Quantitative assessment of representativeness of the long-term agroecological research (LTAR) network of sites

Authors: KUMAR, JITENDRA - OAK RIDGE NATIONAL LABORATORY; HARGROVE, WILLIAM - U.S. FOREST SERVICE (FS); Coffin, Alisa; Baffaut, Claire; PONCE-CAMPOS, GUILLERMO - UNIVERSITY OF ARIZONA; Witthaus, Lindsey

Submitted to: US-International Association for Landscape Ecology
Publication Type: Abstract Only
Publication Acceptance Date: 2/21/2022
Publication Date: 4/13/2022
Citation: Kumar, J., Hargrove, W.W., Coffin, A.W., Baffaut, C., Ponce-Campos, G., Witthaus, L.M. 2022. Quantitative assessment of representativeness of the long-term agroecological research (LTAR) network of sites. US-International Association for Landscape Ecology. virtual. 2022.

Interpretive Summary:

Technical Abstract: The performance of any network of experimental sites depends on how well the environmental conditions at those sites reflect or represent the conditions occurring throughout the greater area that the network is intended to represent. Many experimental networks, including the Long-Term Agricultural Research (LTAR) network, have often added sites in an undirected, ad hoc way, through a process of simple organic growth. Using methods from previous regionalization efforts (e.g. NEON), we quantified and statistically analyzed the representativeness of existing sites within the LTAR network to identify regions that are well-, or poorly-represented. We also derived spatial constituency maps that delineates the regions represented the best by each LTAR site from the current constellation of LTAR sites. LTAR representativeness analyses were conducted across North America at 1km spatial resolution, with multivariate environmental conditions characterized by 15 climatic, edaphic and topographic variables. Such quantitative network analysis provides insight for both LTAR network managers and LTAR researchers. Network managers can use analysis to assess how well the distributed network of sites sample the conditions nationally, quantify regions with poor representation, and provide statistically guided recruitment of new sites to the network to optimize the overall network representativeness. Analysis can also be used by researchers to upscale the point observations to larger regions over which the observations are likely to remain valid. Our analysis shows that the current network of LTAR sites provide a fair representation of croplands and grazingland sites nationally. Cropland sites show a relatively low variability and higher mean representativeness within their constituencies, while grazinglands show a higher variability and lower mean representativeness scores. We also quantified the complementary benefits gained by the LTAR network by leveraging observations collected at NEON and LTER sites in croplands and grazinglands. NEON sites in particular can improve the LTAR’s representation of grazinglands.