Coincidence and Contradiction in the Warming Boreal Forest

By Glenn Patrick Juday

This is the second article in an occasional series by authors invited to trace how their personal thinking about their research has changed over time.

After its purchase in 1867, for most of a century Alaska was a vast expanse of unclassified federal public lands, but in the late 1970s the U.S. government began classifying areas for their long-term management. Early in this process, a far-sighted group of scientists and resource managers developed a plan for a network of ecological reserves, envisioned as experimental treatment areas, as baseline sites for long-term studies needed to inform management, and as sites to protect key biodiversity resources. I began work as the first Alaska Ecological Reserves Coordinator in 1977, and almost immediately climatic oddities with implications for the network began to come to my attention.

Immediate aftermath of the Rosie Creek Fire, June 1983.

Immediate aftermath of the Rosie Creek Fire, June 1983. A severe convective firestorm generated hurricane-force winds of flame that toppled and snapped the trees in this area. Warming and drying of Alaskan boreal forests has led to increased area burned and high fire severity. Figure courtesy of G.P. Juday.

Tundra fires, always uncommon, burned a record high area on the Seward Peninsula during a record warm August 1977. In 1978 mean annual temperature at Fairbanks was second only to 1926 in an observational record back to 1904, as was Anchorage with a record back to 1916. Then came the spectacular warmth of January 1981. The mean monthly temperature at Fairbanks was 16.7˚C (30.0˚F) above normal, the greatest departure from normal of a weather station in the history of North America to that time. