The response of stomatal conductance to seasonal drought in tropical forests
Jin Wu,Shawn P. Serbin,Kim S. Ely,Brett T. Wolfe,L. Turin Dickman,Charlotte Grossiord,Sean T. Michaletz,Adam D. Collins,Matteo Detto,Nate G. McDowell,S. Joseph Wright,Alistair Rogers
DOI: https://doi.org/10.1111/gcb.14820
IF: 11.6
2019-10-17
Global Change Biology
Abstract:<p>Stomata regulate CO<sub>2</sub> uptake for photosynthesis and water loss through transpiration. The approaches used to represent stomatal conductance (<i>g<sub>s</sub></i>) in models vary. In particular, current understanding of drivers of the variation in a key parameter in those models, the slope parameter (i.e. a measure of intrinsic plant water‐use‐efficiency), is still limited, particularly in the tropics. Here we collected diurnal measurements of leaf gas exchange and leaf water potential (<i>Ψ<sub>leaf</sub></i>), and a suite of plant traits from the upper canopy of 15 tropical trees in two contrasting Panamanian forests throughout the dry season of the 2016 El Niño. The plant traits included wood density, leaf‐mass‐per‐area (LMA), leaf carboxylation capacity (<i>V</i><sub>c,max</sub>), leaf water content, the degree of isohydry, and predawn <i>Ψ<sub>leaf</sub></i>. We first investigated how the choice of four commonly used leaf‐level <i>g<sub>s</sub></i> models with and without the inclusion of <i>Ψ<sub>leaf</sub></i> as an additional predictor variable influence the ability to predict <i>g<sub>s</sub></i>, and then explored the abiotic (i.e. month, site‐month interaction) and biotic (i.e. tree‐species‐specific characteristics) drivers of slope parameter variation. Our results show that the inclusion of <i>Ψ<sub>leaf</sub></i> did not improve model performance and that the models that represent the response of <i>g<sub>s</sub></i> to vapor pressure deficit performed better than corresponding models that respond to relative humidity. Within each <i>g<sub>s</sub></i> model, we found large variation in the slope parameter, and this variation was attributable to the biotic driver, rather than abiotic drivers. We further investigated potential relationships between the slope parameter and the six available plant traits mentioned above, and found that only one trait, LMA, had a significant correlation with the slope parameter (R<sup>2</sup>=0.66, <i>n</i>=15), highlighting a potential path towards improved model parameterization. This study advances understanding of <i>g<sub>s</sub></i> dynamics over seasonal drought, and identifies a practical, trait‐based approach to improve modeling of carbon and water exchange in tropical forests.</p><p>This article is protected by copyright. All rights reserved.</p>
environmental sciences,biodiversity conservation,ecology