Spatiotemporal variability and key factors of evergreen forest encroachment in the southern Great Plains.

Woody plant encroachment has been long observed in the southern Great Plains (SGP) of the United States. However, our understanding of its spatiotemporal variability, which is the basis for informed and targeted management strategy, is still poor. This study investigates the encroachment of evergreen forest, which is the most important encroachment component in the SGP. A validated evergreen forest map of the SGP (30 m resolution, for the time period 2015 to 2017) from our previous study was utilized (referred to as evergreen_base). Sample plots of evergreen forest (as of 2017) were collected across the study area, based on which a threshold of winter season (January and February) mean normalized difference vegetation index (NDVIwinter) was derived for each of the 5 sub-regions, using Landsat 7 surface reflectance data from 2015 to 2017. Then a NDVIwinter layer was created for each year within the four time periods of 1985-1989, 1995-1999, 2005-2009, and 2015-2017, with winter season surface reflectance data from Landsat 4, 5, and 7. By applying the sub-region specific NDVIwinter thresholds to the annual NDVIwinter layers and the evergreen_base, a SGP evergreen forest map was generated for each of those years. The annual evergreen forest maps within each time period were composited into one. According to the resulting four composite evergreen forest maps, mean annual encroachment rate (km2/year) was calculated at sub-region and ecoregion scales, over each of the three temporal stages 1990-1999, 2000-2009, and 2010-2017, respectively. To understand the spatiotemporal variability of the encroachment, the encroachment rate at each temporal stage was related to the corresponding initial evergreen forest area, mean annual precipitation (MAP), and mean annual burned area (MABA) through linear regression and pairwise comparison. Results suggest that most of the ecoregions have seen a slowing trend of evergreen forest encroachment since 1990. The temporal trend of encroachment rate tends to be consistent with that of MAP, but opposite to that of MABA. The spatial variability of the encroachment rate among ecoregions can be largely (>68%) explained by initial evergreen forest area but shows no significant relationship with MAP or MABA. These findings provide pertinent guidance for the combat of woody plant encroachment in the SGP under the context of climate change.