More Thoughts on Tree Mortality Study

I got a copy of the Science article by Van Mantgem et. al. on tree mortality referred to in my previous post here.  I have not done a comprehensive review, but I have now read it and its supplements and have a few immediate reactions.

This article struck me as an absolutely classic academic study, for the following reason:   The study can be broken up into two parts – measurement of a natural phenomenon and possible explanations for the measurements. The meat of the effort and real work is in the first part, the measurement of tree mortality, with very weak work on the second part, on the links to global warming.  Many academic studies are guilty of this to some extent.  I once had a professor tell me every study was a year of intensive data gathering and analysis followed by 2 hours of a group of grad students trying to brainstorm causes and implications, the more exciting the better.  Unfortunately, the press releases and media attention in climate tend to focus on this hypothesizing as if it had as much credibility as the actual data analysis.  Let me be specific.

The first part of the study, the measurement of tree die-off rates, appears to be where the bulk of the work is, and their findings seem fairly reasonable — that tree die-off rates seem to have gone up over the last several decades in the western US, and that this die-off seems to be consistent across geography, tree size, and tree type.   My only complaint is that their data shows a pretty clear relationship between study plot size and measured mortality.   Most of the measured tree mortality is in plots 1 hectare or less (about 2.5 acres, or the size of a large suburban home lot).  There is not nearly as much mortality in the larger study plots — I would have liked to see the authors address this issue as a possible methodological problem.

Anyway, the finding of large and spatially diverse increases in mortality of trees is an important finding, and one for which the authors should feel proud to identify.  The second part of the study, the hypothesized causes of this mortality, is far far weaker than the first, though this is not atypical of academic studies.  Remember, in the press summaries, the authors claimed that global warming had to be the cause because they had eliminated everything else it could possibly be.  So here is what their Science article mentions that they considered and rejected as possible causes:

  • Changes in forest density and fire exclusion policies
  • Old trees falling and crushing new trees
  • Ozone levels (they claim they look at “pollution” but ozone is the only chemical discussed)
  • Activity of fungal pathogen, Cronartium ribicola, in certain pines
  • Forest fragmentation

Wow, well that certainly seems comprehensive.  Can’t think of a single other thing that could be causing it.  By the way, the last one is interesting to me, because I suggested forest fragmentation and micro-climate issues in my first post.  So, just to give you an idea of the kind of scholarship that passes peer review, let’s see how they tested for forest fragmentation:  they compared mortality of trees inside national parks vs. mortality of trees outside of national parks.  The logic is that National Park trees would see less fragmentation over time since they are protected from logging, but that of course is a supposition.

This is really weak.  I guess it’s not a bad test if you had to come up with such a test in an afternoon without the time to do any extra work, but it is a very course macro test of a very micro problem.  For example, the top of Kilimanjaro is protected as  a National Park, but evidence is pretty strong that snow on the mountain is being reduced by land-use-related changes in precipitation and local climate due to logging outside the national park.

A lot of folks in the comments of the last post mentioned, reasonably, the massive infestations of western pine bark beetles.  The only mention of the  bark beetle infestations was, interestingly, in their last paragraph, where they said:

First, increasing mortality rates could presage substantial changes in forest structure, composition, and function (7, 25), and in some cases could be symptomatic of forests that are stressed and vulnerable to abrupt dieback (5). Indeed, since their most recent censuses, several of our plots in the interior region experienced greatly accelerated mortality due to bark beetle outbreaks, and in some cases nearly complete mortality of large trees

I guess that is a handy way to deal with an exogenous factor you don’t want to admit drove some of your observations – just reverse the causality.  So now mortality is not caused in part by beetles, beetles are caused by mortality!

By the way, before I head into temperature, I had a question for those of you who may know trees better than I.  Do trees have demographics and generations, like human populations?  For example, we expect a rise in mortality among humans over the next 30 years because there was a spike in birth rates 50 years ago.  Do forests have similar effects?  It struck me that humans cleared a lot of western forests from 1860-1920, and since then the total forested area in the US has expanded.  Is there some sort of baby boomer generation of trees born around 1900 that are now dying off?

