. 6C). To assess the performance of each haplotype under varying periods, the environmental data were incorporated into the established multi-locus model. Using a colored bar chart representation on a map, the haplotype with the highest SOC was displayed alongside its corresponding SOC value. Additionally, the gray bars in the bar chart depict the population mean (Fig. 6d).