Agroecoregions resulting from novel clustering methods: Human dimensions variables

Authors: Hurst, Zachary; Archer, David; Coffin, Alisa; FRIEDRICHSEN, CLAIRE - University Of Idaho; Van Huysen, Tiffany; Goslee, Sarah; Pisarello, Kathryn; WULFHORST, J. - University Of Idaho

Submitted to: US-International Association for Landscape Ecology
Publication Type: Abstract Only
Publication Acceptance Date: 1/1/2022
Publication Date: 4/14/2022
Citation: Hurst, Z.M., Archer, D.W., Coffin, A.W., Friedrichsen, C., Van Huysen, T.L., Goslee, S.C., Pisarello, K., Wulfhorst, J.D. 2022. Agroecoregions resulting from novel clustering methods: Human dimensions variables. US-International Association for Landscape Ecology. Abstract.

Interpretive Summary:

Technical Abstract: Identifying agroecoregions defined by  variables related to human dimensions  is useful for applied agricultural research and understanding how communities may interact with working landscapes.  We pinpointed human dimension variables to be clustered and ultimately synthesized with two additional study domains, production and environment, to map agroecoregons in the continental United States (CONUS). We based our selection of human dimensions variables on social-ecological systems frameworks. We evaluated six different socio-ecological frameworks with respect to their applicability to spatial analysis of agroecosystems.  Based upon our framework selection criteria, we chose to use the Ostrom Social Ecological Systems Framework (SESF) to guide our selection of human dimension variables.  Using a literature review and existing datasets, we identified candidate county-level clustering variables within the Tier 2 level of the Governance and Actors SESF categories. We then scored each variable based on its applicability within the SESF as well as the quality and availability of data to represent the variable; the top scoring variable in each Tier 2 category was used as inputs to the clustering algorithms. Clusters were derived using a combination of 14 different algorithms and 2-30 clusters and then mapped. We selected the final algorithm and number of clusters based upon interpretability and complementarity with other analyses (e.g., production and ecological regions). We used the means of variable values in each cluster to provide a characterization of these resulting regions as related to the human dimensions of agroecosystems.  Our analyses and resulting boundaries are critical for improving our understanding of agricultural management within a broader context that considers social and ecological systems. With an integration of human dimensions, we can better understand the landscape processes that contribute to management outcomes.