Our findings suggest that the mixed feeders and grazers were resilient to the change of habitat by maintaining the level of diet quality in both, sub-humid and semi-arid savanna ecosystems, even if those substantially differ in the vegetation structure and plant species composition. Few differences found in zebra and buffalo in the sub-humid ecosystem with higher NDF and ADF fibre content indicated a higher proportion of fresh grass biomass in grazers’ diet. Similar proportion of lignin in zebra’s diet in semi-arid and sub-humid conditions indicated that zebra did not incorporate any higher proportion of browse to its diet despite grass limited environment, similarly as reported by . Such feeding strategy put other grazing herbivores, here namely roan antelope and buffalo, under a strong competition. Higher proportions of lignin in the diet of roan, and buffalos in the sub-humid savanna suggest that they include browse in their diet as elsewhere, specifically for buffalo see , , . It seems contradictory as buffalos are primarily grazers and roan antelopes switch their diet to browsing especially during dry seasons when resources are scarce, switching from strict grazing (> 95% grass) to mixed feeding (< 50% grass) , whereas in our study, we expect there were more resources in the sub-humid savanna. Our results suggest that the vegetation structure of the sub-humid savanna that is composed by denser woody plants canopy and tall grasses (Andropogon gayanus, Schizachyrium sanguineum) and herbs in the undergrowth  does not offer for grazers adequate grass resources, i.e. appropriate leaf-stem ratio . Roan antelope are sensitive to habitat change and competition from sympatric grazers and are susceptible to nutritional deficiencies . Therefore, facing the spatial limitation, i.e. no option to migrate, and competition from other bulky feeders, i.e., zebras, the species that are physiologically able to switch from grazing to browsing on trees, shrubs and herbs , do so and maintain their diet quality.
Browsers are considered most resilient, not changing the diet, nor the use of space during drought events because trees are more resistant to drought and provide forage resources more or less continuously [3; 13]. Derby elands browse on a wide variety of woody species and are not particularly selective [33; 15], therefore we expected the animals to find adequate food resources in both environments, despite different vegetation. Their diet, however, differed in quality, specifically the animals’ diet was lower in macroelements and fibres in the semi-arid savanna in comparison to the sub-humid one. The levels of nutrients, however, did not fall under critical limits which could lead animals to nutritional deficiency. For large African herbivores, suggested critical faecal nitrogen concentration lies within the range 13–16 g/kg of dry matter (1.3–1.6 % of dry matter) and critical faecal phosphorus concentration is estimated to 2 g/kg (0.2% of dry matter) [34; 35]. Especially pronounced difference in the diet quality parameters between the two sites for browsers, i.e. for Derby eland, but also for common eland, was found in the content of lignin and its proportion as indigestible component in the diet. The sub-humid savanna offers broad-leaved tree species, for instance Terminalia spp., Combretum spp., Saba senegalensis , these species thus provide more biomass, while they also contain more lignin, tannins, and/or other secondary metabolites which simultaneously increase nitrogen content irrelevant as nutritive element due to its non-protein origin. On the other hand, the semi-arid savanna vegetation is composed dominantly by Acacia spp. with more narrow leaves, less biomass, with less antinutritive compounds. The consumption of Acacia spp. by browsing animals is indirectly confirmed by very high concentration of Ca and strongly biased Ca:P ratio, especially in Derby eland diet, as legume, including Acacia spp., and other tree and shrub species alike are rich source of Ca [37; 38]. Elands having Ca rich and P poor diet are likely unable to absorb Ca from the diet and then Ca leave the body without utilization .
There was no specific consistent pattern in the microelement concentrations in relation to feeding type, species, or savanna type. In tropical regions, soils often have limiting supplies of trace elements . This environmental background translates into plant food and as a consequence, domestic ungulates in tropical regions feeding on natural resources were reported as mineral deficient . Data from wild large savanna herbivores on microelements are virtually inexistent. In our study, we have not found any low values of microelements which would indicate any deficiency. Nevertheless, Mn concentration at all species at both localities are very high in comparison to, for instance, 4 µg/kg maintenance requirements of ewes . High concentrations of Mn are even sharply increased at roan antelopes and buffaloes in the sub-humid savanna. On the other hand, high Mn concentrations can be toxic, leading to depressed iron status and hematologic changes or damage of rumen microbial flora. Despite the fact, that Mn is one of the least toxic of the essential elements  concentrations of Mn, as well as of other microelements, and their sources from environment should be investigated further.
We must pay a close attention to the supplementary feeding and its role in the diet quality during the hot dry season when the supplement is supplied to animals. The food supplement is composed by Acacia albida (synonymous to Faidherbia albida) pods and peanut hay (Arachis hypogea). Both are legume species, rich in nitrogen, phosphorus, and calcium [44; 45; 46] and supplied to maintain the animals’ fitness during the most critical period of scarce natural resources, especially in regard to conservation program of critically endangered western subspecies of Derby eland. The supplement may change the interplay of all diet components and prevents a meaningful comparison to free ranging wildlife. For instance, faecal nitrogen and phosphorus concentrations in the animals’ diet in our study were higher at both sites and for all species in comparison to the same and related species during the dry season in Kruger national park . Similarly, the Ca:P ratio in diet of browsing species, i.e. both browsers and intermediate, was higher in the semi-arid savanna reserve where the supplementing management is more regular and intensive in comparison to the reserve in sub-humid area. The multivariate analysis in the sub-humid savanna showed the importance of ongoing season from the wet season on one side, through dry season with low explanatory value, to the hot dry on the other side, and this pattern suggests that a supplementary feeding was complementary only, without any substantial impact on natural dietary associations. Two species of mixed feeders, roan antelope and common eland, showed opposite association to nutrients and seasons, confirming thus their distinct features, thus incomparability despite their ‘mixed feeders’ designation. The position of roan antelope shows its similarity to buffalo and previous consideration of buffalo’s partial browsing. Derby eland related especially to lignin and nitrogen, and zebra appear independent on the season, each species on the opposite side. In contrast, the multivariate analysis of nutrients, species and seasons in the semi-arid savanna revealed the importance of wet and dry seasons, while the hot dry season appeared to be of low importance for the nutrients in the animal diet. We associate this to the effect of supplementary feeding which changes the relationships of nutrients in animal diets and similarities among animals. Under these conditions, common eland diet was more similar to Derby eland’s diet, and their relation to Ca suggest that they benefit from supplementary feeding, alike buffalo was more similar to zebra, all these species independent of season. Diet of roan antelope was more distinct from other species and remained on the axis of seasons.