Climate change is anticipated to affect freshwater resources, but baseline data on the functioning of tropical watersheds is lacking, limiting efforts that seek to predict how watershed processes, water supply, and streamflow respond to anticipated changes in climate and vegetation change, and to management. To address this data gap, we applied the distributed hydrology soil vegetation model (DHSVM) across 88 watersheds spanning a highly constrained, 4500 mm mean annual rainfall (MAR) gradient on Hawai‘i Island to quantify stream flow at 3-h time-steps for eight years in response to the independent and interactive effects of (1) large observed decrease in MAR; (2) projected warming and altered precipitation; and (3) four scenarios of forest invasion by the high water-demanding non-native tree species Psidium cattleianum. The model captured 62% of variability in measured flow at daily time scales, 95% at monthly time scales, and 98% at annual time scales. We found that low DHSVM modeled flow (Q90) and storm flow (Q10) responses to observed declines in rainfall dwarfed those of projected temperature increase or invasion, with flow decline positively correlated with MAR. As a percentage of streamflow, temperature and invasion reductions were negatively correlated with MAR. By comparison, warming alone had little effect on Q90 or Q10, but both decreased with increasing P. cattleianum cover, and projected effects of declining MAR were accentuated when combined with P. cattleianum and warming. Restoration mitigated some effects of climate warming by increasing stream base flows, with the relative effects of restoration being larger in drier versus wetter watersheds. We conclude that potential changes in climate in tropical environments are likely to exert significant effects on streamflow, but managing vegetation can provide mitigating benefits.
