Scalogram habitat measures as predictors of bird abundance

Abstract

Birds select habitat characteristics, such as variability in habitat structure, across multiple spatial scales (grain and extent). Measuring habitat variability at multiple scales can better capture factors that influence avifauna communities than focusing on one scale only. One valuable tool in assessing habitat heterogeneity is the cumulative dynamic habitat index (DHI), which is derived from satellite data and captures temporal variability in vegetation productivity. Our goals were to develop new habitat measures from the cumulative DHI at multiple scales based on scalograms, and to test their performance in models of bird abundance. We counted birds at 188 plots during three breeding seasons (2007–2009) at Fort McCoy military installation, USA, to assess the abundance of forest (ovenbird), shrubland (indigo bunting), and grassland (grasshopper sparrow) bird specialists. We then calculated NDVI based on PlanetScope (3 m), Sentinel‐2 (10 m), Landsat‐8 (30 m), and MODIS (250 m) data to quantify cumulative DHI. We summarized the averaged NDVI cumulative DHI within multiple extents around each bird survey and developed 11 new habitat measures to test their predictive power in models of bird abundance. We found positive relationships between cumulative DHI at different extents and the abundances of both ovenbirds and indigo buntings, a forest and a shrubland specialist, respectively; and a negative relationship with grasshopper sparrows, a grassland specialist. In multiple linear regression models that incorporated single‐ and multi‐grain predictors, the scalogram habitat measures explained moderate to high levels of variability in bird abundance, with R2 = 0.77, 0.37, and 0.75 for our forest, shrubland, and grassland specialists, respectively. Our results show that scalograms are an effective tool for capturing multiscale habitat configuration, because they capture the variability of habitat conditions in forests, shrublands, and grasslands. The scalogram habitat measures that we developed can be computed using our new R package ‘scalogram'.