Overstory–Understory Relationships in Mature Forests of Western Washington

(click here for full paper)

Donald McKenzie*, Charles B. Halpern, and Cara R. Nelson College of Forest Resources Box 352100 University of Washington Seattle, WA 98195-2100 donaldmckenzie@fs.fed.us *Current address: Pacific Wildland Fire Sciences Lab, USDA Forest Service, 400 N. 34th Street, Suite 201, Seattle, WA 98103.

Understanding the relationships between forest overstory and understory communities is essential for predicting changes in the abundance and distribution of understory plants through successional time and in response to forest management. In this study, we used correlation analysis, multiple regression, and non-parametric models to explore relationships between overstory characteristics in mature coniferous forests and the abundance of species in the herb and shrub layers. Data used in this study were obtained from pre-treatment measurements of vegetation plots (818 in total) in the four DEMO blocks of western Washington (see Study Areas).

In conventional analyses of "mean" response, direct interactions between overstory and understory may not be detectable through variation induced by numerous other factors (e.g., stand history, site environment). Thus, we employed an additional approach, in which we estimated "maximum" responses. This approach is analogous to that used to describe biomass-density relationships in pure, even-aged forests or tree-density maxima in uneven-aged, mixed-species stands. Models of maximum response are useful for quantifying thresholds or limits, and the extent to which a predictor (e.g., tree cover) constrains a particular response variable (e.g., herb cover) within the context of other influences.

Our goal was to identify those dependent and independent variables that have strong and predictable relationships and to interpret these relationships in light of our understanding of plant life histories, environmental influences, and the successional development of forests. Given the large number of species-level comparisons possible, we restricted our analyses to the responses of broad groups of plants that occupy distinct vegetation layers or that share common successional patterns or responses to disturbance.

Overstory variables explained >50% of the variation in the mean response of total shrub cover and ca. 50% of the variation in cover of vine maple (the most common shrub) and late-seral herbs (species that reach their greatest abundance in old forests). Stronger relationships (80-90% variance explained) were found between overstory variables and the maximum cover of total shrubs, vine maple, herbs, and each of three functional groups of herbaceous species (Figures 1 and 2). We interpreted these empirical relationships to represent both the direct effects of resource limitations and “time-dependent” responses for which overstory characteristics (e.g., tree size) may be surrogates.

Figure 1.  Maximum abundance models of (a) tall shrub cover as a linear function of stand density index (SDI), and (b) vine maple cover as functions of SDI.  In (b), straight line = fitted values of linear regression, curved line = fitted values of a LOESS model.

Figure 2. Maximum abundance models (3D LOESS) of three categories of herb cover as a function of stand density index (SDI) and quadratic mean diameter (QMD). Numbers on the contour lines represent the predicted maximum of percent cover associated with that line. For (a) total herb cover, and (b) cover of “dominant” herbs, contours suggest a unimodal distribution of cover maxima in the space of the two predictors. Maximum cover is highest at intermediate values of the predictors and lower at their extremes. For (c) cover of “late-seral” herbs, contours suggest a monotonic increase of cover maxima as SDI decreases and QMD increases.