Environmental data
Based on the environmental requirements of GAS, the following climatic variables were selected: annual precipitation; precipitation of driest month, maximum temperature of warmest month and mean temperature of coldest quarter. These were extracted from the 10 arc-minute resolution WorldClim dataset (http://www.worldclim.org).
Precipitation data often does not reflect the amount of water input into irrigated areas. As such, the chosen variables relating to precipitation, annual precipitation and precipitation of driest quarter, were modified to reflect irrigation patterns. The irrigation correction was based on the general principle that irrigation compensates for water loss through evapotranspiration (Brouwer & Heibloem, 1986) and uses methods fully described in Federman et al., (2013). In summary, the difference between evapotranspiration and precipitation was calculated for both annual data and for data representing the driest quarter. These differences were then applied to annual precipitation and the precipitation of the driest quarter in irrigated areas, to give variables adjusted for irrigation. Irrigated areas were identified from (Siebert et al., 2005). Evapotranspiration data was extracted from ENVIREM (Title & Bemmels, 2018), whilst the precipitation data was extracted from the WorldClim dataset.
Multicollinearity amongst the four variables was tested for using the Pearson correlation coefficients. A value of more than 0.7 is considered to indicate variables which covary too much, however, this didn’t apply to any of the four variables.