Results
Over more than a decade of experimental manipulation, growing conditions were altered in a consistent and significant way at our research plots. The warming treatment elevated temperature above and belowground by 3.3°C on average across all years, sites, and canopies (see Table 1, and Figure S1). Warming treatment had a significant effect on soil moisture; reducing VWC by 13% in the closed canopy plots and by 24% in open plots. In the open canopy plots reduced rainfall treatment caused an 11% decrease in VWC, and warming with reduced rainfall together reduced VWC by 35% (see Table 1 and Figures S1 and S2 for more details).
Analysis of 11 growing seasons of leaf gas exchange data across multiple species showed that rainfall reduction and warming treatments led to more conservative water use on average evidenced by decreasedg1 (the slope of the USO model serving as a proxy of the marginal water cost of carbon gain – λ) (P ≤ 0.0087, Table 4). However, species differed in their responsiveness to both drivers. We organize the presentation of results around the hypotheses.
H1: g1 decreases with reduced rainfall for all species. – Our hypothesis was supported asg1 was lower in reduced rainfall treatments (P = 0.006; Table 4 and Figures 1-5). This effect was consistent in all tested models (for selected additional models see methods and Table S1). Overall, plants grown under the rain reduced treatment regime reduced g1 by 10.5% on average compared to plants in ambient plots. This decrease of the g1parameter was generally consistent across both sites and all years (see Figure 5b). The role of VWC in these responses is presented below with respect to both rainfall and warming treatment effects.
H2: g1 will decrease with climate warming due to soil moisture reduction induced by elevated temperature. Mixed effect models showed that warming treatment strongly reducedg1 in both canopies (P < 0.0001; Table 4 and Figures 1-5). This effect was generally consistent across all models and years (see Table 4 and Figure 5). Overall, plants grown in warmed treatments reduced g1 by 25% in the understory and 18% in open canopy plots (see Table 4 and Figures 1-5). These responses support H2 (as further documented below).
Assessing soil moisture regulation of g1 . – As both the warming treatment and reduced rainfall had significant effects on VWC (Table 1) we explored the role that soil moisture might play in regulating g1 . Estimates ofg1 for plants experiencing different levels of soil moisture in each treatment (binned into three categories, i.e., low, medium, and high soil VWC – refer to modeling and data analysis section of methods for additional details on VWC bins) showed thatg1 declined when soil water content was low either due to reduced rainfall or elevated temperatures, or both (Table 5, and Figure 4a,b). Moreover, when we add soil VWC as a covariate the significance of the main effects (i.e., warming and rainfall reduction) is eliminated (P > 0.3855, Table 6) in open canopy but not in the closed (P < 0.0001, Table 6) while soil VWC alone has a significant (P < 0.0124, Table 6) effect on g1 in both canopies. Also, a combination of soil VWC with warming becomes significant (P< 0.019) in both canopies that collectively demonstrates strong influence of soil moisture on g1 that overtakes the effect of warming particularly in the open canopy. Slightly different response of plants in closed canopy is likely due to differences in microenvironments where plots in closed canopy have overall higher soil VWC on average that does not consistently become low enough for plants to drive changes in g1 .
H3: reduced rainfall and warming will have an additive effects on g1 because the primary mechanism of both warming and reduced rainfall effects on g1 will be via the same pathway, of reduced VWC on stomatal behavior. – Warming and reduced rainfall did not show significant interaction in any model (P ≥ 0.3621, for details see Table 4, 5, 6 and S1, and Figures 1b, 2b, 3b, 4b, and 5b) confirming our hypothesis. Across all other sources of variation (in open plots), reduced rainfall alone caused 8.3% decline while warming alone resulted in 15.6% decline ofg1 , and both treatments acting together reducedg1 by 26.5% (see Table S2 and Figures 1b, 2b, 3b, and 5b).
H4: species adapted to either drier and/or warmer conditions will on average have lower and less sensitive g1 than species adapted to more mesic or cooler conditions. Species (for details about species see Tables 3 and S2) differed in their g1 parameter (P< 0.0001, Tables 4, 6, S1 and S2, and Figure 1). Species average g1 in ambient growth conditions ranged in open canopy from 2.8 for P. banksiana to 5.5 in F. alnus , and in closed canopy from 2.3 for A. balsamea to 6.1 for R. cathartica . The four invasive species (i.e. , F. alnus, L. morrowii, L. tatatrica , and R. cathartica ) and native T. americana had the highest g1 of all species (Figure 1 and Table S2). The boreal species had on average the lowest g1 with native temperate species in between invasive and boreal groups. Species with higher drought tolerance indices had slightly higher g1 on average. For more details on the average g1values across species, their respective groupings (e.g., biome association, drought tolerance, etc.) and treatment effects, see Figures 1-3 and Table S2.
There were few differences among species and their respective higher groupings (i.e., drought tolerance, biome association, and phylogenic associations) in sensitivity of g1 , (that is the decline of g1 in response to rainfall reduction or warming) (Figures 2-3, Table S1) as most species and groups responded to warming and reduced rainfall by significantly reducingg1 (Table 4, Figure 1). For example, in closed canopy plots there was a large individualistic variation in responses (P = 0.0384, Figure 1) to warming, with change ing1 ranging from a 10.6% increase in P. glauca to decreases for all other species that ranged from 3.3% forQ. rubra , to 60.5% in T. canadensis .