Correlative ecological niche model applications to predicting landscape-scale woody plant encroachment in Kansas tallgrass prairie systems

Abstract

Woody plant encroachment (WPE) in grassland ecosystems has been a pervasive process across the Great Plains, yet a predictive understanding of where it will occur has been elusive. As an exploration of tools of potential utility in this challenge, we mapped WPE processes over the years 2015-2021 in a set of 9 counties in central Kansas. We developed and tested two correlative models based on landscape features: one that assessed distribution of evergreen trees in 2015, and another that assessed areas of WPE in succeeding years. Both models were successful, being able to predict 2015 forest distributions and being able to predict WPE during 2015-2021, as functions of characteristics of landscapes. These simple, correlative models will certainly not be able to predict WPE processes globally, or even regionally, but provide first proof-of-concept explorations for the central Great Plains region.

Fig 1. Summary of the flow of information in the analyses reported in this paper.

Raster data layers that supply occurrence information (for forest presence or for woody plant encroachment) are shown in green, whereas data layers used as environmental variables are shown in purple. Gray shapes indicate phases of model testing.

Fig 2. Map showing the study area, and the areas over which 249 models were calibrated and evaluated, in central Kansas.

Models were calibrated across Wabaunsee County, in the center of the map, and evaluated in the 8 surrounding counties (outlined in bold black lines. The inset (blue outline) shows a detailed view of a small part of southwestern Wabaunsee County: Yellow areas were forested in 2015, whereas green areas represent areas that became forested between 2015 and 2021.

Fig 3. Summary of calibration phase of development of models of 2015 evergreen forest extent across Wabaunsee County.

(top left) Distribution of evergreen forest in 2015 and woody plant encroachment during 2015–2021, and (top right) model output (light green = low suitability, dark green = high suitability). The bottom two panels show the two environmental variables that were included in the “best” models that were detected and explored in this study, each shown with lighter colors indicating higher values.

Fig 4. Test of the predictions of 2015 evergreen forest extent in the 8-county test area in central Kansas.

Model predictions range from low suitability (white) to high suitability (dark green); actual 2015 evergreen forest extent in the region is indicated by black specks. Small gray areas within the 8-county test area are those in which the model transfers from the calibration region to the test area were extrapolative, and so no predictions were made.

Fig 5. Summary of model results for woody plant encroachment (WPE) in terms of increased extent of evergreen forest in central Kansas between 2015 and 2021.

Top panel shows full extent of the study area, with 2015 evergreen forest extent shown in green, and WPE during 2015–2021 shown in black; both are plotted on top of the model predictions (light blue = unsuitable, dark blue = highly suitable; light gray = no predictions made under extrapolative conditions), which were calibrated in Wabaunsee County (center of map) and evaluated across the 8 surrounding counties. The bottom two panels show “zooms” to two areas in the evaluation area, such that predictions and WPE extent (in black) are independent.

Fig 6. Comparison of old and new views on Ft. Riley, Geary County, Kansas (Kansas Rephotography ID number 2226).

Top photo shows a view from 1899, showing military maneuvers on Pawnee Flats, on Ft. Riley (photo courtesy of the University of Kansas Libraries). Bottom photo was taken from 39.0901°N, 96.7538°W in 2023. Note the original restriction of forested areas to the valley, but the extension of the treed areas in the valleys and onto some of the higher areas in the recent photo. Top portion of figure modified from an image provided at https://digital.lib.ku.edu/ku-pennell/842, under a CC BY license, with permission from University of Kansas, original copyright 1899.