Last Friday, the US Senate was recessed
without taking further action
on the administrations forest initiative,
Yet another devastating year of wildfires has
come to a close,
and the only thing we're left with is the memory of the terrifying destruction, and the desolation of the aftermath.
The opposition to the Bush plan was simply too strong, and the plan was far too vague to calm the fears of a massive
plundering of our greatest natural resource.
However, and as I mentioned previously, the Bush plan has merit. Not because of any political affiliation, but because
it's based on the process of forest restoration.
Granted, the Bush plan had some
elements which were not widely well-regarded. It was also clearly oriented to benefit logging companies, although
from his comments, GWB did seem to understand that there's more to restoration than big logging. What really
scared the naturalists and conservationists however, is that forest restoration requires cutting a lot of
trees. When applied to the overgrown and dense forests of the southwest, forest restoration can be a radical
change, in fact. That's because at its core, forest restoration relies upon the simple conclusion
that many dominant forest species naturally tend to grow in a
clumped spatial pattern.
Now, that's not to say that forest restoration isn't also a very complex process. The sheer number of decisions and
considerations involved in any forest restoration can be overwhelming, and tackling them is perhaps best accomplished
as a community - especially when the community is surrounded by the forest which they want to restore. The community
of Flagstaff, Arizona is one such place, which I also mentioned in the
for forest restoration is the one which the Bush administration claimed that Healthy
Forests was based upon. The Flagstaff plan was developed based on research performed at
Northern Arizona University (NAU),
who's campus is located in right in the middle of Flagstaff. The research considers the many and
various aspects of performing a forest restoration, from social to economic issues, to the monitoring of the old-growth
trees, as well as monitoring birds, grass, insects, brush, and every other biological entity to be found in the forest,
before, during, and after the restoration is completed.
While researchers are working to provide reasonable answers to the many open questions which
remain about forest restoration, there are two basic questions which must be asked, and answered, before a single tree
can be removed. Namely,
- What to cut - Based on age, which trees should be removed, and which should remain.
- How to cut it - What do we want the forest to look like, when finished?
As I alluded to in the first paragraph, the fundamental assertion beneath forest restoration is that the forests were
better off before we ever touched them. Therefore, when answering these two questions, we must first establish a
specific date for any given forest, before which there were only natural processes to affect them. After that's done,
we simply need to establish what the forest looked like at that particular time in history, and cut accordingly,
so it seems. However, since very few photographs exist of that time, the restoration artist must use other techniques
to build our picture of the forests of old.
What a forest is supposed to to look like
The statistical analysis of spatial patterns of trees within
the forest is most important when deciding how, why, and where the forest should be restored. Just as the artist
studies previous works, the restorer of the forest must understand how it existed in it's natural state. Specifically,
the first hypothesis which is tested is:
H0: Trees are naturally distributed with complete randomness in the
Several techniques are
utilized to study the species at question in this hypothesis, including examination of
before and after
photographs, and written/spoken historical accounts of the forests of old. The most important technique, however,
involves going out into relatively undisturbed regions of forest, the so-called "true old growth" stands, and then
laying out sample plots, and measuring distances between the trees, tree-stumps,
and large ash-piles (remnants) of trees which pre-date(d) human disturbance and intervention.
Using statistical analysis upon these distance-measurements, researchers extrapolate conclusions about the true
spatial tendencies of a given species of tree.
Studying tree spacing: Theoretical and historical basis
This technique of measuring tree-to-tree distances was studied closely back in 1954, by two researchers named Clark and
Evans1. They proposed a statistical measure for analyzing spatial growth patterns which is still widely
cited today, and based on raw mathematical derivations from
Paul Hertz back in 1909.
Their statistical measure is called Nearest Neighbor analysis5.
