Simulating drought impact and mitigation in cassava using the LINTUL model

We adapted and used a crop simulation model based on light interception and use efficiency (LINTUL-Cassava) to improve our understanding of water-limited yields of cassava under rain-fed conditions in Southern Togo. Data collected in four different fields in two locations, Sevekpota and Djakakope, during two consecutive growing seasons from 2012 to 2014 were used. Data from Sevekpota in Year 2, when a larger amount of rainfall was received than in Year 1, were used for model parameterisation and calibration. The model was evaluated with data from Year 1 in Sevekpota and Years 1 and 2 in Djakakope. The model calibration and testing results indicated an overall good agreement between simulated and observed storage roots and total biomass dry matter. A decline in leaf area index (LAI) towards the end of the cropping season and the regrowth at the onset of the new rainy season matched fairly well with the simulated dormancy and recovery from the dormancy phase. The model also captured the decline in yield of storage roots due to leaf regrowth at the recovery from dormancy as observed in Sevekpota. Best harvest periods to minimise storage root losses can be identified on that basis. The assessment of the effect of drought as the difference between simulated potential yields, assuming water content at field capacity, and water-limited yields indicated that drought can cause 9-59% loss of yield. The largest yield loss was recorded in Sevekpota in Year 1, and was mainly due to water stress occurring between 78 and 125 days after planting. The best planting period simulated was around mid-February, which is one to two months earlier than the usual planting time in Southern Togo. Further experimental studies are required to confirm this finding and assess how this can practically fit into existing cropping systems. These findings enhance our understanding of water-limited yield of cassava and unveil possibilities of improving it in future.