RÉSUMÉ
Sequential statistical methods were used to maximise carotenoid production by a strain of Rhodotorula mucilaginosa, isolated from the Brazilian ecosystem. Initially, a factorial 2(5-1) experimental design was used, and the variables were pH and the levels of glucose, yeast extract, MgSO4.7H2O and KH2PO4. The nitrogen source (yeast extract) was the most important variable in enhancing carotenoid production; MgSO4.7H2O and KH2PO4 had a negative influence. The initial pH had no significant effect on carotenoid and cell productions. We further investigated the effects of glucose and yeast extract effects, using a second-order central composite design (CCD) to optimise carotenoid production, which was adequately approximated with a full quadratic equation obtained from a two-factor-2-level design. The analysis of quadratic surfaces showed that after 5 days of cultivation at 25ºC, the maximum carotenoid concentration (745 µg l-1) was obtained with 15 g l-1 of yeast extract and 20 g l-1 of glucose. The maximum carotenoid production (152 µg g-1) was obtained with 5 g l-1 yeast extract and 10 g l-1 glucose. Carotenoid formation was more sensitive to changes in yeast extract concentration than to changes in glucose concentration. Maximum cell production was achieved with 15-17 g l-1 of yeast extract and 15-20 g l-1 of glucose.