Anticipating changes in variability of grassland production due to increases in interannual precipitation variability. Issue 5 (22nd May 2014)

• Record Type: Journal Article
• Authors: Hsu, Joanna S.; Adler, Peter B.
• Is Part Of: Ecosphere. Volume 5:Issue 5(2014)
• Page Start: 1
• Page End: 15
• Publication Date: 2014-05-22
• Subjects: aboveground net primary production -- lag effects -- legacies -- precipitation variability -- primary production; Ecology -- Periodicals; Ecology; Periodicals; 577.05
• DOI: 10.1890/ES13-00210.1
• Languages: English
• ISSNs: 2150-8925

Abstract

Abstract : Expected increases in interannual precipitation variability due to climate change will lead to increases in the variability of primary production, with potentially important consequences for natural resource management. Previous work has suggested that various biotic and abiotic processes might amplify or buffer variation in production in response to variation in precipitation. In particular, production to rain variability ratios (PRVR), the coefficient of variation of production divided by the coefficient of variation of precipitation, indicate that production is often relatively more variable than precipitation. We used 37 long‐term data sets from grasslands across the globe to test how future increases in precipitation variability might alter the variability of aboveground net primary production (ANPP). We demonstrate that PRVR is not a useful metric: it is predicted by a site's precipitation‐production relationship and a PRVR greater than 1 need not imply that increases in the variability of ANPP will be disproportionately greater than increases in precipitation variability. Instead, it is the form of the precipitation‐ANPP relationship that determines how increases in precipitation variability will impact ANPP variability. We fit linear, lag effect, and nonlinear precipitation‐ANPP relationships to each data set. At most sites, the precipitation‐ANPP relationship is weakly nonlinear, though the lag effect model, incorporating previous year ANPP, performed best at several sites. To test whether the three models project differences in the response of ANPP to future increases in precipitation variability, we directly perturbed the observed precipitation time series and quantified the results of this perturbation on ANPP variability. Under simple linear or lag effect models, relative increases in ANPP variability were always equal to the relative increases in precipitation variability. When we modeled ANPP as a nonlinear, saturating function of precipitation, projected increases in ANPP variability were disproportionately high, with production dropping more in dry years than it increases in wet years. In six cases, increases in ANPP variability were twice as large as increases in precipitation variability. Based on Akaike model weights, a 5% increase in precipitation variability would cause a 6.3% increase in ANPP variability on average.