Curriculum of Ecology, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, UNITED STATES OF AMERICA.
Species richness is the number of species in a given area or sample and is the most fundamental measure of biodiversity. It results from the aggregation of individual species whose distributions are influenced by processes operating on a wide range of scales. Estimating and understanding species richness at landscape scales (1,000-1,000,000 ha) is not easily achieved from small sample areas that can be completely inventoried. In particular the spatial structure of environments makes the richness observations across a landscape non-additive. This dissertation develops the vital links between the spatial structure of ecological factors that are hypothesized to control species richness, spatial variation in species composition, and the sampling strategies used to measure species richness.
I present a method for objectively and iteratively assessing patterns of biodiversity. This method builds upon “ecological zipcodes” that classify the landscape by energy flux, temperature, and precipitation. I also present a model of human energetic expenditure during walking that can be applied at landscape scales. I use this model to analyze sampling bias associated with accessibility for vegetation surveys. I used both the "ecological zipcodes" and the model of accessibility to design efficient and representative biodiversity samples based on clustered-stratified sampling. Finally, I assess the reliability of richness estimators that incorporate turnover in species composition.
My results illustrate that efficient and representative richness assessment is possible, even with little a priori knowledge about the spatial structure of species richness. They also demonstrate that typical biodiversity assessments show a strong bias in accessibility that is both a product of the spatial structuring of samples as well as environment. This bias is significant even for small biases in sample accessibility. Also, I show that though clustered sampling designs capture multiple scales of aggregation, their representativeness is very sensitive to stratification. Finally, my results show that species richness estimates that incorporate turnover are confounded by the interaction between sample size and environmental heterogeneity. Only when controlling for these effects, can information about the spatial turnover in species composition be effective in estimating species richness.