Climate change causes declines and greater extremes in wetland inundation in a region important for wetland birds
Data files
Sep 29, 2023 version files 232.87 MB
-
README.md
3.79 KB
-
wetland_cov_dataframe_fall.csv
120.65 MB
-
wetland_cov_dataframe_spring.csv
112.22 MB
Abstract
Wetland ecosystems are vital for maintaining global biodiversity, as they provide important stopover sites for many species of migrating wetland-associated birds. However, because weather determines their hydrologic cycles, wetlands are highly vulnerable to the effects of climate change. Although changes in temperature and precipitation resulting from climate change are expected to reduce inundation of wetlands, few efforts have been made to quantify how these changes will influence availability of stopover sites for migratory wetland birds. Additionally, few studies have evaluated how climate change will influence inter-annual variability or the frequency of extremes in wetland availability. For spring and fall bird migration in 7 ecoregions in the south-central Great Plains of North America, we developed predictive models associating abundance of inundated wetlands with a suite of weather and landcover variables. We then used these models to generate predictions of wetland inundation at the end of the century (2069–2099) under future climate change scenarios. Climate models predicted the average number of inundated wetlands will likely decline during both spring and fall migration periods, with declines being greatest in the eastern ecoregions of the Great Plains. However, the magnitude of predicted declines varied considerably across climate models and ecoregions, with uncertainty among climate models being greatest in the High Plains ecoregion. Most ecoregions also were predicted to experience more-frequent extremely dry years (i.e., years with extremely low wetland abundances), but the projected change in inter-annual variability of wetland inundation was relatively small and varied across ecoregions and seasons. Because the south-central Great Plains represents an important link along the migratory routes of many wetland-dependent avian species, future declines in wetland inundation and more frequent periods of only a few wetlands being inundated. resulting in an uncertain future for migratory birds as they experience reduced availability of wetland stopover habitat across their migration pathways.
README: Climate change causes declines and greater extremes in wetland inundation in a region important for wetland birds.
https://doi.org/10.5061/dryad.wwpzgmsqv
We estimated the number of inundated wetlands in the southern great plains for each year from 1991-2020 using the USGS Dynamic Surface Water Extent Product. The DWSE is a free remote sensing data product that is available online through the USGS. Number of inundated wetlands we estimated for two time periods relevent to migratory birds in each year the spring migration as 1 March - 31 May and fall migration as 1 September - 31 November. Wetlanbds were determined to be inundated at in a given season if they were observed to have water in at least 2 landsat/dswe images.
Data was summarised into 10 x 10km grid covering the entire southern great plains in addition to the number of inundated wetlands per grid cell in each year we also recorded a number of variables related to weather and land cover. The weather variables include: total precipitation (mm), mean daily average temperature (oC), precipitation balence (measured over 3 time scales 3-months, 6 months and 12 months prior to migration), and mean daily maximum temperature. Land cover included % cover of crops, grasslands, trees, and development as well as Terrain roughness index and the number of wetland basins. A full description of how each of these variables were derived can be found in the full text of the associated manuscript.
Description of the data and file structure
Included are two Excel files, one for each season with all variables needed to recreate our random forest data-analysis. Each row is associated with a 10x10 km grid cell/ year combination (each cell should occur 30 times, once for each year). Cell ID is recorded as FID_sample in the spreadsheet and the year is recorded in the year or y1 columns. The Latidude and Longitude of the centroid of each grid cell is recorded in the Lat, Long columns respectively. The number of inundated wetlands in each cell per year is recorded in the wet.num column, rows with NA indicate rows cells where no wetlands occur. Seasonal precipitation, mean temperature and maximum temperature and represented by the variables season*.ppt, season*.tmean, season.tmax where season* corresponds to fall, spring, summer, or winter in the excel sheet. precipitation balance is represented by bal3 (3-month precipitation balance), bal6 (6-month precipitation balance), and bal12 (12-month precipitation balance). Precipitation is in millimeters, temperature is in degrees Celsius and precipitation balance (bal3, bal6, bal12) is a unitless metric that represents the unstandardized difference between precipitation total and evapotranspiration potential. The TRI represents terrain roughness index which was calculated as the average difference between neighboring pixels in a 30 m resolution Digital Elevation Map of the southern Great Plains. The crop10, dev, tree, grass represent the percent cover of croplands, developed areas, forests, and grasslands respectively. All percent cover variables were derived from the USDA NASS cropland dataset. All analyses were conducted at the ecoregion level, with each grid cell's ecoregion being denoted in the ecoregion column. The Area column denotes the area of each grid cell in kilometers squared.
Sharing/Access information
This is a section for linking to other ways to access the data, and for linking to sources the data is derived from, if any.
DSWE data can be downloaded from
https://earthexplorer.usgs.gov/
Future climate data is not included here due to file size, but can be accessed at
https://gdo-dcp.ucllnl.org/downscaled_cmip_projections/dcpInterface.html
Code/Software
Methods
We estimated the number of inundated wetlands in the southern Great Plains for each year from 1991–2020 using the USGS Dynamic Surface Water Extent Product. The DWSE is a free remote sensing data product that is available online through the USGS. Number of inundated wetlands was estimated for two time periods relevant to migratory birds in each year the spring migration as 1 March–31 May and fall migration as 1 September–31 November. Wetlands were determined to be inundated at in a given season if they were observed to have water in at least 2 Landsat/DSWE images.
Data was summarised into 10 x 10km grid covering the entire southern great plains in addition to the number of inundated wetlands per grid cell in each year we also recorded a number of variables related to weather and land cover. The weather variables include: total precipitation (mm), mean daily average temperature (°C), precipitation balance (measured over 3 time scales: 3 months, 6 months and 12 months prior to migration), and mean daily maximum temperature. Land cover included % cover of crops, grasslands, trees, and development as well as terrain roughness index and the number of wetland basins. A full description of how each of these variables was derived can be found in the full text of the associated manuscript.