The technique employs a result which, when restated for the forestry context, hypothesizes
that the distance from any given tree to it's nearest neighbor is a random variable, R, which is distributed
according to the
distribution, when the trees grow in a completely spatially random (CSR) pattern over the land. That is to say, that
the probability of finding x trees in the circular area proscribed by the tree-to-tree distance, is
p(x) = (d πr2)x/x! * e-d π r2
Where d is the mean density of the trees per area, and r is the distance to nearest neighbor. . When
considering nearest neighbor distance, we're letting x = 0 (e.g. - the probability of find no trees in the
given area), so the discrete bits reduce to 1, and we're left with the cumulative distribution function of R in
R = 1 - e-d π r2
In this form, the Poisson distribution easily yields a
probability density function (pdf), via
differentiating with respect to r, which is
It's worth pointing out that this pdf is actually a form of the
random variable, a widely used density in the fields of failure analysis and QA. The
or mean of R, is
E(r) = 1/2 * sqrt(d)
with the variance and standard deviation:
σ2r = (4-π)/(4πNd), and
σr = 0.26136/sqrt(Nd)
where N is the number of measurements taken. Now, that sort of makes sense. It says that the expected value
of the distance to nearest neighbor in a CSR population is the inverse of twice the density's root. Thus, to test for
differences in the actual value of R versus the expected value, Clark and Evans proposed the following ratio to
be standard normal:
z = (rA - rE) / σrE
and submitted that rejection on the negative side implied aggregation, while rejection on the positive side implied
evenly-spaced trees. They also showed that R is maximized at about 2.14, when the trees (or any other
measured species) are spatially distributed in a perfectly hexagonal pattern, on a unit area.
That's an elegant and generally well-understood bit of math, to back up the theory of nearest-neighbor statistical
testing. Unfortunately, and despite the beauty of their result, it holds only when the area in question is
unbounded. (infinite area). Additionally, in order to apply this measure, an estimate of the density must be had in
When measuring nearest neighbor distances on a bounded sample plot, as is usually the case, a bias is introduced in the
statistic which is called called "edge-effect". Edge effect leads to under-estimation of the true value of R,
because the nearest neighbor for any given tree may well lie outside of the sample area. Obtaining a proper estimate of
the true density for the species in question in advance can also be quite challenging.
Since Clark and Evan's paper,
a cascade of other works has evolved which cite it, and then provide various techniques which compensate for
edge-effect, and try to work around the density-requirement. Most researchers who utilize the nearest-neighbor
statistic technique these days seem to have settled upon an estimator introduced in 1982, and called
Ripley's K statistic,
to answer H0 above. Ripley's tool examines neighbors out to the kth nearest neighbor,
and uses powerful software to provide the result from the neighbor-distance measurements in a form that can be used to
again reject the above hypothesis, and additionally provide a good estimate of the actual nature of the clumping
tendency exhibited by the species in question.
Translating the theoretical result into a general rule for thinning
In the early research and testing phases of the Flagstaff plan, the scientists at
NAU/Ecological Restoration Institute (ERI)
used Ripley's K statistic to reject H0 (above) for the Ponderosa pine tree, and eventually propose
an initial clump radius of 30 feet. Formally, the foresters (taggers) were told to find every existing old-growth
tree, or remnant, mark it, and also mark any other old growth, as well as some "medium" growth replacement trees within
a thirty foot (~10m) radius. The logging crews then came in and removed trees outside of the radius, via taking all
unmarked trees, and a clumped (or aggregated) spatial pattern of trees was finally restored in the forest.
It's important to note, that every single old growth tree remained in the restoration area, along with it's new
Since the early tests were concluded for the Flagstaff experiments back in the late 90's, the clump radius has been
increased in length, first to 60 feet, and most recently to 90 feet. There were objections to the patchy look of the
forest when the natural radius was used, even though the statistics are sound for that value. Increasing the radius,
and thereby leaving less open space, gives us the feeling of protection and solitude that we crave from the forest.
Increasing clump size also quite naturally implies easier spread of wildfire (crown fires), and limits the protection
afforded for the long-term survival of the forest by establishing the relatively distant clumps, and low undergrowth
which was prevalent long ago.
In any case, now we have a reasonable plan for how to thin the forest, but we still don't know exactly
what to cut. We still need a guideline for deciding what actually is an old-growth remnant, around which
we will reestablish the clumps.
How old is old-growth?
