Assessing ecosystem variance at different scales to generalize about pasture management in southern Wisconsin

Pasture-based agroecosystems are an expanding enterprise in the northcentral U.S., but their structure and function have received little attention. We implemented a manipulative experiment to test pasture management effects on several agroecosystem variables at a university research farm in southcentral Wisconsin. In concert, we established a mensurative experiment across eight grass-based farms in the same region where we measured the same variables under existing land uses. Here we describe how total and relative variation in greenhouse gas fluxes, ground cover, soil arthropods, and soil inorganic N were distributed across nested spatial scales, management treatments, and unexplained realms of these systems. This approach seeks to improve our understanding of pasture management effects on agroecosystem stocks and fluxes, while expanding our inferential space to a wider geographical area. Greenhouse gas fluxes (CH4 and N2O) were an order-of-magnitude more variable than soil N pools, ground cover, or arthropod abundances. Most of the variance in these fluxes was at the subsample level, indicating the need for either more sampling chambers per management treatment or an efficient stratification scheme or both. Methane was more variable farm-to-farm than among management treatments within farms, while N2O showed the opposite pattern indicating that farmers are likely to have more influence on N2O fluxes via management than they can have on CH4 fluxes. Variability in arthropod groups and ground cover was rather uniformly distributed across nested spatial scales, with the exception of beetles and harvestmen, which displayed little subsample variability. Soil inorganic N pools were more variable from farm-to-farm than among management treatments, but most variation was at the subsample level. The least variable response was from the greenhouse gas CO2, which is driven by both auto- and hetero-trophic organisms. Calculation of total and relativized coefficients of variation allowed for a standardized comparison of agroecosystem variables possessing disparate units of measure and facilitated conclusions about the generality of results across a broader geographical area than just a single farm.