Plant community recovery from intense deer grazing depends on reduction of graminoids and the time after exclosure installation in a semi-natural grassland

Exclosures that exclude large herbivores are effective tools for the protection and restoration of grazed plant communities. However, previous studies have shown that the installation of an exclosure does not ensure plant community recovery. Our study aimed to determine the effects of the domination of unpalatable plants and the timing of exclosure installation on the plant community recovery process in montane grassland overgrazed by sika deer (Cervus nippon) in Japan. In this study we compared plant species composition and their cover with inside and outside exclosures installed at different times. Furthermore, we also compared them with those in 1981, when density of sika deer was very low. We used quadrats inside and outside fenced areas established in 2010 and 2011 to record both the cover and the height of species in each quadrat between 2011 and 2015. Plant cover, with the exception of graminoid species, increased in later years in all treatments. Non-metric multidimensional scaling (NMDS) plots showed significantly differentiated treatment trends. The species composition within the 2010 fenced area gradually shifted to greater similarity with the species composition reported in 1981. The plant community in the 2011 fenced area was slower to recover. Compositions of plant communities outside the fenced areas hardly changed from 2011 to 2015. Chao’s dissimilarity index decreased over time between the plant community surveyed between 2011 and 2015 and the past plant community in 1981 within the exclosures, and was higher in the 2011 fenced area than in the 2010 fenced area. In conclusion, we show that the reduction of graminoids and the time after exclosure installation were important for plant community recovery from deer grazing damage. A delay in exclosure installation of one year could result in a delay in plant community recovery of more than one year.