Anyway, on to temperature.   Here is the key statement from the Science article:

We suggest that regional warming may be the dominant contributor to the increases in tree mortality rates. From the 1970s to 2006 (the period including the bulk of our data; table S1), the mean annual temperature of the western United States increased at a rate of 0.3° to 0.4°C decade−1, even approaching 0.5°C decade−1 at the higher elevations typically occupied by forests (18). This regional warming has contributed to widespread hydrologic changes, such as declining fraction of precipitation falling as snow (19), declining snowpack water content (20), earlier spring snowmelt and runoff (21), and a consequent lengthening of the summer drought (22). Specific to our study sites, mean annual precipitation showed no directional trend over the study period (P = 0.62, LMM), whereas both mean annual temperature and climatic water deficit (annual evaporative demand that exceeds available water) increased significantly (P < 0.0001, LMM) (10). Furthermore, temperature and water deficit were positively correlated with tree mortality rates (P ≤ 0.0066, GNMM; table S4).

The footnotes reference that the temperature and water correlations are in the supplementary online material, but I have access to that material and there is nothing there.  I may be unfair here, but it really looks to me like some guys did some nice work on tree mortality, couldn’t get it published, and then tacked on some stuff about global warming to increase the interest in it.   Note that Science recognizes what the study is about, when it titles the article “Widespread Increase of Tree Mortality Rates in the Western United States,” without mention of global warming.  But when it moves to the MSM, it is about global warming, despite the fact that none of the warming and drought data and regressions are considered important enough or persuasive enough to make the article or even the supplementary material.

OK, if this paragraph is all we have, what can we learn from it?  Well, the real eye-catcher for me is this:

From the 1970s to 2006…the mean annual temperature of the western United States increased at a rate of 0.3° to 0.4°C decade−1, even approaching 0.5°C decade−1 at the higher elevations typically occupied by forests

They are saying that for the period 1971-2006 the temperature of the Western US increased 1.1°C to 1.4°C, or 2-2.5°F.  And it increased as much as 6.3°F in the higher elevations.    This seems really high to me, so I wondered at the source.  Apparently, it is coming from something called the PRISM data base.  These guys seem to have some sort of spacial extrapolation program that takes discreet station data and infills data for the area between the stations, mainly using a linear regression of temperature vs. altitude.  I have zero idea if their methodology makes any sense, but knowing the quality of some of the station data they are using, it may be GIGO.  (By the way, someone at Oregon State, who apparently runs this site, needs to hire a better business manager.  Their web site repors that in an academic environment awash with money for climate research, their climate data base work has been suspended for lack of funding).

As a back check on this number, LaDochy in 2007 looked at California temporal and spatial temperature trends in some depth.  He found that when one pulls out the urban stations, California rural areas experienced a 0.034C per decade temperature increase from 1950-2000, an order of magnitude lower than the numbers this study is using (click to expand slide below):

ladochy

Satellite data from UAH, which does not have the same urban bias problems, shows near-surface temperatures rising 0.1-0.3C per decade from 1978-2006 in the study areas, higher than the LaDochy numbers but still well below the study numbers.

This is a real problem with the study.  If you really want to measure the impact of temperature and participation on a 2.5 acre lot, you have to actually measure the temperature and precipitation, and even better things like the soil moisture content, somewhere near the 2.5 acre lot, and not look at western averages or computer interpolated values.

The study authors conclude:

Warming could contribute to increasing mortality rates by (i) increasing water deficits and thus drought stress on trees, with possible direct and indirect contributions to tree mortality (13, 23); (ii) enhancing the growth and reproduction of insects and pathogens that attack trees (6); or (iii) both. A contribution from warming is consistent with both the apparent role of warming in episodes of recent forest dieback in western North America (5, 6) and the positive correlation between short-term fluctuations in background mortality rates and climatic water deficits observed in California and Colorado (13, 24).

