Comparative 1H-NMR profiling of plasma and serum metabolites in giraffes and white rhinoceroses

Background Blood metabolite profiling can provide descriptive information on physiological variation in wildlife, but interpretation is sensitive to sampling design, specimen matrix and analytical batch. Objectives To characterise quantified blood metabolites in apparently healthy giraffes and white rhinoceroses, examine sex-associated differences within each species, and conduct an exploratory matrix-matched interspecies comparison. Methods Proton nuclear magnetic resonance spectra were available for 24 valid giraffe sampling occasions from 19 animals and for 45 white rhinoceroses. Metabolites were quantified jointly from the original spectra. Design-defined comparisons used log₂-transformed concentrations, Welch tests, Benjamini–Hochberg false-discovery correction, effect sizes and permutation, rank-based, leave-one-out and annotation-confidence sensitivity analyses. Results No sex-associated metabolite differences were detected among the 2017 Rooipoort giraffe serum samples. Nine metabolites met the primary false-discovery criterion in the rhinoceros sex comparison, although only N-acetylglucosamine showed consistently strong evidence across all sensitivity analyses. In the comparison of six verified giraffe and 45 rhinoceros EDTA-plasma samples, 36 of 50 metabolites differed by the primary analysis; 28 formed a conservative core supported by all statistical tests and all leave-one-giraffe-out analyses. The core remained significant after adjustment for sex. Collection year was completely confounded with serum versus plasma in giraffes and was not tested inferentially. Conclusions The two study groups had markedly different measured plasma metabolite profiles, with a coherent difference among branched-chain amino-acid-related metabolites. However, species was inseparable from NMR acquisition batch, and the small giraffe group limits biological and pathway-level inference. The concentrations provide exploratory baseline data rather than formal reference intervals.