USING A MECHANISTIC MODEL TO EVALUATE SAMPLING DESIGNS FOR LIGHT TRANSMISSION THROUGH FOREST PLANT CANOPIES

We develop a mechanistic model that predicts instantaneous photosynthe tically active radiation (PAR) to evaluate four spatial and temporal q uestions about measuring light in a plant competition study. Light sam pling schemes are evaluated for a 2-year-old, plant density study with jack pine (Pinus banksiana Lamb.), large-leaved aster (Aster macrophy llus L.), and wild red raspberry (Rubus idaeus L. var. strigosus (Mich x.) Maxim.). The model simulates PAR transmission through a plant cano py using two modules: a radiation module that simulates direct and dif fuse components of global PAR and a distant-dependent vegetation modul e that simulates light interception by individual plants. Calibration and testing of the model using vegetation measurements and a subset of PAR data from the field study revealed that model predictions agreed closely with observed PAR measurements. Means and the standard deviati ons for prediction errors were -0.17% and 6.38% for aster, and 2.02% a nd 6.41% for raspberry, respectively. Using the model, we found that(1 ) positioning a line quantum sensor in diagonal positions on 1 x 1 m s ubplots underestimated average PAR transmission under aster and raspbe rry by 6.25% and 6.47%, respectively, at the ground level, and 7.47% a nd 6.91%, respectively, at the crown level; (2) there was no differenc e in predicted PAR transmission at the ground and crown levels between systematic and random arrangements of aster and raspberry plants in t he density study; (3) instantaneous PAR transmission throughout the da y under a clear sky was highly correlated with the daily average; and (4) the amplitude and standard deviation of predicted PAR transmission were relatively stable for sampling windows that spanned nearly the e ntire daylight period. Using this approach, we demonstrate how mechani stic models can be used to evaluate alternative experimental and sampl ing designs in field studies.