I guess I won’t argue with the word “consistent,” and I suppose it is unfair to hammer these guys too hard for the way the MSM over-interprets the conclusions and latches on to the global warming hypothesis, but really, isn’t that why the warming material is included in the paper, to get attention for the authors?  Because this paragraph would be a nice summary in a paper proposing a new study, and the hypothesis is a reasonable one to test, but it certainly isn’t proven by this study.

Postscript: From the map, some of the test plots are almost right on top of the California bristlecone pines used for climate reconstruction.  Remember, Mann and company begin with the assumption that tree growth is positively correlated with temperature.  This article argues that warming is stunting tree growth and causing trees to die.  While these are not impossible to reconcile  (though its hard considering the authors of this study said their findings were consistent across tree age, size, and variety) I would love to see how the RealClimate folks do so.

Update: Note that I still have not read the complete study itself, so I am sure there are climate regressions and such that did not make the publication or the online exhibits in Science.  So this quick reading may still be missing something.

Update #2: The best reconciliation I have received on this study vs. dendro-climatology work is the following, and is suggested on this page.   Certain trees seem to be growth-limited by temperatures, and certain trees are growth limited by water  (I presume there are other modes as well).  Trees that are temperature-limited will have their growth gated by temperature.  Trees that are water-limited will have growth controlled primarily by precipitation levels.  Grassino-Mayer states:

…sites useful to dendrochronology can be identified and selected based on criteria that will produce tree-ring series sensitive to the environmental variable being examined. For example, trees that are especially responsive to drought conditions can usually be found where rainfall is limiting, such as rocky outcrops, or on ridgecrests of mountains. Therefore, a dendrochronologist interested in past drought conditions would purposely sample trees growing in locations known to be water-limited. Sampling trees growing in low-elevation, mesic (wet) sites would not produce tree-ring series especially sensitive to rainfall deficits. The dendrochronologist must select sites that will maximize the environmental signal being investigated. In the figure below, the tree on the left is growing in an environment that produced a complacent series of tree rings.

So I suppose that while most trees are suffering from higher temperatures via the moisture mechanism, so may be benefiting, and the key is to pick the right trees.

Of course, given that bristlecones were selected as much for the fact that they are old as the fact their growth is driven by one thing or another, the problem is how one knows a particular tree’s  is temperature or moisture driven, and how one can have confidence that this “mode” has not changed for a thousand or more years.

Are bristlecones driven by temperature (as they are at fairly high altitude) or by precipitation (as they are in a very arid region of the southwest).  One might expect that given divergence issues in the bristlecone proxies, the Mannian answer of “temperature” might be wrong.  The NASA site offers this answer on the bristlecones:

Douglas’ [bristlecone] rings [from the White Mountains of CA, the same ones Mann uses] tell about rainfall in the southwestern United States, but trees also respond to changes in sunlight, temperature, and wind, as well as non-climate factors like the amount of nutrients in the soil and disease. By observing how these factors combine to affect tree rings in a region today, scientists can guess how they worked in the past. For example, rainfall in the southwestern United States is the factor that affects tree growth most, but in places where water is plentiful, like the Pacific Northwest, the key factor affecting tree ring growth may be temperature. Once scientists know how these factors affect tree ring formation, scientists can drill a small core from several trees in an area (a process that does not harm the tree) and determine what the climate was in previous years. The trees may also record things like forest fires by bearing a scar in a ring.

  • RPJ

    Given that you posted false data in your knee-jerk reaction post about a paper you hadn’t even read, and have not clarified where you got this graph from, when USHCN gives something strikingly different, why do you think your views on this study specifically, and climate generally, should be given any credibility?

  • Larry Sheldon

    Is there anything in the paper on the topic of tree mortality in connection with:

    Repeatedly boring holes in the trees to admit foreign stuff like bugs and diseases, and causing (one presumes) loss of moisture and disruptions of sap-flow?

    Cutting trees up in photogenic slices and stacking them around not-so-photogenic “scientist’s” offices?

