Recent climate variability and anomaly in the Great Lakes region provided a valuable opportunity in examining the response and regulation of ecosystem carbon cycling across different ecosystems. A simple Bayesian hierarchical model was developed and fitted against three-year (2011–2013) net ecosystem CO2 exchange (FCO2) data observed at three eddy-covariance sites (i.e., a deciduous woodland, a cropland, and a marsh) in northwestern Ohio. The model was designed to partition the variation of gross ecosystem production (GEP), ecosystem respiration (ER) and FCO2 that resulted directly from the short-term environmental forcing (i.e., direct effect) and indirectly from the changes of ecosystem functional traits (e.g., structural, physiological, and phenological traits) (i.e., indirect effect). Interannual variation of FCO2 was mainly driven by indirect effects, accounting for 54%, 89%, and 86% of the interannual variation at the woodland, cropland, and marsh sites, respectively. On the other hand, direct climatic effects accounted for 33% of interannual FCO2 variation at the woodland site and became irrelevant (<10%) at the cropland and marsh sites. In general, annual GEP and ER at each site tended to co-vary and dampen the interannual variability in FCO2. Yet, year-to-year changes of GEP and ER were not spatially synchronous, suggesting that the ecosystem's response to climate was strongly site-specific in terms of the annual net CO2 uptake. Future research should focus on the disparate response among ecosystems and develop a suitable framework to examine the mechanisms that drive differences in closely co-located ecosystems.
