| dc.description.abstract | Two end-user cassava root quality traits namely softness of boiled roots and starch content, greatly influence consumer acceptance, industrial value and market prices of cassava varieties in East Africa. Thus, with the increasing demand of cassava for food and non-food uses, investments have been made by breeding programmes to sustain and/or increase the rate of genetic improvement on the end-user traits. However, existing methods for assessment of these two end-user traits are tedious and have low throughput, hence curtailing the selection process in cassava breeding. Furthermore, there is no clear information on the diversity of starch content and root softness to guide systematic breeding. This study therefore was undertaken to develop high-throughput methods to enhance selection for the two cassava root quality traits. Specifically, the study; i) determined variations in root softness, ii) determined variations in starch content and iii) developed Near Infrared Spectroscopy models for prediction of starch content and root softness in cassava. Softness of boiled root samples was assessed using a penetrometer across three cooking times (15-, 30- and 45 minutes, across 57 cassava clones established in Namulonge. Similarly, softness of boiled root samples was assessed using a penetrometer across three root parts; proximal, middle and distal at 45 minutes cooking time among 119 cassava clones established in Namulonge. Starch content was analyzed on flour samples on 89 cassava clones established in Namulonge and Serere using acid hydrolysis and determination of sugars method. NIRS scans were taken on intact root samples for boiled root softness and on flour samples for starch content. NIRS Models were developed using WINISI software. Results showed that there was significant variation in root softness among clones at all cooking times (p ˂ 0.001). Wide variability was observed for root softness at 15 minutes cooking time and narrowest variability was observed at 45 minutes cooking time. Furthermore, there was significant variability among root parts (p ˂ 0.001). Noteworthy, highest variability was observed with the proximal section (0.489N - 6.36N) and least variability with the distal section (1.30N - 2.89N). There was significant variation in starch content among clones (p ˂ 0.001). Starch content ranged from 23.94 % to 75.23% on dry basis at Namulonge, whereas at Serere, it ranged from 21.34% to 76.32% dry basis. Likewise, clone by location interaction was significant (p ˂ 0.001), suggesting that starch content was dependent on location. For starch content NIRS calibrations, robust models with high coefficient of determination of calibration (R2c = 0.850, and 1-VR = 0.52) were obtained, indicative of higher performance and robustness for screening purposes. For root softness, low calibrations were developed (R2 c = 0.543, and 1-VR = 0.22). Standard Normal Variate and Detrending (SNVD) and Multiple Scatter Correction (MSC) were the best fit scatter corrections for NIRS prediction for starch content and root softness in cassava. Overall, the study results prove that there is indeed significant variation in root softness and starch content among cassava germplasm in Uganda and thus, justifying systematic genetic improvement for root softness and starch content, key end-user quality traits. Furthermore, the study shows potential for employment of NIRS technology to efficiently analyze starch content and root softness in cassava, as an alternative to traditional chemical methods. | en_US |
