Charles D. Canham and María Uriarte. 2006. Analysis of neighborhood dynamics of forest ecosystems using likelihood methods and modeling. Ecological Applications 16:62–73.

Appendix B. Results of a reanalysis of patterns of seedling dispersion for five tree species originally reported by LePage et al. (2000).

We reanalyzed patterns of seedling dispersion for five tree species originally reported by LePage et al. (2000). The data are from samples in mapped stands in "partial" canopy sites. See LePage et al. (2000) for field methods and descriptions of the study sites and species. We used likelihood methods to fit two different dispersal functions: a lognormal function:

and an exponential function:

where dbh_i and dist_i are the size and distance to i = 1..n trees within a fixed radius of a seedling quadrat (20 m radius for hemlock and cedar, 30 m for the other three species).  is a normalizer that equals the arcwise integration of the dispersal kernel. We also estimated a set of eight substrate favorability terms as done in the original analysis (see LePage et al. 2000 for details). Maximum-likelihood parameters were estimated using simulated annealing, an algorithm for global optimization. We compared the models using AIC (Table B1):  the best model (lowest AIC) for each species is indicated in bold type. In order to compare our results to the original analyses reported by LePage et al. (2000), we also did a set of analyses with the exponential function, but in which the values of the and   parameters were fixed at 2 and 3, respectively.

TABLE B1.  Parameter estimates, log-likelihood, and AIC for three alternate models for seedling dispersion of five tree species of interior cedar hemlock forests of British Columbia. Data are from LePage et al. (2000).

LITERATURE CITED

LePage, P. T., C. D. Canham, K. D. Coates, and P. Bartemucci. 2000. Seed abundance versus substrate limitation of seedling recruitment in northern temperate forests of British Columbia. Canadian Journal of Forest Research 30:415–427.