Characterizing conditions of California sage scrub communities in Mediterranean-type ecosystems using remote sensing.

摘要

Biodiversity loss is an urgent global issue. For California's Mediterranean-type ecosystems, protecting biologically diverse vegetation communities such as the California sage scrub (CSS) community type is vital to conserving rare, threatened, or endangered species, as well as overall species richness of the southern and Baja California region. While existing monitoring methods such as field surveys and vegetation type mapping provide ecologically valuable information, they do not provide information about internal conditions of CSS communities. Fractional cover of plant life forms is frequently utilized to examine conditions of (semi-)arid vegetation communities. For the CSS community type, however, the utility of life-form fractional cover has not received adequate attention as an effective monitoring variable indicating ecological integrity; thus, no reliable, cost-effective methods have been developed. This dissertation investigates the effectiveness of fractional cover of true shrub, subshrub, herb, and bare ground for quantifying CSS community conditions, tests remote sensing approaches to obtain spatially comprehensive life-form cover fractions, and explores the utility of life-form fractional cover maps for sustainable, effective long-term monitoring of CSS communities of southern California.;Past studies indicate that fractional cover of plant life forms is an effective measure for quantifying CSS community integrity, and remote sensing is the only means to estimate spatially exhaustive cover fractions cost-effectively over large extent. Among the remote sensing approaches tested, object-based image analysis using pansharpened QuickBird imagery shows the most promise for estimating life-form fractional cover within CSS communities because of its high accuracy (e.g., RMSE as low as 6.4%) and robustness in estimating cover fractions and ability of providing life-form-level landscape metrics. Multiple Endmember Spectral Mixture Analysis using SPOT imagery is capable of estimating cover fractions with comparable accuracy and is beneficial for retrospective analysis for life-form cover changes and cost-effective ecological monitoring. Using spatially exhaustive life-form cover fractions, maps indicating CSS community conditions and species' life-form cover preference were generated. Such maps can fill information gaps between field-based data and vegetation type maps and provide valuable information about habitat recovery, habitat suitability, and ecological integrity of CSS communities. By combining these methods, more effective CSS community monitoring can be achieved.