Co-ordinating Editor: Geoffrey Henebry
Modeling the occurrence of 15 coniferous tree species throughout the Pacific Northwest of North America using a hybrid approach of a generic process-based growth model and decision tree analysis
Version of Record online: 14 MAR 2011
© 2011 International Association for Vegetation Science
Applied Vegetation Science
Volume 14, Issue 3, pages 402–414, August 2011
How to Cite
Coops, N. C., Waring, R. H., Beier, C., Roy-Jauvin, R. and Wang, T. (2011), Modeling the occurrence of 15 coniferous tree species throughout the Pacific Northwest of North America using a hybrid approach of a generic process-based growth model and decision tree analysis. Applied Vegetation Science, 14: 402–414. doi: 10.1111/j.1654-109X.2011.01125.x
Coops, N.C. (corresponding author, firstname.lastname@example.org): Department of Forest Resource Management, 2424 Main Mall, University of British Columbia, Vancouver, Canada V6T 1Z4 Waring, R.H. (email@example.com): College of Forestry, Oregon State University, Corvallis, Oregon 97331, USA Beier, C. (firstname.lastname@example.org) & Roy-Jauvin, R. (email@example.com): Department of Forest Resource Management, 2424 Main Mall. University of British Columbia, Vancouver, Canada V6T 1Z4 Wang, T. (firstname.lastname@example.org): Centre for Forest Conservation Genetics, Department of Forest Sciences, 2424 Main Mall. University of British Columbia, Vancouver, Canada V6T 1Z4
- Issue online: 4 JUL 2011
- Version of Record online: 14 MAR 2011
- Received 15 November 2010, Accepted 21 January 2011
- 3-PG model;
- Climate analysis;
- Decision tree analysis;
- Species geographical distribution
Question: Can we interpret how climatic variation limits photosynthesis and growth for one widely distributed species, and then relate these responses to model the geographic distributions of other species?
Location: The forested region of the Pacific Northwest, United States and Canada.
Methods: We first mapped monthly climatic data, averaged for the period 1950 to 1975 at 1 km resolution across the region. The recorded presence and absence of 15 native tree species were next mapped at 1 km resolution from data acquired on 22 771 field survey plots. To establish seasonal limits on photosynthesis and water use, a process-based growth model (3-PG, Physiological Processes to Predict Growth) was parameterized for Douglas-fir (Pseudotsuga menziesii), one of the most widely distributed species in the region. Automated decision tree analyses were used to predict the distribution of different species by creating a suite of rules associated with the relative constraints that soil drought, atmospheric humidity deficits, suboptimal and subfreezing temperatures would impose on the growth of Douglas-fir.
Results: The 3-PG process-based modeling approach, combined with automated decision tree analyses, predicted presence and absence of 15 conifers on field survey plots with an average accuracy of 82±12%. Predictive models of current distribution for each species differed in the number of, order in, and physiological thresholds selected. A deficit in the soil water balance, followed by departures from optimum temperatures in the summer were the two most important variables selected in predicting species distributions.
Conclusions: Although empirical models using different sampling techniques and statistical analyses may be more accurate in predicting current distribution of species, the hybrid approach presented in this paper provides a greater mechanistic understanding of the limits to growth and tree distributions. These attributes of process-based models make them particularly useful in designing mitigating strategies to projected changes in climate.