Crop yields in East Africa, for example in central Ethiopia are low and year-to-year variability is large because of poor soils, harvest losses and high variation in water supply. Spreading the risk by diversifying production could be potential solution for smallholders.
The case of agricultural diversification
A major assumption in research about agricultural diversification is that it is a risk management strategy, almost like a “natural insurance” against extreme events. This is because if one crop fails another might grow. In this case we would expect to find that farming households with more stable crop yields also have higher diversity and vice versa. However, it is also plausible to assume that harvest losses due to extreme events are correlated and hence all crops fail.
It is difficult to test these hypotheses as multi-year household level micro data is typically not available. There is also no clear evidence on the relative importance of other drivers of yield stability. Understanding which strategies have stabilized yields in the past might help develop strategies to mitigate impacts from future extreme events and natural rainfall variability.
We use climate and survey data in several mixed effects models to understand statistical associations between the variables of interest and where we have reasons to believe that they influence each other.
For Ethiopia, four years of data from the Ethiopia Socioeconomic Surveys are available, spanning over 8 years from 2011 to 2019 but with data gaps for some years. There is significant variations in seasonal rainfall during this time leading to variation in crop growth conditions. In 2015 for example, rainfall was less than 50% of normal during the primary agricultural season June to September across large areas of central and eastern Ethiopia. There can be large variations in the timing and rainfall of the main and secondary rainy season between years.
We develop two separate mixed effects models, the first for farm diversity and prevalence of pastoralism with temperature and precipitation variability as predictors. We develop a second model for temporal crop yield stability with farm diversity, fertilizer use, irrigation use, pesticide use, a drought index, distance to markets, and livestock ownership as predictors.
We find a strong statistical association between farm diversity and climate variability. This is a result of long-term agricultural specialization to different climate zones. Farm diversity is highest in cooler and wetter areas with higher temperature variability and low rainfall variability. Pastoralism, with low farm diversity overall, is more common in warmer areas with lower temperature variability.
Crop diversity positively affect temporal yield stability, and showing a greater effect than irrigation, fertilizer, and pesticide usage. Application of fertilizer can increase yield but also crop water requirement and yield variability if water supply is inadequate. Quantities of agricultural inputs are also very difficult to measure in agricultural surveys. Together, these findings suggest that shifts in typical ranges of climate variability could destabilize farm-level crop yield for smallholders by limiting diversification opportunities
For more information
Read the research article published in Regional Environmental Change.
Read a science summary about the 2015 Ethiopian drought.