Indigestible neutral detergent fiber (iNDF) is routinely used as an internal marker to estimate dry matter digestibility (DMD) of feeds in ruminants but has been applied less frequently to horses. Measurement of iNDF requires prolonged incubation of samples in microbial inoculum. Despite inherent differences in microbial fiber degradation between species, we hypothesized that bovine and equine sources of inoculum would produce similar measures of iNDF and both would yield accurate estimates of DMD in horses. The aim of this study was to directly measure fecal output and DMD in horses using total fecal collection and compare it to estimates derived from iNDF that had been quantified by different methods. Five mature Quarter Horse geldings (552 ± 14 kg) were fed 5 hays differing in fiber composition in a 5 × 5 Latin square design experiment. Hays offered at 1.6% BW included: alfalfa (Medicago sativa, ALFA), orchardgrass (Dactylis glomerata, ORCH), and bermudagrass (Cynodon dactylon) harvested at 4 (BG-4), 6 (BG-6), and 8 wk of regrowth (BG-8). Each period consisted of an 8-d diet adaptation followed by a 3-d total fecal collection (TC). Hay and fecal samples were dried and ground, placed in 25-μm filter bags, and incubated for 288 h in 4 different inocula to determine iNDF: in situ in rumen-cannulated cows, in vitro in rumen fluid (RF), in vitro in cecal fluid (CF), and in vitro in equine fecal inoculum (EF). Following incubation samples were analyzed for residual neutraldetergent fiber (ie, iNDF). Data were compared using a split plot within a Latin square whole plot design using mixed model ANOVA and linear regression. Mean iNDF recovery for in situ and EF (95%, 91%) was greater (P < 0.05) than CF and RF (88%, 85%; SEM 1.6%). Fecal output estimated by in situ and EF did not differ from that measured by TC but was overestimated by RF and CF methods (P < 0.05). An interaction of hay*method was observed for DMD (P = 0.01). For ALFA and BG-6, DMD determined by all iNDF methods aligned with DMD measured by TC. For ORCH and BG-4, only DMD estimates from in situ and EF were similar to TC. All iNDF methods underestimated DMD for BG-8. Linear regression of DMD for all hays measured by TC showed a stronger relationship with in situ and EF iNDF methods (R2 = 0.65, 0.64; RMSE 7.4, 7.2) than RF and CF iNDF methods (R2 = 0.56, 0.61; RMSE 8.6, 7.7), though all were significant (P < 0.01) and all equations approached unity (slope 1.0, intercept 0). Results show iNDF can be used in horses to estimate DMD across a variety of hays and may align better with TC when rumen in situ and EF methods are used to determine iNDF.
Read full abstract