Fire after the fire: Effects of shrub control versus salvage logging

[originally posted at www.foreststeward.com on May 28, 2010]

Article reviewed: Fuel buildup and potential fire behavior after stand-replacing fires, logging fire-killed trees and herbicide shrub removal in Sierra Nevada Forests

By T.W. McGinnis, J.E. Keeley, S.L. Stephens, and G.B. Roller, published in Forest Ecology and Management 2010 Vol 260 pp 23-35

The plot line: Four areas that burned with intense wildfires in the Sierra Nevada were examined in order to explore salvage logging and herbicide spraying effects on species composition and predicted future fire behavior. The researchers conclude that logging had small effects on species composition and fire behavior, especially when compared to the effects of spraying shrubs with herbicides. As would be expected, herbicide-treated areas had lower amounts of shrubs present and greater amounts of grasses and forbs (including some exotic grasses and forbs). Herbicide-treated areas had lower predicted flame lengths and rates of fire spread, but mortality to small trees was still expected to be high in herbicide-treated areas. In the case of the four fires used in this study, it was post-fire management treatments such as shrub removal, thinning, and pruning (and not salvage logging) that most influenced forest change and future fire behavior following wildfires.

Relevant quote: “Ultimately, the amount of fuel remaining in any given stand after logging was under the control of individual Forest Service managers…”

Relevance to landowners and stakeholders:

The debate continues. Should we do salvage logging after wildfires? This study looks at the issue with respect to the effect of logging on forest structure and composition, but there are of course many other effects that could be considered.

Although this study is limited by a lack of experimental control (they found areas that happened to be treated differently, rather than controlling and assigning treatments experimentally), the stark difference between the effects of logging versus herbicide treatments seemed convincing. It was not the logging, per se, that influenced what plant species were present or how vulnerable the forest was to fire. It was the actions that occurred after the logging that made the difference. In central and southern Sierra Nevada forests, shrub communities profoundly influence how a forest develops following disturbances. It therefore makes sense that management treatments which influence the shrub community (like spraying herbicide) would influence forest development.

There is a need to improve upon this study and conduct a variety of treatments (including controls where nothing is done) in an experimental fashion following wildfires in the Sierra Nevada. Rather than doing nothing because there is uncertainty in what the effects of active management are (after all, there is plenty of uncertainty in the outcome of doing nothing), different alternatives can be tested in order to hone in on preferred treatments for meeting given objectives. This is the essence of active adaptive management.

Relevance to managers:

Disturbances of moderate or high intensities in Sierra Nevada mixed conifer forests tend to initiate a “shrub response.” Shrubs can germinate from dormant seeds or sprout from existing plants to quickly occupy a site and its plentiful resources (light, water, and nutrients). Shrubs can dominate a site for decades to centuries to indefinitely. Shrub removal has been a common and effective treatment for managers aiming to ensure or accelerate the time it takes for the site to be dominated by trees, but there is of course biological and social baggage associated with using herbicides. Rapid tree dominance following fires may not always be an objective, but where it is an objective, it is hard to beat herbicides in terms of treatment effectiveness in meeting that objective. In this study, it was not surprising that spraying shrubs with herbicides reduced shrubs (duh), or that there were more exotics (because there are more of ALL species when resources are plentiful, not just exotics). The more relevant results were the effects of herbicides on the fuel structure.

Having a lot of shrubs creates a certain fuel structure that facilitates a certain type of fire (often a canopy fire), while trading shrubs for trees and grass/forbes (via spraying herbicide) creates a different type of fire (often a surface fire). The researchers predicted that either structure would promote a fire behavior that would kill many of the trees while the trees are small. But eventually big trees will become established (if they aren’t killed by fire) and become more resistant. And the time it takes to grow big trees is shorter when shrubs are controlled. Again, this assumes that tree dominance (as opposed to shrub dominance) is an objective.

For a manager wanting to greatly reduce the probability that a young stand of trees is lost to wildfire, the modeling done in this study actually implies that a relatively intense host of treatments might be necessary to reduce risk to a minimal level. Assuming unlimited resources (impossible, I know), a manager really trying to reduce risk of loss in a young stand of trees might do the following:

• Maintain, via thinning, wide spacing to maximize individual tree growth (and target smaller trees for removal when thinning)

• Reduce or maintain low surface fuels by whole tree harvesting when thinning or by burning (prescribed or piles)

• Reduce exotic and grass understory biomass via either prescribed burns or herbicide application

• Prune up trees as high and as frequently as feasible while avoiding loss of growth from pruning too much

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

The primary limitation is the lack of experimental control. For example, two of the controls had higher pre-fire basal area than the corresponding treated areas. This means that the areas that were logged (treatment areas) had higher tree densities than the areas not logged (i.e. the controls). The authors state the problems with the controls, but then never explain why this was OK in their opinion for the various inferences made or what it might mean for limiting the scope of the study (the area for which they are making inferences appears to be the entire Sierra Nevada).

It is definitely worthwhile to do studies like this that create experiments retrospectively (case studies, in other words), but they are inherently limited when compared to experiments designed before treatments are applied.