1. Liquid-associated bacteria (LAB) were harvested from the liquid phase (LAB1) and from the solid phase of rumen contents after washing and manual shaking (LAB2). Solid-adherent bacteria (SAB) were recovered after washing and pummelling the total particles (SAB1). The distribution and the chemical composition of these three bacterial compartments were investigated in four dairy cows fitted with rumen fistulas. The animals received successively a diet consisting of one part hay and one part barley-based concentrate (diet C) and the same diet containing free soya-bean oil (79 g/kg dry matter (DM); diet So).
2. The efficiency of removal of SABI from total particles of rumen digesta collected I h after feeding, was calculated from the diaminopimelic acid content in particles and of the corresponding detached bacteria. It was 24% on diet C and 18% on diet So (P < 0.05), using a combination of homogenizing and ‘stomaching’ treatments in saline (9 g sodium chloride/1) (reference treatment). For diets C and So respectively it was lowered by Tween in saline solution (1 g/l; 22.7 and 17.8 %, not significant), but was increased when using a previous chilling (6 h at 4°) of homogenized particles before stomaching in saline (28.8 and 24.7%, P < 0.05) and in Tween 80 in saline (1 g/l; 26.6 and 20.8%, P < 0.05).
3. The extent of removal of SABI from the solid fraction of rumen digesta by the reference treatment decreased with decreasing particle size; it was at the highest for particles retained on 4 and 2 mm sieves (62.1–82.1 %) and still elevated for particles retained on 0.8, 0.4 and 0.1 mm sieves (41.3–57.9%). It was very much reduced for particles smaller than 0.1 mm (11.7–14.5 %), suggesting the occurrence of favourable conditions for the adhesion of SAB firmly resistant to removal (SAB2).
4. The concentration of total SAB (SABI +SAB2) in particles collected l h after feeding was lower (P < 0.05) in diet C (190 g/kg DM) than in diet So (234 g/kg DM). Values averaged 595–645 g/kg DM for particles smaller than 0.1 mm, but only 61 and 81–98 g/kg DM for particles retained on 4 and 0.4 mm sieves, and on a 0.1 mm sieve respectively. No significant differences were noted between diets but the effect of particle size was highly significant (< 0.1 mm v. others).
5. Postprandial variations of concentrations of total SAB on total particles exhibited a large increase I h after feeding in diet So (P < 0.05). Similar but amplified variations were observed for LAB in both diets (P < 0.05).
6. Total bacterial mass amounted to 213 and 231 g DM/kg whole-rumen contents DM in diets C and So respectively 6 h after feeding. Mean percentages of total SAB (69.8), LAB1 (7.3) and LAB2 (22.9) in total rumen contents were not significantly modified by the lipid level of the diet.