Old-growth is a very subjective term. It's thrown around by the media, and well-meaning conservationists as though it
were some magical aura, exuding from every tree which stands taller than 10 meters or so. Granted, a tree takes a
long time to grow, but old-growth as a purely scientific descriptor means something more than any given tree with a
trunk thicker than your leg. To complicate matters further, there's the fact that it's impossible to positively
determine the age of a tree, without cutting it down. So, once again, statistical analysis must be used to provide
the forester with an estimator he needs to restore the forest. Before we take a look at how that's done, lets consider
the seminal work of Swetnam and Brown13.
Swetnam and Brown produced their paper in 1992,
which is also cited by the restoration researchers at NAU, and which analyzed the oldest stands of Ponderosa pine
(among other conifers) in the southwest. While they allow for the claims that forest age may not be a discrete variable,
they also emphasize several important points:
- True old-growth trees have high scientific value.
- True old-growth trees tend to grow in difficult-to-reach locations (or else they likely would not have lived this
- True old growth trees are often affected by environmental conditions which cause a disparity in the relationship
between their age and their diameter.
- True old-growth trees in the Ponderosa pine forests of the southwest were typically around in
Daniel Boone's time,
and even earlier. (~ 1700AD).
Examination of Swetnam's paper will also show you a map which makes it easy to see that true old-growth stands
really are frightfully rare. Perhaps they should be among the first, therefore, to be restored. In any case, when
considering old-growth for the purpose of forest restoration, we define the term in a much broader sense.
Statistically speaking, defining the relative age of a particular stand of forest is more complex, and the techniques
utilized do not have the clean and simple mathematics that we saw for R. For the purposes of forest
restoration, we're primarily interested in a single minimum age requirement. As we've seen, any tree meeting that
requirement, along with some of it's neighbors (who need not be as old), are left untouched by the restoration artist.
In the southwest, the researchers at NAU have determined the minimum age requirement, also called the
to be around 1880. That is, the date before which non-native settlers began affecting the forests in the southwest.
The establishment of the pre-settlement date itself is relatively academic, and considers domesticated animal grazing
disturbances, as well as aggressive harvesting of the early 1900s, mass replantings, and the interruption of natural
As mentioned, the only completely reliable way to date a tree is to cut it down, the next most reliable technique is to
drill a core sample, and count rings. In the process of forest restoration, and for the sake of simplicity and
scalability for many workers, old-growth is defined by thickness of the trunk. Depending upon which plan is agreed upon
and utilized, an old-growth Ponderosa pine tree in the southwest will have a trunk diameter of somewhere between 12-16"
(30-40 cm), which corresponds to an age of somewhere between 80-120 years. When a tree, or remnant of this diameter or
greater is found, it is treated as old-growth, and the agreed radius is proscribed about it to determine which trees
should be retained to establish it's new "clump".
Since the chosen diameter specification is applied universally on a given restoration plot, it is also a mean value. It
is a very mean value, in fact. There's typically more disagreement and contention about the chosen old-growth diameter
than any other variable which is considered in forest restoration. The important point, when deciding
minimum trunk diameter, is that it's not just a few bucks more (or less) for the lumber sale, it's about giving the
forest the best chance to reestablish natural growth.
Like any good creative process, forest restoration also makes a big mess. Some of the mess is reusable, but much of it
is actually just fertilizer - fertilizer that is loaded with nitrogen, and loves to burn. This by-product of forest
restoration is also called
biomass. Biomass, essentially, is everything alive in the forest, except for the creatures. It includes
the trees, brush and shrubs, grasses, roots, and other organisms and matter lying on the floor, including the remains of
any of the above. Dealing with biomass is also a tenuous and difficult questions in the process of
restoration. The first, and most fundamental question which is asked by restoration research about biomass is
How much biomass existed at the pre-settlement date?
This is an important question because it represents the amount which should remain in the forest when the
process of restoration is complete. The various experiments in forest restoration which are taking place have little to
report, in the way of statistical tools, for estimating this value. Fulé and Covington et. al. utilize
"unpublished" figures to establish their target for pre-settlement biomass, and purport to acheive that target via
several means in their implementation, including chipping and spreading (and compacting) of smaller trees and shrubs
which must be cut. In spite of their best efforts to utilize the excess biomass from the restoration, they have a lot
Thinning resulted in the removal of a total of 5,500 bd.ft./ac (3,700 bd.ft./ac of 9-16 in dbh trees, 1,800 bd.ft./ ac
of 5-8.9 in dbh trees). Most of the smaller diameter trees (629 trees per acre in the 1-4.9 inch dbh class) were
utilized as latillas for adobe home construction. A major problem in utilization was what to do with the 37 tons per
acre of thinning slash. Because there was no market for this material, it was hauled (70-80,18-wheel dump truck loads)
to a borrow pit and burned. - [Covington et al.]
