Compositional changes in the overstory and understory layers of forests are generally expected to be coupled due to the strong influence of the overstory on the environment of the understory.  We resampled 66 northern forest sites that were originally sampled in the early 1950’s for forest overstory and understory composition.  We used Mantel tests and regression analyses of dissimilarity matrices to examine potential relationships between forest layers.

Results/Conclusions  

In the first and second sampling period, overstory and understory composition were highly correlated (Mantel statistics 0.573 for first period and 0.475 for second period,  both significant at p < 0.005), and both layers responded to similar environmental gradients associated with soil texture and chemistry.  Both overstory and understory change are predictable from initial composition (dissimilarity correlations: 0.282 for overstory and 0.724 for understory, both significant by cross-validation).  However, understory compositional change was uncorrelated with overstory compositional change (Mantel statistic 0.0014, n.s.).  While the understory and overstory at these sites respond to similar environmental gradients, their dynamics over 50 years are decoupled for the range of overstory change observed in this study.