Improving the reliability of the two-stage in vitro assay for ruminant feed digestibility by calibration against in vivo data from a wide range of sources

Abstract

A set of 482 ruminant feed samples with digestibility determined in vivo was obtained from 21 laboratories all over the world and used to investigate the precision and reliability of the Tilley and Terry (1963) two-stage in vitro assay ( TT). Digestibility values in vivo ranged from 0.25 to 0.90 and consisted of one or more of the following digestibility parameters: dry matter digestibility ( DMD), organic matter digestibility ( OMD) and digestibility of organic matter in dry matter ( DOMD). The influence of various feed characteristics and other variables on the digestibility estimate was studied. Simple regressions of digestibility in vivo on that in vitro were calculated for 422 samples of feeds that were suitable and appeared reliable. The resultant residual standard deviations (RSD) were 0.0276, 0.0310 and 0.0218 and the coefficients of determination ( r 2) were 0.922, 0.907 and 0.931 for DMD, OMD and DOMD, respectively. Analysis of the residuals showed that 16% of the unexplained variance was due to the SE of the in vivo data. A multiple regression for the same data identified some significant variables besides TT (source of feed, experiment within source, plant species and feed type), which increased R 2 to 0.958, 0.948 and 0.968 and reduced RSD to 0.0220, 0.0259 and 0.0166 for DMD, OMD and DOMD, respectively. Other factors such as sample age (up to 21 years), intake level, treatment and in vitro run had negligible effects on the accuracy of the calibration. The study led to the following conclusions. 1. (1) The reliability of the in vitro assay is improved when the results are calibrated against in vivo samples from several laboratories. Since good reproducibility between replicates and runs of in vitro determinations are attainable in most laboratories, widely based calibration of such determinations will probably yield more reliable digestibility estimates than those obtained from most individual in vivo experiments. 2. (2) Such a widely based calibration can serve as a bench mark for checking and improving protocol for both new in vivo and in vitro determinations. 3. (3) Coefficients to account for the effects of specific feed characteristics on digestibility estimates should be derived by multifactorial analysis from in vitro values calibrated against a widely based, reliable and balanced set of in vivo values. Such analysis can also provide a more reliable basis for the construction of standard feed digestibility tables. It is suggested that the two-stage in vitro assay is an appropriate basis for a reliable international feed digestibility system that will improve current estimates world wide and especially in countries where facilities for comprehensive in vivo studies are not available.