Abstract:

The potential of NIRS fecal profiling to predict crude protein (CP) and
digestible organic matter (DOM) in the diet of free-ranging sheep and cattle,
and condensed tannins in the diet of sheep, in sub-Saharan Africa (SSA), was
examined. The calibration sets comprised of samples from the semi-arid,
humid, and highland zones of SSA. USA samples were incorporated into the
cattle calibration. Validation samples were independently derived. Cluster
analysis was conducted to aid interpretation of results.
Six sheep CP and DOM equations were created according to the
combination of samples (“All locations”, “Ethiopia", “Ethiopia' and Nigeria”,
“Niger”, “Extrusa” (CP) or "In vitro" (DOM), and “Stall” (CP) or "In vivo" (DOM).
Calibration statistics were satisfactory, (SEC within twice the SEL, and
R2>0.80), except for the “Niger” CP equation (R2 0.70). Selected wavelengths
>0.80, (R2 0.88), indicating the potential for development of a robust CP sheep
equation. DOM validation results were poor; the “Ethiopia and Nigeria”
were biologically significant. Only the “All locations” CP validation had a R2 equation performed best (R2 0.40). Spectral diversity indicated between the
Niger and validation samples was confirmed by cluster analysis.
0.91). Selected wavelengths were highly indicative of tannin-protein
complexes. Validation statistics were less satisfactory: a high SEP 57.79,
indicated laboratory technique differences, a R2 of 0.62 was below the
required R2 0.80, but was indicative of the potential of the methodology to
predict tannin in the diet of sheep. Spectral diversity was confirmed by a
cluster analysis.
Two equations were formulated for the cattle CP and DOM calibration:
“Africa”, and “USA and Africa”. Both CP and DOM calibration statistics were
deemed satisfactory. The CP validation statistics were similar to each other
though not of predictive capacity. The DOM validation statistics were poor;
SEP and R2 (“Africa” 12.27 and 0.09) and (“USA and Africa” 11.49 and 0.06),
respectively. T outliers indicated differences in determination techniques.
The findings substantiate the potential of NIRS fecal profiling as a
methodology to predict the diet of free-ranging livestock across SSA. More
diverse calibration sets, and standardization of laboratory techniques is
required.
The sheep tannin calibration statistics were satisfactory (SEC 9.02, R2