Premium
MULTI‐MODEL ANALYSIS OF TREE COMPETITION ALONG ENVIRONMENTAL GRADIENTS IN SOUTHERN NEW ENGLAND FORESTS
Author(s) -
Papaik Michael J.,
Canham Charles D.
Publication year - 2006
Publication title -
ecological applications
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.864
H-Index - 213
eISSN - 1939-5582
pISSN - 1051-0761
DOI - 10.1890/1051-0761(2006)016[1880:maotca]2.0.co;2
Subject(s) - relative species abundance , ordination , basal area , ecology , competition (biology) , abundance (ecology) , akaike information criterion , interspecific competition , canopy , environmental gradient , biology , habitat , statistics , mathematics
Robust predictions of competitive interactions among canopy trees and variation in tree growth along environmental gradients represent key challenges for the management of mixed‐species, uneven‐aged forests. We analyzed the effects of competition on tree growth along environmental gradients for eight of the most common tree species in southern New England and southeastern New York using forest inventory and analysis (FIA) data, information theoretic decision criteria, and multi‐model inference to evaluate models. The suite of models estimated growth of individual trees as a species‐specific function of average potential diameter growth, tree diameter at breast height, local environmental conditions, and crowding by neighboring trees. We used ordination based on the relative basal area of species to generate a measure of site conditions in each plot. Two ordination axes were consistent with variation in species abundance along moisture and fertility gradients. Estimated potential growth varied along at least one of these axes for six of the eight species; peak relative abundance of less shade‐tolerant species was in all cases displaced away from sites where they showed maximum potential growth. Our crowding functions estimate the strength of competitive effects of neighbors; only one species showed support for the hypothesis that all species of competitors have equivalent effects on growth. The relative weight of evidence (Akaike weights) for the best models varied from a low of 0.207 for Fraxinus americana to 0.747 for Quercus rubra . In such cases, model averaging provides a more robust platform for prediction than that based solely on the best model. We show that predictions based on the selected best models dramatically overestimated differences between species relative to predictions based on the averaged set of models.