In every case of restoration, the restorer should have a plan for what to do with the biomass which cannot remain in
the forest, either due to exceeding the biomass target, or due to it's crown-fire enabling
properties as a so-called ladder fuel. I pointed out in the previous article some of the alternatives being
considered for this problem, including utilization as a power-generation fuel. There are many other ideas, and they
seem to be open to new ideas, as well.
Biomass is, as one might guess, is also studied using statistical tools. Biomass is also hypothesized to be
distributed spatially over the forest, just like tree-location, and age. It has averages and variance that may even be
predictable, someday. Much important work in studying biomass is being done via satellite-based remote
sensing4, which as we know, continues to improve. In fact, researchers seem quite confident that they can
learn just about anything they want to know about the forest from satellite imagery. They're now able to analyze most
visible and non-visible bands for information about the forest. It's unclear to me whether, or if, it will ever be
possible to actually count trees, much less measure them, via satellite imagery. At least, not via any publicly
Other Considerations in Forest Restoration
There are currently about a
half-dozen large restoration areas around the city of Flagstaff, within the so-called Urban Interface of the
Coconino forest. Another lies slightly north of town in an area called Mount Trumbull. The Mount Trumbull area is
more situated in true wilderness than the other plots around town, and therefore provides a test of open-wilderness
implementation. While the urban interface is considered a higher priority than open-wilderness for restoration
projects, it's also important not to simply forget about the open wilderness, and the danger it faces, if action is not
There are plenty of other interesting research projects going on in and around Flagstaff. While many experiments are
still being actively carried out, some results are now becoming available. Of particular concern to many naturalists
considering forest restoration is the affect which it may have upon wildlife. While researching this article, I
had an exchange with a biologist who studied Western Bluebirds in and around the restored areas of
Flagstaff. He gave me permission to repost this abstract:
We examined the effects of presettlement forest restoration treatments on the nesting success of Western Bluebirds in
ponderosa pine forests of northwestern Arizona, U.S.A. From 1998 to 2001 we monitored 97 active Western Bluebird nests,
41 in current-condition untreated forest and 56 in restoration-treated forest. We found no effect of restoration
treatments on clutch size and little effect on the number of nestlings per nest. However, in treated forest stands
number of fledglings per nest averaged 1.6 times greater, and probability of a nest surviving to successfully fledge at
least one young was up to 4.2 times greater than in untreated forest. Probability of a nest succeeding averaged 0.39 +/-
0.11 (SE) and 0.75 +/- 0.06 from 1999 to 2001 in untreated and treated forests, respectively. In addition, in treated
forest, average number of nests infested with the blowfly parasite Protocalliphora sialia was up to 4.3 times greater,
and number of parasites per fledgling was up to 10.7 times greater than in untreated forest. Overall, the data suggest
that in treated forest Western Bluebirds have a higher probability of successfully fledging young, but they are at
greater risk of parasitic infestations, of which the ultimate effects on post-fledging survival are unknown. -
[Germaine and Germaine]
Insects in general, as indicated by their effect on the bluebirds, seem to thrive in the re-opened forest. One of the
accomplishments reported by Covington, Fulé, Moore and the rest of the researchers in their Progress
Report12 earlier this year is that butterfly population was increased in abundance and species-richness in
restored areas, possibly due to the additional sunlight. Their entire report is worth reading, in fact. It gives
synopsys of many of the ongoing experiments which are measuring everything from grazing effects on restored forest, to
the human dimension of ecological restoration, even among the native Indian tribes of northern Arizona.
