Elevated CO2 and warming effects on grassland plant mortality are determined by the timing of rainfall

Abstract

Background and aims Global warming is expected to increase the mortality rate of established plants in waterlimited systems because of its effect on evapotranspiration. The rising CO2 concentration ([CO2]), however, should have the opposite effect because it reduces plant transpiration, delaying the onset of drought. This potential for elevated [CO2] (eCO2) to modify the warming effect on mortality should be related to prevailing moisture conditions. This study aimed to determine the impacts of warming by 2 °C and eCO2 (550 μmol mol-1) on plant mortality in an Australian temperate grassland over a 6-year period and to test how interannual variation in rainfall influenced treatment effects. Methods Analyses were based on results from a field experiment, TasFACE, in which grassland plots were exposed to a combination of eCO2 by free air CO2 enrichment (FACE) and warming by infrared heaters. Using an annual census of established plants and detailed estimates of recruitment, annual mortality of all established plants was calculated. The influence of rainfall amount and timing on the relative impact of treatments on mortality in each year was analysed using multiple regression techniques. Key Results Warming and eCO2 effects had an interactive influence on mortality which varied strongly from year to year and this variation was determined by temporal rainfall patterns. Warming tended to increase densityadjusted mortality and eCO2 moderated that effect, but to a greater extent in years with fewer dry periods. Conclusions These results show that eCO2 reduced the negative effect of warming but this influence varied strongly with rainfall timing. Importantly, indices involving the amount of rainfall were not required to explain interannual variation in mortality or treatment effects on mortality. Therefore, predictions of global warming effects on plant mortality will be reliant not only on other climate change factors, but also on the temporal distribution of rainfall.