.png){kind=icon}
.png){kind=icon}
Return
Quantifying ecological indicators along field boundaries in US Midwest cropland
.png){kind=icon}
OA
AI
F
J
T
A
N
B
DOI：10.1007/s10980-026-02436-6.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Habitat fragmentation is a major ecological concern across intensively cultivated landscapes of the U.S. Midwest. Within the corn–soybean production landscape of the U.S. Midwest, natural ecosystems have largely been replaced by monoculture cropping systems and minimal contiguous tree cover, or vegetative buffer areas remain, leading to severe habitat isolation and loss of biodiversity. This study aims to (1) quantify the spatial extent and connectivity of vegetated buffer zones embedded within dominant corn–soybean agricultural systems across the U.S. Midwest using satellite-based remote sensing with deep learning classification methods, and (2) assess their ecological integrity through widely-recognized landscape metrics relevant to biodiversity and ecosystem functionality. We process Sentinel-2 NDVI time series observations and train a Long Short-Term Memory (LSTM) neural network to classify corn-soybean area under cultivation. Landscape metrics, including Effective Mesh Size and Contagion, are evaluated across ten states. The LSTM-based classification proved highly effective, achieving overall accuracy ranging from 0.88 to 0.97. We found that vegetated and natural buffer zones are unevenly distributed across the region. Natural areas embedded within corn–soybean landscapes in Iowa, Illinois, and Minnesota exhibited the lowest connectivity, here the corn–soybean landscape is characterized by dense, continuous crop cover, indicating limited ecological functionality. In contrast, Ohio, Missouri, and Indiana formed an intermediate cluster, where some natural patches persisted within the dominant agricultural matrix, reflecting moderate spatial variability. Meanwhile, Michigan, Wisconsin, Kentucky, and Tennessee displayed greater natural connectivity, resulting in a more fragmented agricultural landscape and enhanced potential for ecosystem service provision. This study provides a spatially explicit assessment of buffer zone integrity across the corn–soybean landscapes of the U.S. Midwest. The classified high-resolution layer represents a valuable resource for evaluating landscape functionality at both macro and micro scales, supporting sustainable land use and guiding targeted interventions. Dually, when combined with additional data, it can be used to explore dynamics related to management and variability in the agronomic context.
Keywords:
Crop ecology
U.S. Midwest
Remote sensing
Landscape metrics
Buffer zones
Journal
IF:
3.7
Papers:
3.8K
Citations:
1.4W
Organization
Cited Papers