Ecosystem responses to increased variability in rainfall, a prediction of general circulation models, were assessed in native grassland by reducing storm frequency and increasing rainfall quantity per storm during a 4-year experiment. More extreme rainfall patterns, without concurrent changes in total rainfall quantity, increased temporal variability in soil moisture and plant species diversity. However, carbon cycling processes such as soil CO2 flux, CO2 uptake by the dominant grasses, and aboveground net primary productivity (ANPP) were reduced, and ANPP was more responsive to soil moisture variability than to mean soil water content. Our results show that projected increases in rainfall variability can rapidly alter key carbon cycling processes and plant community composition, independent of changes in total precipitation.

Rainfall variability is a key driver of ecosystem structure and function in grasslands worldwide. Changes in rainfall patterns predicted by global climate models for the central United States are expected to cause lower and increasingly variable soil water availability, which may impact net primary production and plant species composition in native Great Plains grasslands. We experimentally altered the timing and quantity of growing season rainfall inputs by lengthening inter-rainfall dry intervals by 50%, reducing rainfall quantities by 30%, or both, compared to the ambient rainfall regime in a native tallgrass prairie ecosystem in northeastern Kansas. Over three growing seasons, increased rainfall variability caused by altered rainfall timing with no change in total rainfall quantity led to lower and more variable soil water content (0-30 cm depth), an approximately 10% reduction in aboveground net primary productivity (ANPP), increased root to shoot ratios, and greater canopy photon flux density at 30 cm above the soil surface. Lower total ANPP primarily resulted from reduced growth, biomass and flowering of subdominant warm-season C4 grasses while productivity of the dominant C4 grass Andropogon gerardii was relatively unresponsive. In general, vegetation responses to increased soil water content variability were at least equal to those caused by imposing a 30% reduction in rainfall quantity without altering the timing of rainfall inputs. Reduced ANPP most likely resulted from direct effects of soil moisture deficits on root activity, plant water status, and photosynthesis. Altered rainfall regimes are likely to be an important element of climate change scenarios in this grassland, and the nature of interactions with other climate change elements remains a significant challenge for predicting ecosystem responses to climate change.

Predicted climate changes in the US Central Plains include altered precipitation regimes with increased occurrence of growing season droughts and higher frequencies of extreme rainfall events. Changes in the amounts and timing of rainfall events will likely affect ecosystem processes, including those that control C cycling and storage. Soil carbon dioxide (CO 2 ) flux is an important component of C cycling in terrestrial ecosystems, and is strongly influenced by climate. While many studies have assessed the influence of soil water content on soil CO 2 flux, few have included experimental manipulation of rainfall amounts in intact ecosystems, and we know of no studies that have explicitly addressed the influence of the timing of rainfall events. In order to determine the responses of soil CO 2 flux to altered rainfall timing and amounts, we manipulated rainfall inputs to plots of native tallgrass prairie (Konza Prairie, Kansas, USA) over four growing seasons (1998)(1999)(2000)(2001). Specifically, we altered the amounts and/or timing of growing season rainfall in a factorial combination that included two levels of rainfall amount (100% or 70% of naturally occurring rainfall quantity) and two temporal patterns of rain events (ambient timing or a 50% increase in length of dry intervals between events). The size of individual rain events in the altered timing treatment was adjusted so that the quantity of total growing season rainfall in the ambient and altered timing treatments was the same (i.e. fewer, but larger rainfall events characterized the altered timing treatment). Seasonal mean soil CO 2 flux decreased by 8% under reduced rainfall amounts, by 13% under altered rainfall timing, and by 20% when both were combined (Po0.01). These changes in soil CO 2 flux were consistent with observed changes in plant productivity, which was also reduced by both reduced rainfall quantity and altered rainfall timing. Soil CO 2 flux was related to both soil temperature and soil water content in regression analyses; together they explained as much as 64% of the variability in CO 2 flux across dates under ambient rainfall timing, but only 38-48% of the variability under altered rainfall timing, suggesting that other factors (e.g. substrate availability, plant or microbial stress) may limit CO 2 flux under a climate regime that includes fewer, larger rainfall events. An analysis of the temperature sensitivity of soil CO 2 flux indicated that temperature had a reduced effect (lower correlation and lower Q 10 values) under the reduced quantity and altered timing treatments. Recognition that changes in the timing of rainfall events may be as, or more, important than changes in rainfall amount in affecting soil CO 2 flux and other components of the carbon cycle highlights the complex nature of ecosystem responses to climate change in North American grasslands.