  • RPJ

    Given that you posted false data in your knee-jerk reaction post about a paper you hadn’t even read, why do you think your views on this study specifically, and climate generally, should be given any credibility?

  • RPJ

    “I once had a professor tell me every study was a year of intensive data gathering and analysis followed by 2 hours of a group of grad students trying to brainstorm causes and implications, the more exciting the better” – either the professor was in a really poor department, or you’ve just made this anecdote up. I suspect the latter.

  • admin

    To RPJ:

    The USHCN data I posted is the raw data. The chart you link is from another study, and has been adjusted, and without knowing the adjustments they make I can’t comment on the validity.

    I do know that when the USHCN and GISS make adjustments to the Grand Canyon in their spatial modeling and roll-ups, the temperature gets adjusted up. The agencies claim that this is to take out urban heat island effects, but it is hard to imagine the Grand Canyon having a negative urban heat island effect (rural cold island?) What tends to happen in these models is that the urban upwards bias does not get reduced, but gets spread around the surrounding geographies like peanut butter in an averaging process.

    I had a discussion specifically about this station and its adjustments by the GISS here

    As to the anecdote, it is certainly true, but I have no problem with readers discounting anecdotes of unnamed (for obvious reasons) sources. I would certainly discount them. However, you don’t have to rely on an anecdote. This paper is a great example. Solid observational data, weak causality discussion. It is very common that the weakest hypothesis in a paper gets the most media attention, often to the embarrassment of the folks who authored the paper.

    And by the way, it is all against a backdrop of my believing a) temperatures have increased some over the last decades, though not by several degrees and b) my sense that this certainly would stress some trees

  • admin

    By the way, here are pictures of the Grand Canyon and Tucson sites mentioned in my comment above.

  • RPJ

    No, admin, you are wrong. The graph I posted was from the USHCN web site. It has not been adjusted. It was annual mean temperatures, as measured. You haven’t given a link to where you got your graph from.

    Your claims about urban heat islands are untrue. This is the raw data. This is the UHI-corrected data.

    “What tends to happen in these models is that the urban upwards bias does not get reduced, but gets spread around the surrounding geographies like peanut butter in an averaging process.” – not so. Urban stations are corrected such that their trends match those observed at nearby rural stations. There is no averaging process. This is all very well documented in GISS publications.

    If your anecdote about the professor was true, then this must have been in a very poor department.

  • RPJ

    No, admin, you are wrong. The graph I posted was from the USHCN web site. It has not been adjusted. It was annual mean temperatures, as measured. You haven’t given a link to where you got your graph from.

    Your claims about urban heat islands are untrue. I already posted the link to the UHI-corrected data. This is the raw data. The difference is minuscule.

    “What tends to happen in these models is that the urban upwards bias does not get reduced, but gets spread around the surrounding geographies like peanut butter in an averaging process.” – not so. Urban stations are corrected such that their trends match those observed at nearby rural stations. There is no averaging process. This is all very well documented in GISS publications.

    If your anecdote about the professor was true, then this must have been in a very poor department.

  • Keith W

    RPJ, you need to read the headers of your links. The graph you posted says at the top, “Annual Mean of Urban Heat Adjusted monthly mean temperature (F) – 1903-2006” . Seems to me that the graph was made first by creating monthly averages, adjusting those for UHI, and then computing an annual average. Warren wins on this one.

  • Larry Sheldon

    What I think RJP is trying to point out, and Warren and others are failing to see, is that all aspects of the Global Warming Model prove and predict all outcomes.

    So there is no conflict that the trees used to prove Global Warming because of the evidence of health and so on are being killed by it.

    When I lived in California, the answer from the experts when asked about the obvious death of trees was “acid rain”.

  • An Inquirer

    RPJ and admin,
    There are plenty of opportunities for miscommunication and for talking past each other when discussing temperature records and adjustments. Before raw temperatures get to GISS, they are adjusted for factors such as Time of Observation and other factors, but not UHI. GISS reverses these adjustments and then adjusts for UHI and other factors. (It has been probably a year since I studied this process so the process could have been changed since then — I know that GISS made a minor improvement in the process, apparently after being ridiculed at CA for introducing needless work and not getting certain data more directly, but it was a minor change.) The bottom line, both of you could be right in what you are referencing and not in full understanding of what the other person is referencing.

