Valorisation nutritionnelle d’un mélange manioc–son de riz fermenté par Saccharomyces cerevisiae dans l’alimentation du tilapia du Nil (Oreochromis niloticus) et de la carpe commune (Cyprinus carpio)

Abstract

Feed is a major constraint in tropical inland aquaculture. This study evaluated a 60:40 (DM basis) cassava (Manihot esculenta)–rice bran mixture treated for 72 h at 30 °C with 2% baker’s yeast (Saccharomyces cerevisiae), 3% molasses and 1% urea in sealed bags after manual air removal. Five diets formulated to be approximately isonitrogenous (about 30% crude protein) and close in gross energy were prepared with 0, 15, 30, 45 and 60% treated mixture and fed in triplicate to 25 fingerlings per tank for 90 days. Apparent crude protein, measured by Kjeldahl nitrogen (N × 6.25), increased from 7.2 to 13.0% DM. The quantitative urea-nitrogen balance showed that 10 g urea/kg DM supplied 4.665 g N/kg, equivalent to 2.92 percentage points of crude-protein equivalent. This accounts for 50.3% of the total 5.8-point increase, meaning that 49.7% of the observed increase cannot be explained by the nitrogen initially supplied as urea. Crude-protein values are therefore interpreted as apparent crude protein derived from total Kjeldahl nitrogen. Fermentation also reduced crude fibre of the mixture (7.2 to 5.7%), cassava cyanide (42.1 to 8.3 mg HCN eq/kg), rice bran phytic acid (5.42 to 0.46 mg/g), and the rice bran free fatty acid index (12.8 to 4.6% oleic acid). F30 showed the best observed performance: weight gain increased by 18.4% in tilapia and 20.1% in carp relative to the control, while feed conversion ratio decreased from 1.63 to 1.44 and from 1.72 to 1.50, respectively. Quadratic fits located the weight-gain vertex at 29.8% for tilapia and 32.2% for carp, with 95% intervals of 27.5–32.1% and 29.9–34.8%. The operational interpretation is therefore an optimal zone close to 30% for both species. Survival was analysed using a generalized linear model with a binomial distribution, whereas water-quality parameters are presented as mean ± SE over the 90-day trial.