Evaluation of some factors influencing the reliability of buffer soluble N recovery in feeds

Solubility of feed N in borate–phosphate buffer (buffer soluble N; BSN) is widely used in modern feed evaluation systems as a cornerstone for calculation of rumen N degradability. Two experiments were completed with the same eight feed samples of dried grass, barley grain, unheated rapeseed cake, soybean meal, sunflower seed cake, lupins, field beans and peas to address methodological problems with the borate–phosphate buffer procedure. In the first experiment, samples were incubated for 60 min followed by four methods to separate solutes and solids: centrifugation at 3000 × g (39 °C) for 15 min (CNT); passive filtration (FLT); centrifugation followed by sedimentation at 4 °C for 16 h (CNT-C) and centrifugation followed by passive filtration (CNT-FLT). Treatments were repeated on two days. Only the CNT-FLT method was apparently free from separation problems. There were filtration problems with lupins, peas and field beans and centrifugation problems from floating particles with grass, barley, field beans and peas. Standard deviation between days (SDRD) among methods was >50 g BSN/kg N for grass, barley and sunflower seed cake, while field beans and peas had (SDRD) < 10 g BSN/kg N. Sunflower seed cake was most sensitive to separation method, with a method standard deviation of 85 g BSN/kg N. Replication between days, method and feed × method contributed 0.48, 0.26 and 0.26, respectively of the variance not attributable to feed sample. Among all samples, SDRD was 57.0, 6.6, 9.8 and 28.4 g BSN/kg N for CNT, FLT, CNT-C and CNT-FLT, respectively. In the second experiment, only the CNT-FLT separation was used in a factorial design with incubation times (25, 60, 240 and 1440 min), and samples milled on either a hammer or a knife mill, both fitted with 1-mm screens. BSN values increased by 2, 8 and 17% for 60, 240 and 1440 min, relative to 25 min and was 21% higher for the hammer mill. Mill × feed interaction contributed 0.60 of the total variance (exclusive of feed), with large legume seeds (field beans, peas, lupins) having 30 to 40% higher BSN values after hammer milling. Effect of mill type was smaller or absent in small seeds. Overall SDRD was approximately 12 g BSN/kg N for both mill types. Results show that standardization of mill type is most important to obtain comparable BSN results among laboratories. The CNT–FLT method is recommended as the separation method of choice with low reproducibility error, and it was the only method without practical problems.