  • papertiger

    Let’s preface this comment by restating the fact that trees by design maintain their own internal temperature to facilitate photosynthesis.
    There is no such thing as a tree-om-meter.

    Treerings do respond to enhanced co2 conditions, as long as they have adequate water and nutrients in the soil same as any other plant.

    One possiblility that the scientists didn’t cover in their brainstorming session, the enhanced co2 conditions cause old growth groves to depleat the soil of nutrients quicker then they used to. Natures ability to regenerate nitrogen from forest detris might be outstripped by the co2 induced growth spurt.

  • Bob

    Trees like all living plants are subject to the law of minimums in which that which is in least suppy will limit growth. The top three things for tree growth are oxygen, heat and moisture. This is why they raise seedlings in greenhouses. It is very unlikely that a temperature increase would have a negative effect on growth or mortality. I would think that all the other possibilities that the researchers indicated as a cause for the increased mortality are far more likely than any temperature change.

  • hunter

    Of course the AGW promoters would dismiss the reasons forestry experts have found for tree growth changes.
    They don’t care about the facts of the case, and they already know the answer.
    For AGW promoters, the answer to any question is always, always, AGW.
    For AGW believers, no matter how tortured or opaque or obviously contrived the evidence, what AGW promoters have just said is always and forever right, until AGW promoters contradict what they said earlier, when they are still always and forever right.
    Because the most important thing about AGW is that AGW is always true.

  • Keith H

    They discounted beetle infestation as a cause? Unbelievable. I’ve personally seen the effects of the Spruce Bark Beetle, for example, in Cedar Breaks National Monument. Cedar Breaks has mostly fir trees and spruce trees. They look somewhat similar except for one thing… All the spruce trees are dead. It was stunning to walk through a forest and see dead spruce trees everywhere with similar fir trees growing just fine next to them. Of course, it must be global warming that’s causing the spruce beetles to spread since what else could it be? (That’s sarcasm in case someone couldn’t tell.)

  • cfdman

    The data has absolutely been adjusted, not only does it say so directly, but if you follow to source – it is referenced as having been adjusted there as well. I am not well enough informed to precisely say how it was adjusted, but claiming that is has not been showes ZERO credibility.

    RPJ = LIAR

  • RPJ

    cfdman – you’re an ignorant cunt and you don’t have a clue. How dare you call me a liar? Typical denialist yapping idiot. The only positive to people like you existing is that you rapidly destroy any last shred of credibility that the denialists ever had.

  • papertiger

    Usually bald faced liars impliment an adjustment of charm to offset and begile the target of their prevarication.

  • Forester

    Read the article, the original post and this post. I am a work in the woods forester with only a BS degree but have been practicing forestry for 36 years. If I was the faculty adviser for these guys they would get a C at best. Trees die for lots of reasons. The dynamics of a forest are complex. Finding that more trees have died only tells you, that more trees have died. Someone with appropriate knowledge might be able to determine the cause of death from examining recently dead stems. That these researchers did not develop an exhaustive list of potential causes of mortality shows either a lack of knowledge of their subject or they were trying to prove a prior answer. Either way it’s poor science. There is lots of research published on what might be causing increased tree mortality in forests. It’s found in various journals and other publications. A decent literature review would have given the authors the knowledge to better discuss the consequences of their findings.

  • hunter

    “Their web site repors that in an academic environment awash with money for climate research, their climate data base work has been suspended for lack of funding”

    Here in one sentence, you present a fantastically wrong misconception, and evidence that directly contradicts it. Such are the contortions of thought that deniers constantly have to work themselves into, to maintain their fantastic beliefs.

  • Mike Davis

    As the owner of a tree farm I agree with what forester stated. RPJ and HUNTER: this thred is about a report that was badly written and investigated. your personal attacks devalue any opinion you present.

