Integrating dynamic processes into waterfowl conservation prioritization tools
Data files
Dec 16, 2021 version files 3.68 MB
-
testdata.csv
574.21 KB
-
traindata.csv
3.11 MB
Sep 08, 2022 version files 2.87 MB
-
testdata.csv
574.21 KB
-
traindata.csv
2.30 MB
Abstract
Aim: Traditional approaches for including species’ distributions in conservation planning have presented them as long-term averages of variation. Like these approaches, the main waterfowl conservation targeting tool in the United States Prairie Pothole Region (US PPR) is based primarily on long-term averaged distributions of breeding pairs. While this tool has supported valuable conservation, it does not explicitly consider spatiotemporal changes in spring wetland availability and does not assess wetland availability during the brood rearing period. We sought to develop a modeling approach and targeting tool that incorporated these types of dynamics for breeding waterfowl pairs and broods. This goal also presented an opportunity for us to compare predictions from a traditional targeting tool based on long-term averages to predictions from spatiotemporal models. Such a comparison facilitated tests of the underlying assumption that this traditional targeting tool could provide an effective surrogate measure for conservation objectives such as brood abundance and climate refugia.
Location: US PPR
Methods: We developed spatiotemporal models of waterfowl pair and brood abundance within the PPR of the US. We compared the distributions predicted by these models and assessed similarity with the averaged pair data that is used to develop the current waterfowl targeting tool.
Results: Results demonstrated low similarity and correlation between the averaged pair data and spatiotemporal brood and pair models. The spatiotemporal pair model distributions served as better surrogates for brood abundance than the averaged pair data.
Main conclusions: Our study underscored the contributions that the current targeting tool has made to waterfowl conservation but also suggested that conservation plans in the region would benefit from the consideration of inter- and intra-annual dynamics. We suggested that using only the averaged pair data and derived products might result in the omission of 46-98% of important pair and brood habitat, respectively, from conservation plans.
These data were collected through three different studies:
Carrlson, K. M., Gue, C. T., Loesch, C. R., & Walker, J. A. (2018). Assessment of repeat-visit surveys as a viable method for estimating brood abundance at the 10.4-km2 scale. Wildlife Society Bulletin, 42(1), 72–77.
Kemink, K. M., Gue, C. T., Loesch, C. R., Cressey, R. L., Sieges, M. L., & Szymanski, M. L. (2019). Impacts of oil and gas development on duck brood abundance. The Journal of Wildlife Management, 83(7), 1485–1494.
Walker, J., Rotella, J. J., Schmidt, J. H., Loesch, C. R., Reynolds, R. E., Lindberg, M. S., Ringelman, J. K., & Stephens, S. E. (2013). Distribution of duck broods relative to habitat characteristics in the Prairie Pothole Region. The Journal of Wildlife Management, 77(2), 392–404.
Methods for data collection is detailed in each of these manuscripts. All data were amalgamated and processed in program R.
The traindata.csv file was used to run the spatial brood models and the testdata.csv file was used to test the model fit. Column names are described as follows, each row represents a wetland surveyed in a given year:
x/y - UTM coordinates: +proj=aea +lat_1=40.38611111111111 +lat_2=47.27722222222222 +lat_0=37 +lon_0=-105.6855555555556 +x_0=0 +y_0=0 +datum=NAD27 +units=m +no_defs +ellps=clrk66
+nadgrids=@conus,@alaska,@ntv2_0.gsb,@ntv1_can.dat
C1/C2 - AM and PM brood counts at wetland
scalePE - scaled value of basin level percent emergent cover
scale PE2 - scalePE*scalePE
LWA - scaled value of log(July basin wet area)
basinWA - scaled value of July basin wet area
scaleJWA - scaled value of the total wet area within the 10.36 sq km plot on which the wetland was located
scaleMWC - scaled value of the total number of wet basins counted in May on the 10.36 sq km plot that the wetland was located
scalePC - scaled value of the proportion of the 10.36 sq km plot that was covered in perennial vegetation.
Y09 - Y17 - year in which the count took place (Y08 was treated as the intercept)
logDate - scaled value of the log(day) within the study each year
SEAS/SEMI/TEMP - regime of the basin (Lake was treated as the intercept)
WalkIn - represented whether the survey was conducted on foot (0) or from a vehicle (1).