Global primary productivity, respiration, litter decomposition, and climate: Revisiting a paradigm

Background/Question/Methods

The relationship between climatology and ecosystem pattern and process, specifically net primary productivity (NPP), is at this point dogma, but the actual relationship has only been documented for a spatially and biologically limited set of data. In doing so authors have found that the former is a linear function of certain climatic indices, namely Actual Evapotranspiration (AET), however, the affect of AET as well as elevation, temperature, and precipitation has largely been assumed a priori. It remains to be seen at the global scale whether these indices drive in a similar manner gross primary production (GPP), respiration, whether aboveground autotrophic (R_AA) or soil heterotrophic (R_SH), and actual decomposition of leaf and course woody debris (LCWD). This analysis was designed to revisit the linear ecosystem process-climate axioms sensu lato via a spatially broad and biologically rich data-set. AET at the global scale is inferior to elementary climatic indices; including winter low, summer high – winter low, and annual average temperature (°C).
Results/Conclusions

Annual precipitation and elevation are the weakest predictors of ecosystem pattern and process, with average summer high intermediately robust. The slopes of the log-log NPP relative to AET function under a current, historic, and both data-sets were 0.736x, 1.697x, and 3.135ln(x), with respective R² of 0.426, 0.903, and 0.669. This disparity is undoubtedly a function of data-set breadth, both spatially and with reference to encompassed ecosystems. The aforementioned climatic indices most robustly describe global patterns in NPP and GPP, with a slight decline in capability relative to R_AA and R_SH. None of the variables explain more than 20% of the variability associated with carbon-use efficiency (NPP:GPP) and R_SH:R_AA. It is clear from this analysis that the efficacy of AET as a predictor of ecosystem process is more an artifact of data-set size and the fact that non-random selection of data-points may lead to erroneously inflated correlations. Studies that are designed to analyze this type of correlation a/o causation would benefit from data-mining that insure a continuous rather than discrete range dependent and independent variables. This was the objective of the current study proving that the very dynamic AET, while biologically relevant may lack the ability to account for processes taking place in cold/ and warm/dry environments, while temperature indices at the global scale are not hindered in their predictive capability under such scenarios.