Effects of spontaneous heating on estimates of total digestible nutrients for alfalfa-orchardgrass hays packaged in large round bales1

Large round or large square hay packages are more likely to heat spontaneously during storage than hay packaged in conventional (45 kg) bales, and the effects of this phenomenon on the associated energy estimates for these hays can be severe. Our objectives for this project were to assess the relationship between estimates of total digestible nutrients (TDN) and spontaneous heating and to describe any important differences in energy estimates that may result specifically from 2 methods of estimating truly digestible fiber (TD-Fiber). Using the summative approach to estimate TDN, TD-Fiber can be estimated from inputs of protein-corrected neutral detergent fiber (NDFn) and acid detergent lignin (TD-FiberLIG) or from NDFn and 48-h neutral detergent fiber digestibility (TD-FiberNDFD). Throughout 2006 and 2007, mixed alfalfa (Medicago sativa L.)-orchardgrass (Dactylis glomerata L.) hays from 3 individual harvests were obtained from the same 8.2-ha research site near Stratford, Wisconsin. Both options for estimating TD-Fiber (TD-FiberLIG or TD-FiberNDFD) were then used independently via the summative approach to estimate the total TDN concentrations (TDN-LIG or TDN-NDFD, respectively) within these hays. Estimates of both TDN-LIG and TDN-NDFD then were related to heating degree days >30°C accumulated during storage by various regression techniques. Changes (poststorage - prestorage) in TDN-LIG that occurred during storage (ΔTDN-LIG) were best fitted with a nonlinear decay model in which the independent variable was squared [Y = (11.7 × e−0.0000033×x×x) − 11.6; R2 = 0.928]. For changes in TDN-NDFD (ΔTDN-NDFD), a quadratic regression model provided the best fit (Y = 0.0000027x2 − 0.010x + 0.4; R2 = 0.861). Generally, ΔTDN-LIG estimates were 2.0 to 4.0 percentage units lower than ΔTDN-NDFD estimates when heating exceeded 500 HDD. For regressions on maximum internal bale temperature, both ΔTDN-LIG (Y = −0.38x + 16.3; R2 = 0.954) and ΔTDN-NDFD (Y = −0.25x + 10.2; R2 = 0.848) were best fitted by linear models with heterogeneous (P < 0.001) slopes and intercepts. In both cases, coefficients of determination were high, suggesting that simple measures of spontaneous heating are excellent predictors of energy losses in heated forages. Regardless of method, reductions in TDN were associated primarily with losses of nonfiber carbohydrate, which is known to occur via oxidation of sugars during spontaneous heating. For heated forages, some discrepancy between TDN-LIG and TDN-NDFD existed because the relationship between NDFD and spontaneous heating was shown previously to be very poor, resulting in minimal changes for estimates of TD-FiberNDFD as a consequence of heating. In contrast, TD-FiberLIG declined in close association with heating, largely because TD-FiberLIG was sensitive to changes in concentrations of both NDFn and acid detergent lignin. Discrepancies between TDN-LIG and TDN-NDFD were exacerbated further when neutral detergent fiber rather than NDFn was used to estimate TD-FiberNDFD. Estimates of TDN declined by as much as 13.0 percentage units within severely heated hays, and this is a serious consequence of spontaneous heating.