Using structural equation modeling to evaluate direct and indirect effects on forest regeneration in Pennsylvania

Background/Question/Methods

Forests in North America have experienced changes in community composition over the last century partly due to historically high ungulate densities. Our objective was to determine whether there was support for direct effects (i.e. deer density, canopy structure, herbaceous competition) or indirect effects (i.e. landscape structure, canopy structure and deer density) on woody regeneration in areas of timber harvest. We conducted pellet group counts and vegetation surveys within 24 blocks (2.59km²) distributed throughout the study area (~10,000km²). Each block received a shelterwood timber harvest and an herbicide application, used to control fern understory, within the past 2 years to encourage regeneration. We estimated deer densities with annual pellet surveys along five 1 mile transects spanning each block in mid-Spring. Vegetation surveys sampled seedling density, herbaceous richness, tree overstory and herbaceous cover within 30 permanent subplots within each plot. Canopy openness was calculated using hemispherical photos taken at each of the 30 subplots. We included herbaceous cover as direct effect in the model since herbaceous cover will compete for the same resources as woody regeneration.  We used structural equation models to describe both indirect and direct effects on woody seedling regeneration.

Results/Conclusions

We calculated the standardized regression coefficients, which reflect both the strength of the relationship, and can be compared across all variables within the model. There was a positive direct effect of canopy openness on woody seedling per m² (γ = 29.439), with an indirect effect of 1.9. There was a positive direct effect of deer density on woody seedlings per m² (γ = 0.293), with an indirect effect of 0.023. Both canopy openness and deer density positively affected herbaceous cover (γ = 11.937, γ = 0.146). There was a positive effect of herbaceous cover on seedlings per m² (β = 0.160). The relationship between canopy openness and seedling regeneration concurs with the literature while the positive effect of deer and seedling regeneration is less well explained. The relationship could be due to foraging behavior, with an inverse relationship with the number of seedlings attracting deer to the site. We expect additional years of monitoring and a new LIDAR dataset will help more fully develop these relationships in the future. The use of structural equation modeling for these complex ecological interactions will help create a more comprehensive explanation of how ungulates impact forest composition.