Abstract: To establish a method to use Oxytropis glabra as a silage crop, the influence of different ratios of O. glabra and corn (Zea mays) on microbial characteristics and the diversity of lactic acid bacteria in silage were investigated. We also screened for swainsonine-removing lactic acid bacterial strains. The proportions of O. glabra:corn in the eight treatments were as follows: 10:0 (T₁), 9:1 (T₂), 8:2 (T₃), 7:3 (T₄), 6:4 (T₅), 5:5 (T₆), 4:6 (T₇) and 0:10 (T₈). The microbial community structure and quantities were analyzed after fermentation of silage for 60 days. Lactic acid bacterial strains were isolated from fresh material and silage, and their removal rates of swainsonine were determined. The main results were as follows: 1) The numbers of lactic acid bacteria in fresh material increased with increasing proportions of corn, but did not differ significantly among the silage materials. The mixing ratio did not affect the numbers of coliform bacteria in the fresh material, and coliform bacteria could not be detected after fermentation. The numbers of aerobic bacteria in fresh and silage material decreased with increasing proportions of corn. The amount of yeasts was not affected by the ratio of O. glabra and corn. When more corn was added, the number of molds increased, but molds were beneath the limits of detection in the silage materials. 2) Four lactic acid bacterial strains were isolated from the fresh material: Lactococcus lactis subsp. Hordniae, Lactococcus lactis subsp. Lactis, and Lactococcus taiwanensis. Nine lactic acid bacterial strains were isolated from silage material, including Lactobacillus plantarum subsp. Plantarum, Lactobacillus brevis, Lactobacillus pentosus, and Lactobacillus amylovorus. As the proportion of corn increased, the diversity of lactic acid bacteria increased in fresh material but decreased in silage. 3) The swainsonine-removing rates of all lactic acid bacterial strains were higher than 85%, and 100% for strains JD6E, JD4D, JD10D, JD2E, and JD1F. After pretreatment of silage with boiling water, the swainsonine-removing rates of strains JD10D and JD2E were 100% and 97.76% respectively, but those of other strains decreased by 31.76%-100%. The swainsonine-removing rates of strains JD6D and JD1E dropped to 2.17% and 0, respectively, after the heat treatment. These results showed that with increasing proportions of corn in the mixture, the quantities and diversity of lactic acid bacteria in fresh material increased and the populations of aerobic bacteria both in raw material and silage decreased, which increased the success rate of ensilage. Some lactic acid bacteria were able to remove swainsonine. Strains JD10D and JD2E had the best ability to bind swainsonine, and strains JD6D and JD1E showed the best swainsonine-degradation ability. These four strains can be used as biological detoxification agents to remove swainsonine from O. glabra silage.

Key words: Oxytropis glabra, mixed silage, microbial characteristics, swainsonine-removing