SOURCES OF ERROR WITH INVITRO DIGESTIBILITY ASSAY OF PASTURE FEEDS

A series of experiments was undertaken to determine population statistics for in vitro organic matter digestibility (in vitro OMD) data and to examine the effects of basal diet, donor animal and precollection fasting interval on the activity and specificity of rumen fluid inoculum. The experiments utilized wether sheep, a diverse set of pasture grass and legume feeds prominent in the Australian subtropics and the Tilley and Terry in vitro digestibility procedure running under the operating pressure of a practising feeds evaluation laboratory. The standard errors of in vitro OMD estimates for within and between batch runs were +/- 0.88 x 10(-2) and 0.62 x 10(-2), respectively. These error terms were used to develop protocols to accept, reject or scale raw in vitro OMD data. Differences between donor animals in the activity of rumen fluid were highly significant. Extending the precollection fasting interval beyond 16 h was associated with a substantial decline in inoculum activity. An in vitro-in vivo calibration relationship based on fifteen test feeds and using lucerne (Medicago sativa) as basal diet was described by the linear model y = 1.3 x -0.195 +/- 4.9 x 10(-2), r = 0.79 (y = in vivo OMD, x = in vitro OMD). Despite large effects of basal diet on both the absolute values and relative ranking of test feeds, neither the RSD nor r values were improved using alternative diets to lucerne chaff. The results highlight the need to formally standardize the analytical and biological components of the in vitro digestibility procedure to safeguard the integrity of data.