A test of the seasonal availability of water hypothesis in a C3/C4 mixed grassland

Understanding how cool-season C3 and warm-season C4 grasses will respond to climate change is critical for predicting future grassland functioning. With warming, C4 grasses are expected to increase relative to C3 grasses. But, alterations in the seasonal availability of water may also influence C3/C4 dynamics because of their distinct seasons of growth. To better understand how shifts in the seasonal availability of water can affect ecosystem function in a northern mixed grass prairie in southeastern Wyoming, we reduced early season rainfall (April – June 2021) using rainout shelters and added the amount of excluded precipitation during the latter half of the growing season (July-September), effectively shifting spring rainfall to summer rainfall.  As expected, this shift in precipitation seasonality influenced patterns of soil water availability, leading to increased soil respiration in the summer months and sustained canopy greenness throughout the growing season. Despite these responses, there were no significant differences in C3 aboveground net primary production (ANPP) between the seasonally shifted treatment (SEAS) and the plots that received ambient (AMB) precipitation. This was likely due to the high levels of spring soil moisture present before rainout shelters were deployed that sustained C3 grass growth.  However, in plots with high C4 grass cover, C4 ANPP increased significantly in response to increased summer rainfall. Overall, we provide the first experimental evidence that shifts in the seasonality of precipitation, with no change in temperature, will differentially impact C3 vs. C4 species, altering the dynamics of carbon cycling and canopy albedo in this extensive semi-arid grassland.

Methods
Experimental Design

Before the 2021 growing season, we established twenty 1 m2 plots (n=10 per treatment). Plots were separated by at least 3 m, and aluminum flashing was installed (10 cm belowground and 5 cm aboveground) 20 cm outside of the plot perimeter to reduce surface and shallow soil water movement into and out of each plot.  Rainout shelter roofs (2.44 m × 3.05 m made of clear corrugated polycarbonate, Suntuf, Palram Americas) that were larger than the 1 m2 plots were then placed over ten of the plots. Roofs were initially installed 80 cm above the ground at a slight angle to allow water to drain away from the plot; later in the season, the shelters were raised to 100 cm. Although previous work has demonstrated that these shelters have minimal influence on the microclimate (Loik et al. 2019; Post and Knapp 2020; Hoover et al. 2022), we monitored soil temperatures at 10 cm weekly and evaluated light transmission under the roofs using a 1-m linear quantum light sensor (Decagon AccuPAR, model LP-80).

We altered the seasonal dynamics of soil water availability (seasonally shifted treatment, SEAS) by using the clear roofs to exclude all precipitation from April 10 – June 30 (some minor blow-in of precipitation during storms was inevitable). We removed the shelters July 1 and added the equivalent amount of water excluded in the spring to these plots in addition to the ambient precipitation received.  The additional precipitation was applied manually throughout July – September replicating the distribution of rainfall events from the spring (i.e., additions were similar in event frequency and magnitude, Table S1). This ensured that total growing season precipitation was similar for the SEAS and the ambient (AMB) treatment, which received natural precipitation for the entire growing season. Precipitation was recorded at a nearby NOAA weather station (Cheyenne Weather Forecast Office, (41.1516, -104.80622)). Treatments (n=10) were randomly assigned, and because there was some minor topographical variation in the landscape (10 plots were slightly uphill from the others), we assigned treatments within two blocks to control for any effects topography. Block effects were non-significant in our analyses, but after plants became active, we noted that the cover of C3 and C4 functional groups varied widely among plots. Thus, we estimated total plant cover by species in each plot to account for this variation in our analyses.