  • Dave

    Hey Coyote,

    I am a professional Geographer (MA); minored in Ecology. I have worked in forestry as an analyst and cartographer for almost a decade. I have also done some work in climatology. I’m up to my eyeballs in a hydrology project, so I’m not going to make an extended comment, but here’s three quick points:

    1. Yes, trees have demographics. The growth of individual trees, stands, and forests tends to approximate a logistical curve. Lots of stuff complicates this.
    2. You may want to examine the study’s sampling methods. That they find more dead trees in one plot size than another makes me suspect that they may be running into what geographers call “the boundary problem.”
    3. Chris Daly’s PRISM is a good spatial interpolation model, but I don’t think it is any better than, say, Steve Running’s MTCLIM. IIRC, PRISM modifies a Digital Elevation Model(DEM)and regresses the elevations against temp or precip until the fit ceases to improve. Then it uses that relationship to interpolate and extrapolate to unsampled locations and elevations. Like any model, its reliability is dependant on the fit of its logic to the reality of the modeled domain, and the quality of the input data. So, yeah, plug in c**p data, and PRISM will output c**p.

    Cheers,

    Dave

  • Old Gut

    Tree mortality in the western US? Try pine bark beetles: http://www.ext.colostate.edu/pubs/insect/05528.html

  • pft

    Interesting article here:

    http://www.treedictionary.com/DICT2003/hardtoget/jk-53/index.html

    “There is also an environmental component of tree longevity. Some trees live longer where site conditions restrict growth rates. Bristlecone pine, Pinus aristata, growing under stressful conditions on mountains, can live for more than 3000 years.”

    So if trees are living in conditions conducive to rapid growth, presumably they die off earlier. In fact, 33% of the country is covered with trees, and the volume of timber is 25% greater than 40 years ago. We have as much forested area as we had 80 years ago. One reason is the elimination of the horse for travel and work led to less crops being grown for food, and agricultural improvements, not to mention higher CO2 levels, led to greater yields and less crop land being required.

    I sometimes wonder if overpopulation of trees in a forest lead to competition for resources (water and sunlight and nutrients from soil) that might affect mortality rates. Regional climate factors are certainly issues as well, but the science of tree mortality seems not very developed from what I could find. More questions than answers, because trees live a long time in rural areas.

  • pft

    Interesting article here:

    http://www.treedictionary.com/DICT2003/hardtoget/jk-53/index.html

    “There is also an environmental component of tree longevity. Some trees live longer where site conditions restrict growth rates. Bristlecone pine, Pinus aristata, growing under stressful conditions on mountains, can live for more than 3000 years.”

    So if trees are living in conditions conducive to rapid growth (higher temperature, higher CO2 levels, etc), presumably they die off earlier. In fact, 33% of the country is covered with trees, and the volume of timber in forests is 25% greater than 40 years ago. We have as much forested area as we had 80 years ago. One reason, interesting enough is the elimination of the horse for travel and work (replaced by fossil burning machines and cars) led to less crops needed to be grown for food, and agricultural improvements, not to mention higher CO2 levels, led to greater yields and less crop land being required, and more land available for forests to grow. The Federal government owns 30% of all land inhibiting development for other purposes, and thats been the case since Teddy Roosevelt

    Overpopulation of trees in a forest leads to competition for resources (water and sunlight and nutrients from soil) that might affect mortality rates for trees due to suppression or starvation.

    Regional climate factors are certainly issues as well, but the science of tree mortality seems not very developed from what I could find. More questions than answers, maybe because trees live a long time in rural areas. Like climate science, the historical data just is not that great, so what seems abnormal today may be very normal, it’s just the first time we have observed it, and not because it never happened before.

  • Fred

    Bob: How is oxygen a requirement for tree growth? Trees breathe in CO2 and expire oxygen.

    If CO2 is the limiting factor, then as CO2 goes up, tree rings would get fatter. There are studies on this, search for ‘co2 fertilization tree ring’ to start.