Using Landsat imagery to map understory shrub expansion relative to landscape position in a mid‐Appalachian watershed

Abstract

AbstractGlobally, shrub expansion is an important driver of ecological change. However, most studies of shrub expansion have focused on dryland ecosystems (e.g., savannas, rangelands, grasslands), or in tundra regions (e.g., arctic, alpine). However, shrubs play a key role in other systems, such as the understory of temperate forests. In the Appalachian forests of the eastern United States, rhododendron (Rhododendron maximum) is the most prevalent species constituent of the understory evergreen shrub community and can affect forest diversity and structure by altering light and moisture regimes, and changing soil chemical and physical properties. We examine the spatial patterns and temporal dynamics of the evergreen shrub layer within a mid‐Appalachian forest over the period from 1986 to 2011 using Landsat TM data to explore how shrub expansion is related to landscape position (e.g., elevation, aspect, and distance‐to‐stream). We use a combination of remotely sensed vegetation indices (e.g., NDVI, tasseled cap indices) derived from snow‐free, leaf‐off Landsat 5 surface reflectance data from 1986 to 2011 paired with a Random Forest classification model to examine shrub dynamics in the Weimer Run watershed near Canaan Valley, West Virginia, a first‐order, high‐altitude watershed. We show extensive increases in winter greenness we attribute to expansion of the evergreen shrub community. From 1986 to 2011, there is a 0.14 increase in winter NDVI, with a 52% relative increase in shrub coverage over 25 yr. Shrub expansion is most notable at lower elevations, along streams, and on southerly oriented slopes. We argue that changes in shrub abundance are due to decreased moisture limitations driven by changing climate. There is an increased effort to reintroduce native conifers to this region, but changes in the shrub community jeopardize this effort, as rhododendron inhibits the growth and dispersal of these conifers (e.g., red spruce). Our results show that rhododendron shrubs are expanding into areas of the forest from which they have previously been restricted. Use of these remote sensing methods may allow better habitat suitability mapping, leading to better targeted restoration efforts and more‐informed ecosystem forecasts.