|dc.description.abstract||The main objective of this study was to determine the influence of browse plant species and harvesting heights [browsable (<1.5 m) and non-browsable (>1.5 m)] on the nutritional value of browse plant leaves. Leaves samples were collected from Maytenus capitata, Olea africana, Coddia rudis, Carissa macrocarpa, Rhus refracta, Ziziphus mucronata, Boscia oliedes, Grewia robusta, Phy llanthus vessucosus and Ehretia rigida. Chapter 1 of this thesis comprises of background information, defines the statement of the problem and the justification of the study. Chapter 2 is the review of related literature on the importance of browse plant leaves, factors affecting chemical composition and the response of plant leaves to such factors. In Chapter 3, samples were analysed for proximate components, soluble phenolics, condensed tannins and minerals. The results showed that plant species had an effect (P <0.05) on fibre and mineral content of browse leaves. Plant species and interaction between plant species and harvesting height had a significant (P <0.05) influence in terms of total N, total phenolic content, while harvesting height did not influence ( P > 0 . 0 5) these substances. The leaves contained moderate to high levels of N which ranged from 81.5 g/kg N (Carissa macrocarpa) to 168.6 g/kg N (Z.
mucronata) at both heights enough to meet level required by microbes in the rumen, making them potential sources of supplementary protein during the dry seasons. Total phenol results show the interaction of tree species and harvesting height. Tree species and harvesting height affected (P<0.05) condensed tannin (CTs) content as on average the leaves of non-browsable height (0.61 AUsso nm/200 mg) was higher than that of leaves at browsable height (0.55 AUsso nm/200 mg), Chapter 4 investigated the effects of plant species and harvesting height on buffer nitrogen solubility, dry matter and nitrogen degradability of plant leaves. The results revealed that the browse plants and interaction between browse species x harvesting height had a significant impact on nitrogen solubility index of browse species. The study also showed that plant species
and harvesting height and their interaction had influence (P <0.05) on IVDMD and IVND at 12, 24 and 36 hours of incubation. At non-browsable height, B. oleoides (291.2 g/kg DM) had the highest (P <0.05) immediately degradable dry matter ' a' fractions while at browsable height R. refracta (292.4 g/kg DM) had the highest fraction 'a'. Leaves of E. rigida harvested from both non-browsable (455.8 g/kg DM) and browsable (729.5 g/kg DM) height had the highest degradable part of insoluble ' b' fraction of dry matter. The IVDMD of browse leaves was low for both plant heights, which could be due to high fibre and moderate to high levels of tannins in the leaves. The results obtained in Chapter 3 and 4 were used as reference values to calibrate and validate the NIRS
as a possible tool to predict the nutritive value of browse plant leaves. Leaves were scanned (32 scans per spectra), and spectra recorded at intervals of 2 nm using the SpectraStar XL then spectral data were recorded in diffuse reflectance and expressed as log ( 1/R). The next step was to transfer the reference values into the NIRS spectral data where they were used to generate calibration models with the aid of UCal software. Calibration models were validated using reference values and respective spectral data from an independent set of browse leaves from different areas. All chemical parameters had good calibration statistics with high R2 values (>0.8)
and low standard error of calibration ( 41.58). External validation revealed that the prediction accuracy of calibration model for total N level was high since it was able to explain 88% of the variation in this parameter in independent samples and had a small standard error of prediction (SEP) of 16.34. However, validation statistics were poor for fibre, which could be due to errors in the determination of fibre fractions during laboratory analysis. The results also show that only the OM and N calibration models generated from this study can be utilized to accurately predict these components in browse plant leaves. Thus it is concluded that NIRS can be
used to rapidly predict total N and OM content of these substrates that are frequently used as protein supplements in ruminants and other herbivores.||en_US