Two decades of agricultural drought impacts: remote sensing insights into vegetation productivity and phenological change in semi-arid Botswana.

Vegetation phenology needs to be considered in the assessment of agricultural drought severity, especially in drylands such as Botswana. Using Remote Sensing time-series data (2000 - 2020), this study evaluated vegetation productivity using the annual sum of the Enhanced Vegetation Index (EVIsum) and computed changes as trends (EVIsumtrend) in agricultural lands comprising grasslands and croplands. To assess agricultural drought severity, a weighted linear combination was applied to the EVI-based Vegetation Condition Index (VCIwlc) and compared to the conventional Standardized Precipitation Index. To detect how agricultural drought has impacted vegetation phenology, we identified dynamics in vegetation greenup, maturity, peak, senescence and dormancy and correlated these to the EVIsum and VCIwlc. Differentiating between croplands and grasslands, vegetation productivity in grasslands was consistently lower than in croplands during droughts. The seasonal agriculture-related phenology, such as late vegetation greenup - the start of the season - is correlated with reduced vegetation productivity and severe agricultural droughts as evidenced by lower EVIsum and VCIwlc values, respectively. Notable phenology patterns include delayed greenup in grasslands compared to croplands, while the peak period often overlaps between grasslands and croplands. Overall, phenological shifts detected over agricultural lands in Botswana were about 58 days delayed greenup between the earliest and latest detection, whereas for peak, senescence and dormancy, delays were between 60 and 61 days. Findings provide valuable insights into how vegetation responds over time to changing environmental stressors such as drought. Understanding and monitoring how drought impacts agriculture-related phenology in dryland ecosystems is essential to informing national drought policies.