Disentangling the Regional Climate Impacts of Competing Vegetation Responses to Elevated Atmospheric CO2

Abstract

Biophysical vegetation responses to elevated atmospheric carbon dioxide (CO₂) affect regional hydroclimate through two competing mechanisms. Higher CO₂ increases leaf area (LAI), thereby increasing transpiration and water losses. Simultaneously, elevated CO₂ reduces stomatal conductance and transpiration, thereby increasing rootzone soil moisture. Which mechanism dominates in the future is highly uncertain, partly because these two processes are difficult to explicitly separate within dynamic vegetation models. We address this challenge by using the GISS ModelE global climate model to conduct a novel set of idealized 2×CO₂ sensitivity experiments to: evaluate the total vegetation biophysical contribution to regional climate change under high CO₂; and quantify the separate contributions of enhanced LAI and reduced stomatal conductance to regional hydroclimate responses. We find that increased LAI exacerbates soil moisture deficits across the sub-tropics and more water-limited regions, but also attenuates warming by ∼0.5-1°C in the US Southwest, Central Asia, Southeast Asia, and northern South America. Reduced stomatal conductance effects contribute ∼1°C of summertime warming. For some regions, enhanced LAI and reduced stomatal conductance produce nonlinear and either competing or mutually amplifying hydroclimate responses. In northeastern Australia, these effects combine to exacerbate radiation-forced warming and contribute to year-round water limitation. Conversely, at higher latitudes these combined effects result in less warming than would otherwise be predicted due to nonlinear responses. These results highlight substantial regional variation in CO₂-driven vegetation responses and the importance of improving model representations of these processes to better quantify regional hydroclimate impacts.

Figure 1

The change in peak growing season LAI (aggregated across vegetation types) from Baseline for (a) 50% increase sensitivity experiment and (b) CMIP6 “pseudo” 2×CO₂ ensemble. Red markers indicate small LAI decreases. Both panels (a) and (b) use colorbar under (b). (c) Fractional (0–1) coverage per gridcell of C4 grass used in modelE. (d) Difference in peak growing season LAI for 50% increase sensitivity experiment minus CMIP6 “pseudo” 2×CO₂ experiment.

Figure 2

AllEffects responses (measured against Baseline as in Table 2) for (a) surface temperature (°C) and (b) soil moisture difference (top 57 cm), normalized by standard deviation of the reference experiment in Table 2 (z‐score). This unit is used for all subsequent soil moisture anomalies shown herein. Stippling indicates responses are not significant via a Student’s t‐test (p < 0.05, sample size adjusted for autocorrelation where relevant). The ratio of BioPhys to AllEffects to a doubling of CO₂ for (c) surfacetemperature and (d) soil moisture, shown for BioPhys significant gridcells. All responses shown for hemisphere‐respective growing seasons.

Figure 3

Change in growing season (a) surface temperature, (b) soil moisture (top 57 cm), (c) precipitation, and (d) transpiration due to Rad_Only effect. Stippling indicates responses are not significant via a Student’s t‐test (p < 0.05, sample size adjusted for autocorrelation where relevant).

Figure 4

Zonally averaged growing season anomalies for (a) soil moisture for top 57 cm, (b) precipitation, and (c) surface temperature for BioPhys effect (green), Con_Only (red), and LAI_Only (blue). Only significant gridcells (via a Student’s t‐test described in the Methods section) were used in the creating the zonal average.

Figure 5

50‐years climatological growing‐season anomalies. Row 1: soil moisture (z‐score, top 57 cm) for Con_Only (a), LAI_only (b), and Biophys (c). Row 2: precipitation for Con_Only (d), LAI_only (e), and Biophys (f). Anomalies are shown for each hemisphere’s respective summer. Stippling indicates responses are not significant via a Student’s t‐test (p < 0.05, sample size adjusted for autocorrelation where relevant).

Figure 6

50‐years climatological growing‐season anomalies. Row 1: surface temperature (˚C) for Con_Only (a), LAI_only (b), and Biophys (c). Row 2: evaporative fraction (×100) for Con_Only (d), LAI_only (e), and Biophys (f). Anomalies are shown for each hemisphere’s respective summer. Stippling indicates responses are not significant via a Student’s t‐test (p < 0.05, sample size adjusted for autocorrelation where relevant).

Figure 7

Row 1 shows Con_Only growing season changes in (a) pressure level × latitude relative humidity, (b) depth of the planetary boundary layer (m), and (c) pressure level × latitude change in temperature. Similarly, Row 2 shows respective changes in these variables for LAI_Only, and Row 3 for BioPhys. Stippling indicates responses are not significant via a Student’s t‐test (p < 0.05, sample size adjusted for autocorrelation where relevant).

Figure 8

AllEffects versus the combined LAI_Only + Con_Only + Rad_Only effect for growing season (a) surface temperature (˚C), and (b) soil moisture (mm, top 57 cm). A perfect correspondence between the AllEffects response and LAI_Only + Con_Only + Rad_Only response would fall on the 1:1 line shown on the figures. The geographic distribution of the nonlinear interaction term, taken as AllEffects – (RadOnly + LAIonly + ConOnly), for (a) surface temperature and (b) soil moisture during the growing season.

Figure 9

The geographic distribution of the growing season nonlinear interaction term, taken as AllEffects – (RadOnly + LAIonly + ConOnly), for (a) vapor pressure deficit, and (b) canopy evaporation, (c) low level cloud cover, and (d) net shortwave radiation.

Figure 10

Growing season responses for experiments prescribed with LAI from “pseudo” 2xCO2 experiment. Ratio of BioPhys to AllEffects for (a) surface temperature and (b) soil moisture for top 57 cm. AllEffects non‐linear responses (as computed in Section 3.5) for (c) surface temperature (native unit ˚C) and (d) soil moisture for top 57 cm (native unit mm).