March 5, 2010

Using chainsaws and matches to grow big trees

Article reviewed:  Thinning and burning in dry coniferous forests of the western United States: Effectiveness in altering diameter distributions

By A. Youngblood, published in Forest Science, Vol 56 pp. 46-59

The plot line: This study evaluated the degree to which thinning, burning, and thinning+burning treatments changed the proportions of big versus small trees. It integrated across several similar studies in the western U.S. to derive common trends (similar to last week’s meta-analysis, but without that study’s inherent limitations). Although the results could have been easily predicted by most foresters, it documented that thinning and understory burning treatments that are designed to reduce small trees relative to big trees often do reduce small trees relative to big trees (surprise!). More importantly (but also predictably), it documented that in general, thinning+burn treatments were more effective in creating bigger shifts from small to big trees (at least in the short-term). The author puts fire hazard reduction treatments in the context of achieving multiple objectives, particularly enhancing old forest-like structure and creating forests that will be resilient to climate change.

Relevant quote: “Active treatments may be steps in creating structures that are better able to resist the influence of climate change and are resilient to change and enable forests to respond to change.”  

Relevance to landowners and stakeholders:

One of the values of studies like this is that fact that they take emphasis away from fixed diameter targets (e.g. “thou shalt not harvest trees less than 24” diameter”), and place emphasis on the relative number of big versus small trees. The physical limits on the range of tree sizes and tree density change from location to location on both small and large scales. Therefore,

It does not make sense to establish widely-applying diameter thresholds if we want forest managers to meet objectives of landowners.

It is not difficult to mathematically describe forest structure in ways other than a simple diameter target or threshold. As the author of this study points out, there are several ways of measuring diameter distributions and the degree to which forests can be achieving desired distributions (i.e. structures). The focus should be on the shape of the distribution (i.e. ratio of big trees to small trees), not just average tree size.

Relevance to managers:

Very commonly, managers use diameter distributions to describe forest structure (despite an assertion in the paper that they are not common). Often, however, the analysis of a distribution goes no further than a visual description of the distribution (i.e. “bell-shaped” or “inverse-J shaped”). Sometimes, this might be enough. But in the face of stakeholders being very focused on tree size and stand structure, it behooves managers to seek ways to quantify diameter distributions. In this paper, a relatively uncomplicated way of describing the distribution is used. The 10th and 90th percentiles are estimated as a way to quantify the trend toward big-tree dominance that occurred after the treatments, and percentiles are intuitive to anyone who has taken the SAT or anyone who has compared how tall their child is compared to the "national average" (mine is 90th!). Excel software can compute percentiles given a list of trees and their diameters. It can also measure “skewness” and “kurtosis,” which are informative measurements of distributions. Standard deviation of diameters is also an informative measure of variability. All of these are relatively easy to compute with standard software. The harder part may getting the data, which managers have to commit to collecting before and after treatments.

Critique and/or limitations (there’s always something, no matter how good the article is) for the pedants:

There was never any discussion of what a desirable distribution would look like. The focus of the study was to determine if the fuel treatments were effective in creating structures that “might persist in late-successional forests.” But the extent of the analysis and interpretation was simply that the treatments modified structure and that the ratio of big trees to small trees generally went up. But did the ratio go up enough to meet objectives or form a new distribution that is likely to “persist in late-successional forests?”

One limitation is the fact that there was likely delayed mortality in the burn and thin+burn treatments that was not captured. The post-burn measurements occurred within a year after the burning was done (sometimes much less than a year). Many big trees may have died 1-3 years following the fire and the related insect activity, but they would have been measured as live trees. This could be significant, especially if the thin+burn treatments had relatively high burn intensities that killed some larger trees.

It seems like there should be six or seven other authors on this paper, given that raw data from seven sites collected by seven different researchers were gathered into a database and then analyzed.