4.1 | Diet characterization
The determination of diet composition of large grazing herbivores has been a difficult problem due to methodological limitations and habitat complexity. In comparison with traditional diet analysis of herbivore fecal samples, metabarcoding approaches provide a fast and high-resolution identification to species level (Vesterinen et al., 2016). Both frequency of occurrence (FOO) and relative read abundance (RAA) were applied to obtain semi-quantitative diet data. Overall, our survey of sheep feces revealed 56 plant taxa (at genus level or species level) (Figure 2). Both approaches have shown accurate estimates when the number of food taxa in samples is small (Mcinnes et al., 2017). However, there are taxa biases in recovery of sequences. FOO tends to overestimate the importance of food groups with small amounts. RRA can reflect more accurate relative proportions of the population-level diet.
Metabarcoding dietary analysis allows quantification of variation in animal diets due to changes in usability and taxa composition (Table 3 and Figure 3). The herbage taxa, such as Lespedeza sp .,Caragana korshinskii , Oxytropis sp. , Salsola sp. ,Bassia sp. , Setaria italica , and Cenchrus sp. , in the diets of Tan sheep as characterized by metabarcoding data were broadly consistent with our knowledge base for the diets of Tan sheep in this area. The high diversity of food taxa observed in sheep diets can, to a large extent, be attributed to the capability of metabarcoding to quickly detect plant items derived from food residues that are often hard to identify by other conventional techniques. Sequencing data based on RRA metrics demonstrated that lambs presented variant diets as a response to grazing type or grazing period, with approximatively all food taxa assigned to the species level.
The selection of dietary components is affected by many factors such as habitat characteristics and availability of plant resources (biomass, nutrients, or palatability), but traditional techniques can only determine at most up to 8–12 and thus might not provide a comprehensive evaluation of diet composition in the natural grassland (Zhang, Jin, Badgery, & Tana, 2017). The observations herein showed that there were significant differences in herbage composition among the grazing patterns that resulted from the distinction of feeding regimes. The energy and albumen contents of concentrated pelleted feed in the time-limited grazing group were greater than in the forage roughage in the grazing group, and this influenced the plant groups consumed by sheep in the concentrate dominant group. Compared to the time-limited grazing group (grazing for 4 h), sheep grazing for 12 h were able to acquire more diverse plant species rather than relying primarily on Lespedeza sp. , and this can be attributed to their having sufficient time for choice. Regarding to the supplementary feeding group, Lespedeza sp. seemed to prevail due to its ubiquity as well as satisfactory palatability. In comparison, the grazing time in the time-limited grazing group was less than in the all-day continuous grazing group, resulting in less selectivity for plant species and higher transformation efficiency for forage.
It has been reported that the feedback for feeding behavior of sheep can be associated with the sensory quality traits of herbage taxa (Provenza, 1995). In the present study, the findings on the diet composition for the two periods showed a significant difference. The diet intake of sheep in the early grazing mainly consisted of Lespedeza sp. , but this changed to Chenopodium sp. , Artemisia, sp. , andLespedeza sp. in the late grazing group. Tan sheep have been shown to forage largely onLespedeza sp. , a common plant item in local desert steppe. Furthermore, metabarcoding diet analysis also demonstrated that Tan sheep indeed consume medicinal plants such as Phellodendron amurense , Arnebia euchroma ,Ixeris tamagawaensis , and Convolvulus sp. , consistent with other studies.