Global climate change is predicted to alter growing season rainfall patterns, potentially reducing total amounts of growing season precipitation and redistributing rainfall into fewer but larger individual events. Such changes may affect numerous soil, plant, and ecosystem properties in grasslands and ultimately impact their productivity and biological diversity. Rainout shelters are useful tools for experimental manipulations of rainfall patterns, and permanent fixed-location shelters were established in 1997 to conduct the Rainfall Manipulation Plot study in a mesic tallgrass prairie ecosystem in northeastern Kansas. Twelve 9 x 14-m fixed-location rainfall manipulation shelters were constructed to impose factorial combinations of 30% reduced rainfall quantity and 50% greater interrainfall dry periods on 6 x 6-m plots, to examine how altered rainfall regimes may affect plant species composition, nutrient cycling, and above-and belowground plant growth dynamics. The shelters provided complete control of growing season rainfall patterns, whereas effects on photosynthetic photon flux density, nighttime net radiation, and soil temperature generally were comparable to other similar shelter designs. Soil and plant responses to the first growing season of rainfall manipulations (1998) suggested that the interval between rainfall events may be a primary driver in grassland ecosystem responses to altered rainfall patterns. Aboveground net primary productivity, soil CO 2 flux, and flowering duration were reduced by the increased interrainfall intervals and were mostly unaffected by reduced rainfall quantity. The timing of rainfall events and resulting temporal patterns of soil moisture relative to critical times for microbial activity, biomass accumulation, plant life histories, and other ecological properties may regulate longerterm responses to altered rainfall patterns.

Climate change is causing measurable changes in rainfall patterns, and will likely cause increases in extreme rainfall events, with uncertain implications for key processes in ecosystem function and carbon cycling. We examined how variation in rainfall total quantity (Q), the interval between rainfall events (I), and individual event size (S E ) affected soil water content (SWC) and three aspects of ecosystem function: leaf photosynthetic carbon gain (A CO2 ), aboveground net primary productivity (ANPP), and soil respiration (J CO2 ). We utilized rainout shelter-covered mesocosms (2.6 m 3 ) containing assemblages of tallgrass prairie grasses and forbs. These were hand watered with 16 I Â Q treatment combinations, using event sizes from 4 to 53 mm. Increasing Q by 250% (400-1000 mm yr À1 ) increased mean soil moisture and all three processes as expected, but only by 20-55% (P 0.004), suggesting diminishing returns in ecosystem function as Q increased. Increasing I (from 3 to 15 days between rainfall inputs) caused both positive (A CO2 ) and negative (J CO2 ) changes in ecosystem processes (20-70%, P 0.01), within and across levels of Q, indicating that I strongly influenced the effects of Q, and shifted the system towards increased net carbon uptake. Variation in S E at shorter I produced greater response in soil moisture and ecosystem processes than did variation in S E at longer I, suggesting greater stability in ecosystem function at longer I and a priming effect at shorter I. Significant differences in ANPP and J CO 2 between treatments differing in I and Q but sharing the same S E showed that the prevailing pattern of rainfall influenced the responses to a given event size. Grassland ecosystem responses to extreme rainfall patterns expected with climate change are, therefore, likely to be variable, depending on how I, Q, and S E combine, but will likely result in changes in ecosystem carbon cycling.