COMPARISON OF ESTIMATES OF DIGESTIBILITY OF 2 DIETS FOR RAINBOW-TROUT, ONCORHYNCHUS-MYKISS (WALBAUM), USING 2 MARKERS AND 2 METHODS OF FECES COLLECTION

Estimates of apparent digestibility using an indirect method were made for two rainbow trout, Oncorhynchus mykiss (Walbaum), diets (NF and L T) containing 450 g kg(-1) fish meal. Two inert markers, titanium diox ide and chromic oxide, were used as digestibility markers. Faeces coll ections were made using two methods: stripping and the Choubert (tank) method. Faecal oil (g kg(-1)) and gross energy (GE; MJ kg(-1)) were h igher for the tank system, but differences were greater for NF (P < 0. 001 and P < 0.05 for method x diet interaction, respectively). Faecal titanium and chromium contents were also higher for the tank system, b ut the differences were greater for LT (P < 0.01 and P < 0,05, respect ively). The main method effects were significant for crude protein (CP ) and carbohydrate (CHO) content in that values were higher (P < 0.001 ) for the stripping method. The main diet effects were significant for ash and CHO contents (P < 0,001), with values being highest for LT an d NF, respectively. The only significant method x diet interaction for apparent digestibility was with oil (P < 0.01) in that tank values we re higher than stripping for NF but not for LT. Marker x method intera ctions were significant with dry matter (DM), ash, CP, GE, CHO (P < 0. 05) and oil (P < 0.01) in that titanium values were lower than chromiu m values for the tank method. Marker x diet interactions were also sig nificant with DM, ash, CP, GE, CHO (P < 0.05) and oil (P < 0.01) in th at titanium values were lower than chromium values for NF, Both method s of faeces collection have inherent weaknesses, but it is suggested t hat the Choubert method gives more accurate and repeatable estimates o f apparent digestibility of diets for fish.