Abstract
In dealing with predicted changes in environmental conditions outside those experienced today, forest managers and researchers rely on process-based models to inform physiological processes and predict future forest growth responses. The carbon and oxygen isotope ratios of tree-ring cellulose (δ(13) C(cell) , δ(18) O(cell) ) reveal long-term, integrated physiological responses to environmental conditions. We incorporated a submodel of δ(18) O(cell) into the widely used Physiological Principles in Predicting Growth (3-PG) model for the first time, to complement a recently added δ(13) C(cell) submodel. We parameterized the model using previously reported stand characteristics and long-term trajectories of tree-ring growth, δ(13) C(cell) , and δ(18) O(cell) collected from the Metolius AmeriFlux site in central Oregon (upland trees). We then applied the parameterized model to a nearby set of riparian trees to investigate the physiological drivers of differences in observed basal area increment (BAI) and δ(13) C(cell) trajectories between upland and riparian trees. The model showed that greater available soil water and maximum canopy conductance likely explain the greater observed BAI and lower δ(13) C(cell) of riparian trees. Unexpectedly, both observed and simulated δ(18) O(cell) trajectories did not differ between the upland and riparian trees, likely due to similar δ(18) O of source water isotope composition. The δ(18) O(cell) submodel with a Peclet effect improved model estimates of δ(18) O(cell) because its calculation utilizes 3-PG growth and allocation processes. Because simulated stand-level transpiration (E) is used in the δ(18) O submodel, aspects of leaf-level anatomy such as the effective path length for transport of water from the xylem to the sites of evaporation could be estimated.