Is the dual-isotope conceptual model fully operational?

Abstract

The stable carbon isotopic composition (δ13C) of organic matter is a valuable measure of plant physiological processes. Unfortunately, the impacts of carbon assimilation (A, demand for CO2) and stomatal conductance (gs, supply of CO2) on Rubisco discrimination (through Ci/Ca) are difficult to separate. Thus, a conceptual model (Scheidegger et al. 2000) was developed to constrain the interpretation of δ13C variation by measuring oxygen isotopic composition (δ18O) on the same material (the ‘dual-isotope’ approach). The idea is that δ18O, as a proxy for evaporative flux, will be modified by gs but not A. Therefore, if changes in environmental conditions cause a long-term change in A and/or gs, this should be reflected in the carbon and oxygen isotope ratios of organic matter, respectively. This model is conceptually sound and a number of papers (Sidorova et al. 2009, Brooks and Mitchell 2011) have used this model for interpreting δ13C variation in tree rings. In the current issue of Tree Physiology, Barnard et al. have utilized the dual-isotope model to infer physiological responses of mature Pseudotsuga menziesii (Mirb) Franco trees to environmental variation associated with site differences and canopy position. They developed a novel approach to eliminate confounding factors by plotting relative changes (from the grand mean) in tree ring δ13C and δ18O values. They used the conceptual model to infer temporal changes in gs, spatial variation in A associated with leaf nitrogen content and different levels of physiological responsiveness to environmental forcing for trees from different canopy positions. This was an ambitious use of the dual-isotope model due to complex environmental variation associated with sites (slope and aspect differences), canopy position (three crown classes), seasonal separation (earlywood and latewood) and inter-annual climate (8 years). The dual-isotope model has not been tested under such complex conditions, and the highly variable results of Barnard et al. (2012) make model interpretation challenging. Our ability to take advantage of this conceptual model is limited by a number of model assumptions and constraints that are often overlooked. Barnard et al. (2012) recognized many of these assumptions and tried to quantify or limit confounding variation; however, they did not test whether model predictions regarding A and gs were realized in their system, which would have been a valuable confirmation of model applicability. They assumed that this conceptual model was operational for interpreting tree ring isotope variation in complex forest ecosystems. The purpose of this commentary is to highlight areas of caution that must be recognized before the dual-isotope model can be routinely utilized and encourage research that could help reduce model uncertainties. 1. The dual-isotope model interprets changes in δ18O as primarily influenced by gs. This requires that environmental influences on evaporative enrichment (Craig and Gordon 1965) including source water δ18O, atmospheric vapor δ18O, ambient humidity and leaf temperature must either be constant over time (for tree rings) or between treatments (common garden) or that all relevant fractionation events also influence gs (e.g., vapor pressure deficit [D] differences modify gs through a welldescribed relationship; thus, increasing D leads to enhanced evaporative enrichment in 18O, but it can also reduce gs leading to a concomitant increase in leaf water δ18O, amplifying the effect of D on organic matter δ18O). Documented inter-annual variation in the δ18O of precipitation, soil evaporative enrichment profiles and the fact that trees can tap different sources of water may reduce one’s confidence that constant source water δ18O is an appropriate assumption (Sarris et al. 2013). The paucity of data on δ18O variation in atmospheric vapor Commentary