Relationship between in vitro and in vivo methane production measured simultaneously with different dietary starch sources and starch levels in dairy cattle

• We compared in vitro and in vivo methane production in dairy cattle simultaneously.
• Increased starch fermentation reduced both in vitro and in vivo methane production.
• In vitro methane is related to in vivo methane expressed per unit fermented organic matter.
• In vitro methane is unrelated to in vivo methane expressed per unit ingested feed.
• In vitro methane measurement can be indicative of the trend of in vivo methane.

To investigate the relationship between in vitro and in vivo methane (CH4) production measured simultaneously using the same rumen-fistulated cows in both experiments, four dietary treatments based on concentrate that accounted for 400 g/kg of the mixed diet DM, were formulated to contain starch varying in rate of fermentation (slowly (S) vs. rapidly (R): native vs. gelatinized maize grain) and level of inclusion (low (L) vs. high (H): 270 vs. 530 g/kg of concentrate DM). Sixteen rumen-fistulated lactating dairy cows were used in a complete randomized block design with these treatments replicated in four periods of 17 d each. In experiment 1, after 12 d of adaptation, the cows were housed in respiration chambers for 5 d to measure CH4 production. In experiment 2, in each period in vitro gas and CH4 production were measured (in duplicate per period) for mixed diet samples from the same diet as fed to the donor cows using rumen inocula adapted to the respective diets for an average of 16 d. In addition, samples of two concentrate ingredients, viz. grass silage and beet pulp, were incubated with four different inocula obtained from individual donor cows. Gas production (GP) was measured using automated GP system with CH4 measured at distinct time points. In vitro (24-h) CH4 production of mixed diet was lower with R than S (42.9 vs. 49.5 ml/g of incubated organic matter (OM); P=0.004), and higher with L than H (49.8 vs. 42.6 ml/g of incubated OM; P=0.002). A significant interaction effect between source and level of starch (P=0.015) was also found, indicating the CH4 production of the RH diet decreased in particular. In vivo, an increased rate of starch fermentation resulted in a lower CH4 per unit of estimated rumen-fermentable OM (eRFOM; 55.6 vs. 61.2 ml/g of eRFOM; P=0.007), and higher level of starch tended (P=0.089) to reduce CH4 per unit of eRFOM, but dietary starch level and source did not affect CH4 per unit of OM consumed. Across the diets tested, 24-h in vitro CH4 (ml/g of incubated OM) correlated well with in vivo CH4 expressed per unit of eRFOM (R2 = 0.54; P=0.040), but not when expressed per unit of OM ingested (R2 = 0.04; P=0.878). For grass silage (the same trend for beet pulp), inocula adapted to R- and H-based diets compared with S- and L-based diets resulted in a lower CH4 production (36.1 vs. 44.8 ml/g of incubated OM, R vs. S; and 37.4 vs. 43.4 ml/g of incubated OM, H vs. L; P<0.001). These results indicate that adaptation of rumen inoculum to different diets affects CH4 production of a substrate differently. In conclusion, in vitro CH4 measurement can be indicative of the trend of in vivo CH4 production from different combinations of sources and levels of starch when in vivo CH4 is expressed per unit eRFOM, but not when expressed per unit OM ingested. This study suggests that complexity associated with rumen fermentation conditions needs to be considered to fully predict in vivo CH4 production from in vitro measurements.