The protective effect of dwarf bamboo on broad-leaved seedlings against deer browsing

Abstract

The forest understory is often densely covered by dwarf bamboo Sasa spp. in Hokkaido, northern Japan. Dwarf bamboo is an important winter forage for sika deer Cervus nippon, and dwarf bamboo inhibits forest regeneration by preventing seedling establishment. Although percentages of browsed seedlings are a useful index of deer browsing pressure, dwarf bamboo is also expected to affect browsing on seedlings by altering physical accessibility to seedlings. We hypothesized that the relationship between seedling height and browsing incidence depends on the coverage of dwarf bamboo because smaller seedlings were protected by dwarf bamboo. The incidence of browsing was recorded for broad-leaved seedlings ≥ 0.2 m in height and < 1 cm in diameter at breast height in 19 plots (40–200 m2 each) located in natural forests. Coverage of dwarf bamboo varied from 3% to 74%, except for one plot where dwarf bamboo was absent. The mean height of the dwarf bamboo in each plot was 0.44–1.17 m. The product of seedling height and coverage of dwarf bamboo was selected as an explanatory variable in the generalized linear mixed model for the incidence of browsing on each seedling. This suggests that smaller seedlings were protected by dwarf bamboo and browsing incidence is low for smaller seedlings within plots with high coverage of dwarf bamboo. This can be an example of associational defense against deer. The probability of browsing incidence was low for seedlings 0.2–0.5 m tall in sites with dense dwarf bamboo, while no significant differences were detected between seedlings 0.5–0.8 m tall and those ≥ 0.8 m. Even though seedlings under dwarf bamboo cover tend to escape from deer browsing, dwarf bamboo directly affects seedling survival and growth negatively. Therefore, it is necessary to manage both the deer population and dwarf bamboo to enable forest regeneration. The percentages of seedlings with browsing incidence on seedlings ≥ 0.5 m tall can approximate the browsing level and is useful for management decisions.