PARENT SESSION
Contributed Oral Session 127: Forest Structure and Communities
Thursday, August 11, 1:30 PM - 5:00 PM, Meeting Room 513 A, Level 5, Palais des congrès de Montréal

Crown shape and competition for canopy space: a realistic model parameterized for North American tree species.

Purves, Drew^*,1, Pacala, Steve¹, ¹ Dept. EEB, Princeton, NJ, USA

ABSTRACT- In mesic forests, the growth rate of a canopy tree relative to other individuals is determined largely by its exposed crown area (ECA). Therefore, a method for predicting the ECA of trees could go a long way in predicting forest growth dynamics, for example self-thinning, the formation of size-hierarchies, and competition between neighboring trees. At first glance, the development of such a method would appear to be a difficult task, because the size and shape of crowns are obviously species- and size-specific, and are also highly plastic, responding to the identity, size and location of neighboring trees. However, the task is made much easier if we begin by assuming that trees exhibit perfect plasticity in growth, such that the canopy is always filled, in which case the aim is to predict the tessellation of the canopy tree crowns. Furthermore, visual observations of forest canopies suggest that this extreme assumption is close to the truth. We present a simple model that is based around the idea that each tree has a 3-D potential crown shape, which becomes trimmed, through interactions with neighbors, to give a realized crown shape, and hence ECA. The model formulation is essentially identical to TASS, a simulation model used extensively in Canadian forest management (as we found out after model development). We present a simple approximation method that allows the ECAs to be predicted very rapidly, without calculating the exact tessellation. This in turn enables the model to sit inside a hierarchical Bayesian estimation scheme that uses public-access forest inventory data (USDA FHM data) to provide parameter values for each tree species in North America. We present the pictures of potential tree shapes generated by the estimator, and then test the model independently by comparing predicted vs observed crown areas and crown ratios (measurements that were not used in the parameterization in any way). As we show, the model gives reasonable predictions of crown areas, with average r-squared around 0.45 and average RMA slope close to 1. We will then present comparisons of observed vs predicted growth dynamics in stands with different species composition, and discuss some general implications of the relationship between tree shape, competition, and canopy structure.

Key words: Tree, Canopy, Competition, Growth