As mentioned previously, forest restoration in the southwest relies on the fact that the native tree species there tend
to clump together. While this result holds for other conifers and tree species, it may not always be the case. If a
species is found which demonstrates other spatial tendencies in its natural environment, then restoration for that
forest might utilize a different marking/thinning technique. Calculation and study of each individual forest, along
with the tendencies of its dominant (and lesser) species, should be a part of any good plan.
The cost to the public of any restoration is also a consideration. As mentioned in the
in this series, some estimates for restoration go as high as $1600/acre (~$4000/ha). That figure, it seems to me,
must be at the very high end of the scale. Fulé, et al.8 found that applying full restoration in a
dense Ponderosa pine forest had an associated cost of about $300/acre ($748/ha), for example.
Naturally, the final cost is dependent on many factors, including how many trees are taken, but also how high a price
the Forest Service is able to get for those trees. There's also a significant cost associated with cleaning up the mess,
as we've seen, at least until better ideas are put forward for managing slash.
Forest restoration is a complex process. This note only considers two or three of the major issues, and there
are hundreds more. Right now, in many ways, restoration scientists are still in the early phases of learning about
these, but I believe that the scientific basis is sound. Naturally, it'll be at least a dozen years before we know if
it truly is as good as it seems, although the early results look promising.
Forest restoration is not a clearcut by any means, neither is it a sustainable-forestry technique. Reducing the
population (logging/thinning) is important in restoration, but only insofar as it leaves the forest in a state which
shall not be altered further (protection). Ideally, after a forest is restored, only natural processes (along
with controlled burning) should be required to maintain its new, and natural state. In practice this doesn't always
hold, but as insight is gained to the craft of restoration, and the trees mature, the odds will improve.
Therefore, it's important to proceed with care, when restoring a forest. It's also important to proceed with some
haste. You see, as I pointed out previously, many forests are facing a crisis larger than mismanagement, it's a crisis
of drought. A crisis which we've
brought upon ourselves, perhaps, or perhaps it is something worse. Whatever the reason, the forests of the western US
are threatened by more than just greedy lumber companies, and politicians looking for votes. It's time to take action,
and give our forests back a chance to fight for themselves, with their own natural defenses. It's not time to
thin, it's time to restore.
- Clark and Evans. 1954. Distance to nearest neighbor
as a measure of spatial relationships in populations. Ecology 35: 445-453
- Covington, Niering, Starkey, Walker. 1998. Ecosystem Restoration and Management:
- Covington, Fulé, Smith, Springer, Heinlein, Huisinga, and Moore 2001. Comparing ecological restoration
alternatives: Grand Canyon, Arizona. Forest Ecology and Management 170:19-41.
- Davis, Melack, Day and Wang 1997. Biomass
Modeling of the Ponderosa Pine Forests of Western North America with SIR-C/X-SAR for Input to Ecosystem Models
- Germaine and Germaine, 2002. Forest restoration treatment effects on the nesting success of Western Bluebirds RESTORATION ECOLOGY
- Dixon, Phillip. 2001. Nearest Neighbor Methods
- Fulé and Covington. 1998. Spatial patterns of Mexican
pine-oak forests under different recent fire regimes. Plant Ecology 134:197-209.
- Fulé, Covington, Smith, Springer, Heinlein, Huisinga, and Moore 2002. Testing ecological restoration
alternatives: Grand Canyon, Arizona. Forest Ecology and Management 170:19-41.
- Fulé, Covington, Hart and Weaver 2001. Modeling ecological restoration effects on
ponderosa pine forest structure Restoration Ecology 9(4): 421-431.
- Mast, Fulé, Moore, Covington, and Waltz. 1999. Restoration of presettlement age
structure of an Arizona ponderosa pine forest. Ecological Applications 9(1):228-239.
- Research Guide for the Flagstaff Plan
- Southwest Fire Initiative Update 2002
- Swetnam, T. W., and P. M. Brown. 1992. Oldest
known conifers in the Southwestern United States: Temporal and spatial patterns of maximum age. In: M. R. Kauffman,
W. H. Moir, and R. L. Bassett, tech. coords., Old Growth Forests in the Southwest an Rocky Mountain Regions, Proceedings
of a Workshop, March 9-13, 1992, Portal, Arizona. USDA Forest Service General Technical Report RM-213:24-38.
- Previous articles in this series: