SEARCH

SEARCH BY CITATION

Keywords:

  • competition;
  • confounding effect;
  • neighbourhood analysis;
  • population dynamics;
  • Ripley's K function;
  • logistic regression;
  • spatial pattern;
  • stand mapping

Summary

1  We mapped the locations of live and dead trees in a large forest plot dominated by pioneer Douglas fir (Pseudotsuga menziesii) with an understorey of the invading late-successional species western hemlock (Tsuga heterophylla) and western red cedar (Thuja plicata) on Vancouver Island, British Columbia, Canada, to test for intra- and interspecific density-dependent effects on tree survival.

2  We analysed both the spatial patterning of trees in the plot and the relationships between neighbourhood density and tree survival. We also examined the effects of additional variables (principally elevation) as covariates in our neighbourhood analyses.

3  Both the spatial and initial neighbourhood analyses suggested strong intra- and interspecific density-dependent effects on tree survival. Douglas fir survival was significantly higher in less dense patches of conspecifics and non-random tree death led to regularly spaced survivors, as expected from intraspecific competition. The significantly lower survival of western hemlock in denser patches of Douglas fir and the resulting negative spatial association between surviving trees of these two species were consistent with interspecific competition.

4  However, having controlled for the influence of elevation on tree survival (probably mediated by variation in soil moisture) in neighbourhood analyses, although the survival of the pioneer Douglas fir trees was still subject to strong density-dependent effects, variation in its density in the overstorey no longer appeared to influence the survival of the invading late-successional species. There was, however, evidence for asymmetric interspecific density dependence between the two late-successional species since western hemlock mortality tended to be higher in denser patches of western red cedar.

5  Our results emphasize the importance of considering confounding factors in studies that seek evidence for density dependence.