Holistic aboveground ecological productivity efficiency modeling using data envelopment analysis in the southeastern...

Aboveground net primary productivity (ANPP) of an ecosystem is among the most important metrics of valued ecosystem services. Measuring the efficiency scores of ecological production (ESEP) based on ANPP using relevant variables is valuable for identifying inefficient sites. The efficiency scores computed by the Data Envelopment Analysis (DEA) may be influenced by the number of input variables incorporated into the models and two DEA settings-orientations and returns-to-scales (RTSs). Therefore, the objectives were threefold to: (1) identify soil-environmental variables relevant to ANPP, (2) assess the sensitivity of ESEP to the number of input variables and DEA settings, and (3) assess local management relations with ESEP. The ANPP rates were calculated for pine forests in the southeastern U.S. where 10 management types were used. This was followed by an all-relevant variable selection technique based on 696 variables that cover biotic, pedogenic, climatic, geological, and topographical factors. Five minimal-optimal variable selection techniques were further applied to create five parsimonious sets that contain a different number of variables used as DEA inputs. After setting ANPP as the output variable, two DEA orientations (input/output) and six RTS were applied to compute ESEP. The variable selection methods succeeded in objectively identifying the major factors relevant to ANPP variation. The site index showed the highest correlation with ANPP (r = 0.39), while various precipitation factors were negatively correlated (r = - 0.15~ - 0.29, p < 0.01). Parsimonious ESEP models observed a decrease in score variances as the number of input variables increased. Various RTS produced similar scores across orientations. Of the DEA settings, an output orientation with decreasing RTS produced the most progressive ESEP with large variation. Results also suggested that macro- and micronutrient fertilization is the best combination of management strategies to achieve high ESEP. This holistic benchmark approach can be applied to other ecological functions in diverse regions.