Predictability of fine-scale dynamics in a mixed forest stand: The effect of abundance measures

Background/Question/Methods

The commonly used measures of tree abundance, stem density and basal area density, differ in their relative weighting of small and large individuals.  Stem density weights small and large individuals equally (weight proportional to dbh⁰), whereas basal area density weights larger individuals more heavily (weight proportional to dbh²).  While these measures of abundance have simple, tangible interpretations, other weightings of size might be useful.  This study compared a range of size-weighted abundance measures (dbh^x, x from 0 to 3) for their utility in the fine-scale prediction of forest stand dynamics in a 1 ha Pinus/Betula stand in northern Minnesota.  All trees with dbh ≥5 cm were mapped, measured and tagged in 2001, and re-measured in 2011.  The compositional changes of twenty five 10 m × 10 m non-contiguous subplots within the stand were analyzed for this study.  For each abundance measure, vectors of abundance × species were compiled for each plot.  The fine-scale predictability of stand dynamics over the 10 year period was measured by the correlation between: a) dissimilarities among vectors of initial composition in 2001, and b) dissimilarities among vectors of compositional change.  Previous applications of dissimilarity correlation have used basal area density as the measure of abundance.

Results/Conclusions

Abundance measures that weighted larger stems more heavily had higher correlations: r = 0.10 for dbh⁰ (stem density); r = 0.45 for dbh¹; r = 0.78 for dbh² (basal area density); and r = 0.86 for dbh³.  The common choice of basal area density is good for studies of stand dynamics, but measures of abundance that weight size more heavily (e.g. dbh³) may also be useful.  The stand analyzed for this study had minimal stem turnover (mortality or ingrowth) in the larger size classes, so size-weighted abundance measures predominantly reflected the growth of large trees, which has previously been shown to be highly predictable from initial composition.  An obvious next move, then, is to repeat this analysis for stands in which there has been more turnover among large trees.