CR 46:77-83 (2011)  -  DOI: https://doi.org/10.3354/cr00975

ABSTRACT: The structure and function of ecosystems are strongly affected by rainfall variability in grasslands worldwide. We used a diversity index (Shannon index [I]) and 3 non-diversity indices (coefficient of variation [CV], standard deviation [SD], and dry days [U_d]) to analyze rainfall variability and the relationship of rainfall variability with 1982–2008 above-ground net primary productivity (ANPP) and 2003–2008 growing season (May–September) net ecosystem CO₂ exchange (NEE). A series of bivariate regressions were performed with precipitation (P) and 1 measure of its temporal variability (CV, I, SD, or U_d) as an independent variable. The regression models were evaluated with regard to their ability to predict ANPP. Only the model with P and I as the predictors was statistically significant at p < 0.01, and had the highest coefficient of determination (R² = 0.50). ANPP increased with increasing I (r = 0.45, p = 0.05), but no consistent relationship with the total amount of annual rainfall was observed (r = –0.17, p = 0.46). Using all data from a 6 yr study period, the Shannon index explained 50% of the change in NEE. In general, the Shannon index had the best spread and sensitivity under different rainfall regimes and was the most appropriate index for carbon flux studies.

KEY WORDS: Rainfall variability · Shannon index · Coefficient of variation · Standard deviation · Dry day