Measured Responses

We measured soil moisture (volumetric water content, VWC) weekly in all plots throughout the experiment (April 10 – September 23) with a 20 cm handheld soil moisture time-domain reflectometry probe (Campbell-Scientific Hydrosense II). This instrument integrates soil moisture in the top 20 cm of soil where most of the root biomass in this grassland is located (Sun et al. 1997; Carillo et al. 2014). To assess treatment effects on plant water status, we estimated mid-day (12:00 - 14:00hr MST) leaf water potential with a Scholander pressure chamber (PMS instruments) for a dominant C3 grasses, Pascopyrum smithii, and the dominant C4 grass, Bouteloua gracilis. Fully expanded, mature canopy leaves (1-2 leaves per plot, n=6 per treatment) were collected each week. Because the C3 and C4 grasses become active at different times of the growing season, we measured P. smithii water status from May 27 – Sept. 16 and B. gracilis from June 15 - Sept. 16.

            To evaluate how differences in the seasonal availability of water influenced C3 and C4 dynamics and ecosystem function, we measured canopy greenness and soil CO2 efflux weekly, photosynthetic rates for the primary C3 and C4 species in June and July, and ANPP at the end of the growing season (September).

            Canopy greenness, measured to assess canopy-scale phenological responses and to serve as a proxy for potential ecosystem carbon uptake, was estimated with repeat digital photography (following the methods of Post and Knapp 2020, Hoover et al. 2022). Briefly, an iPhone camera was positioned directly above a marked 50 cm x 50 cm frame in a corner of each plot, and each photograph was then cropped to contain only the interior area of the frame. These cropped photos were processed using the R package EBImage (Pau et al. 2010) to calculate the average green chromatic coordinate (GCC) index (Filippa et al. 2016). The GCC index accounts for variation in pixel brightness (Filippa et al. 2016), thus avoiding background light levels and potential infrastructure impacts.

Soil respiration was measured weekly (May 5 – Sept. 23) to quantify how the treatments influenced this important carbon flux (Hashimoto et al. 2015). Permanent PVC collars (10 cm in diameter, n=6 per treatment) were installed in locations between grasses at the end of April (2.4 cm belowground and 2 cm aboveground), and all vegetation within the collars was removed (clipped at the base). Before each measurement, any new vegetation growth was also gently removed.  Soil respiration was measured using a 6400-09 soil flux chamber attached to an LI-6400XT (LiCor., Inc, Lincoln NE, USA). Measurements were taken mid-day (between 8:30hr – 12:30hr MST) at ambient CO2 concentration, humidity, and temperature.

            Leaf gas-exchange was measured (June 23-24) prior to roofs coming off and after the roofs were removed (July 24-25). On each date, a portable photosynthesis system (LI-6400, LiCor., Inc, Lincoln NE, USA) was used to measure the CO2 uptake (net photosynthesis, or A) on 12 fully expanded mature upper canopy leaves for both C3 (P. smithii) and C4 (B. gracilis) individuals in each treatment.

The LI-6400 was fitted with a 3×2 cm cuvette head and a red-blue LED light source. For all measurements, flow rate was held constant at 600 μmol s-1. The temperature exchanger was set to an average midday summer temperature of 30 °C. Leaf temperature (Tleaf) was measured with a thermocouple and averaged 31 ± 1.7 °C (standard deviation) across all measurement dates. Relative humidity conditions in the chamber were controlled near ambient levels but did vary slightly depending upon water vapor fluxes from the leaf. Photosynthetic photon flux density in the chamber was set at 1800 μmol m-2 s-1, approximating full-sun conditions to measure light-saturated net photosynthesis (Asat) and stomatal conductance to water vapor (gs, Fig. S1) at a chamber reference [CO2] of 420 μmol mol-1. All measurements occurred between 10:00hr and 15:00hr MST

Finally, ANPP was estimated near the end of the growing season (Sept. 1-2) in all plots as plants began to senesce.  For each plot, all aboveground vegetation within two 0.1 m2 subplots was harvested to ground height, sorted by functional group (C3 grass, C4 grass, forb, woody, or annual grass), and then dried at 60 °C for 48 hours before being weighed to the nearest 0.01g. Previous year’s growth was easily distinguished from current year growth and was not included.