Hyperspectral imagery for biodiversity mapping in a wildland-agriculture matrix
Like true islands, patches of forest in an agricultural matrix can serve as ‘ecological islands’ that may or may not follow biogeographical predictions. The study of such landscapes, sometimes called ‘countryside biogeography’ (Daily 1997), can yield critical insights about the impacts of habitat fragmentation over the short and long term. Here I used data from the Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) collected over the Kellogg Biological Station and surrounding wildland-agriculture matrix in 2008 to assess canopy tree diversity across a range of sizes of forest patches. The land that includes the WK Kellogg Biological Station has been a part of Michigan State University since the 1920s, and the surrounding landscape has been a wildland-agricultural matrix since the settlement of Michigan in the mid-1800s. I compared a number of different metrics of spectral diversity and patch size and shape to assess spatial biodiversity patterns.
I found that some forest patches do follow diversity patterns as predicted by biogeography theory, but some do not, likely due to variations in land use history and topoedaphic patterns. Importantly, many of the approaches to assessing hyperspectral diversity are highly dependent on the specific data selected, and so a portion of this talk will focus on the challenges of different approaches. Overall hyperspectral imagery, or imaging spectroscopy, has much to offer to the fields of biogeography and community ecology.