Quantifying productivity responses to antecedent environmental drivers at multiple time-scales

Background/Question/Methods

Ecological processes are influenced by antecedent (past) conditions over multiple time scales. Understanding the importance of antecedent conditions and their time scales of influence requires sufficient temporal information on the drivers and processes. We describe the “stochastic antecedent modeling” (SAM) approach for quantifying the time-scales, temporal patterns, and influence of antecedent conditions on ecosystem processes. We apply SAM to understand the drivers of productivity in semi-arid grasslands, at annual, seasonal, and sub-daily time-scales. We describe the framework in the context of 52 years of annual net primary productivity (ANPP) to evaluate the importance of past monthly and annual precipitation. We expand upon this “simple” model to evaluate the antecedent drivers of sub-daily gross primary productivity (GPP) from the Prairie Heating and CO₂ Enrichment study involving manipulations of atmospheric CO₂, temperature, and soil water. GPP was estimated as the sum of net ecosystem exchange and ecosystem respiration (measured via chamber methods), yielding GPP data for 88 days over six years, accompanied by continuous measurements of light, temperature, humidity, soil moisture, and manual measurements of vegetation greenness (~bi-weekly) and plant nitrogen (seasonal). We fit a SAM model to the GPP data to evaluate the current and antecedent effects of these different drivers.

Results/Conclusions

The first analysis revealed important lags such that ANPP was affected by precipitation received 1-2 years prior to the growth year. Current precipitation amount and event size explained 47% of the variation in ANPP, and incorporation of antecedent precipitation explained an additional 28% of the variation. Likewise, the analysis of sub-daily GPP found that antecedent conditions were the primary factors governing GPP through their effects on light-use efficiency (LUE) and light-saturated GPP (GPPsat). In general, antecedent soil moisture was the most important factor underlying variation in LUE, while antecedent moisture, vapor pressure deficit, and air temperature were most important for GPPsat. In fact, current conditions were essentially irrelevant once antecedent conditions were considered. The soil moisture and air temperature lags were generally consistent across all global change treatments such that: (1) soil moisture conditions occurring ~2 weeks prior and (2) air temperatures occurring 3-4 days prior to the observed GPP were most important for predicting GPP. The application of the SAM framework to this GPP dataset revealed potential lag responses to environmental drivers, and suggests that such antecedent factors are critical for understanding productivity, especially under future novel climate conditions.