Demographic modeling of Hawaiian silverswords, and its implications for conservation

Abstract

Population viability analyses based on matrix population models provide important information for species management. These analyses enable biologists to predict future population size and structure, assess extinction risk, and identify the stages and transitions in a species' life history that have the largest effects on population growth rate (lambda). It is also possible, using these analyses, to weigh the relative effectiveness of different management strategies. In this study, I constructed a matrix population model for a threatened Hawaiian plant species, the Haleakala silversword (Argyroxiphium sandwicense ssp. macrocephalum), in order to assess the viability of this population under different disturbance and management regimes. Over five years, I assessed annual variation in seed set and quantified reproductive vital rates for use in a matrix model. These data were combined with long-term demographic data in order to construct a stage-based matrix model for the Haleakala silversword. Using this model, I examined temporal and spatial variation in silversword demography. I also used the model to evaluate management strategies for the Haleakala silversword and the related subspecies, the endangered Mauna Kea silversword (Argyroxiphium sandwicense ssp. sandwicense). Using deterministic and stochastic models, I compared the relative impacts of different threats on silversword persistence, and weighed the relative effectiveness of different management options. The Haleakala silversword was self-incompatible, and percent seed set was positively correlated with annual flowering plant abundance. Individuals were pollen-limited only in low flowering years, suggesting a pollinator-mediated Allee effect in this species. Seeds remained viable in the field for multiple years, suggesting that a seed stage should be incorporated into the matrix model. Reproductive output (e.g., number of capitula and capitulum size) was strongly correlated with rosette diameter, allowing for size-based estimates of reproduction. Vital rates and population growth rates varied over time and space. In all years and plots, adult survival had the greatest impact on lambda. Thus, factors influencing adult survival, such as browsing and outplanting, had larger effects on lambda than did factors influencing seed set. Management strategies aimed at increasing germination rates or adult survival rates will be most effective in ensuring the persistence of silversword populations.