The response of native species to removal of invasive exotic grasses in a seasonally dry Hawaiian woodland

Non-native perennial grasses form 30% of the live understory biomass in seasonally dry, submontane forests in Hawaii Volcanoes National Park yet their effects on native species are unknown. We removed these grasses from plots of 20 mx 20 m in 1991 and maintained removal and control areas over the next three years. Two fast growing shrub species, Dodonaea viscosa and Osteomeles anthylidifolia, increased in size significantly more in removal areas than in controls. Individuals of the most abundant shrub species, Styphelia tameiameia showed no net growth response to grass removal. They did, however, change their architecture: many branches along the mid and upper sections of the main trunk died and a proliferation of new leaves and shoots occurred in the lower 40 cm of trunk. Basal diameter increase was very small in Metrosideros polymorpha, the dominant tree species in these sites. All species except Styphelia had significantly increased leaf tissue nitrogen in removal plots by 18 months after removal when compared to shrubs in control areas suggesting that removal plot shrubs had greater access to soil nitrogen. Available soil-N pools, which were generally higher in the removal plots, support this interpretation. Light levels near the soil surface were also higher where grasses were removed than where they were present which may have contributed to increased shrub growth. By contrast, soil moisture was consistently lower where grasses were removed than where they were still present. Shrub tissue carbon isotope values were consistent with the interpretation that shrubs in removal plots had less rather than more water available to them. Hence, the increased growth observed in removal plot shrubs could not be due to release from moisture competition. Lastly, our results showed that seedlings of all woody species except Metrosideros were significantly more abundant in removal plots at both one and three years after removal and initially high sapling mortality was balanced by high recruitment into the sapling class. We believe that over time this will result in increased densities of native shrubs if grasses are kept out. With the presence of grasses, shrub growth in these woodlands is reduced and biomass is shifting towards grasses.