Can the reflectance spectrum be used to trace grass feeding in ewe milk

Fifty Comisana ewes were randomly divided into two groups of 25 animals each at 63 days after parturition. One group (control) received 3.5 kg per head per day of a total mixed ration based on maize silage. The second group (grass) grazed a pasture in Sicily 4 h per day and received 600 g per head of oats hay after returning from pasture. During each milking session (twice a day) all the animals received concentrate supplementation. Milk was sampled after 50 days of treatment. Milk reflectance spectrum and CIE L∗ a∗ b∗ colour were measured using a MINOLTA CM2022 spectro-colorimeter. The spectra were translated to have a value of reflectance at 530 nm = 0. The integral of the translated spectrum was calculated between 450 and 530 nm and was used to evaluate the possibility to discriminate milk samples from the two treatments. Milk from grass-fed animals was lighter in colour compared to milk from control ewes (L∗: 76.42 versus 72.80; P < 0.01) and had a higher b∗ value indicating a more yellow colour (5.13 versus 2.70;P < 0.01). Plotting together a∗ and b∗ allowed to obtain a good discrimination between the two groups. The absolute value of the integral was higher in the grass group compared with the control group (254.5 versus 150.2;P < 0.001). However, a discrimination through the integral value was imperfect as the absolute values of the control group ranged from 69 to 229 and the absolute values of the grass group ranged from 195 to 324. It is likely that a better discrimination would have been possible if the pasture was the only feeding source for the animals. © 2003 Elsevier Science B.V. All rights reserved.