All animals were subjected to a digestibility trial of 12 days, in April 2014. Fecal output was estimated using chromic oxide (CrO 3 ) as a marker, following Smith & Reid (1955) . The marker was packed in a paper cartridge that was provided, daily, orally, in a single dose (10 g animal - 1 ), in the morning of each day, between 06h30 and 08h00. After seven days of adaptation, feces samples were collected on the 8th (16h00), 9th (14h00), tenth (12h00), 11th (10h00), and 12th (08h00) days.
Feces were collected once per day, in the paddock where the animals grazed. Soon after spontaneous defecation, samples of feces were collected from the ground with care so as to avoid contamination by foreign bodies and identi ed. The feces were then frozen at -10 ºC and later processed and analyzed by atomic absorption spectrophotometry to measure the chromium content according to method INCT-CA M005/1, as per Detmann et al. (2012).
To determine individual supplement intake, titanium dioxide (15 g/animal) was mixed with the supplement immediately before its supply, following the same fecal collection scheme described for chromic oxide. The titanium concentration was determined according to method INCT-CA M-007/1 described by Detmann et al. (2012) . Readings were taken with an absorption spectrophotometer at the Laboratory of Physiology of the Department of Basic and Instrumental Studies at the State University of Southwest Bahia. The individual intake of concentrate was estimated by dividing the total TiO 2 excretion by its respective concentration in the concentrate.
Voluntary roughage intake was estimated using the internal marker indigestible NDF (iNDF). The marker was obtained after rumen incubation, in duplicate, of 0.5-g samples of forage, concentrate, and feces previously ground to 2 mm, inside non-woven fabric bags ("TNT", grammage of 100 g m - 2 , 5 × 5 cm), for 288 h, following method INCT-CA F-009/1 described by Detmann et al. (2012) .
Total DM intake (kg day - 1 ) was estimated as follows:
where FO = fecal output (kg day - 1 ), obtained using chromic oxide, MFe = concentration of the iNDF marker in the feces (kg kg - 1 ); MS = concentration of the iNDF marker in the supplement (kg kg - 1 ); SDMI = supplement dry matter intake (kg); and MFo = concentration of the iNDF marker in the forage (kg kg - 1 ).
Feces were collected once per day, in the paddock where the animals grazed. Soon after spontaneous defecation, samples of feces were collected from the ground with care so as to avoid contamination by foreign bodies and identi ed. The feces were then frozen at -10 ºC and later processed and analyzed by atomic absorption spectrophotometry to measure the chromium content according to method INCT-CA M005/1, as per Detmann et al. (2012).
To determine individual supplement intake, titanium dioxide (15 g/animal) was mixed with the supplement immediately before its supply, following the same fecal collection scheme described for chromic oxide. The titanium concentration was determined according to method INCT-CA M-007/1 described by Detmann et al. (2012) . Readings were taken with an absorption spectrophotometer at the Laboratory of Physiology of the Department of Basic and Instrumental Studies at the State University of Southwest Bahia. The individual intake of concentrate was estimated by dividing the total TiO 2 excretion by its respective concentration in the concentrate.
Voluntary roughage intake was estimated using the internal marker indigestible NDF (iNDF). The marker was obtained after rumen incubation, in duplicate, of 0.5-g samples of forage, concentrate, and feces previously ground to 2 mm, inside non-woven fabric bags ("TNT", grammage of 100 g m - 2 , 5 × 5 cm), for 288 h, following method INCT-CA F-009/1 described by Detmann et al. (2012) .
Total DM intake (kg day - 1 ) was estimated as follows:
where FO = fecal output (kg day - 1 ), obtained using chromic oxide, MFe = concentration of the iNDF marker in the feces (kg kg - 1 ); MS = concentration of the iNDF marker in the supplement (kg kg - 1 ); SDMI = supplement dry matter intake (kg); and MFo = concentration of the iNDF marker in the forage (kg kg - 1 ).
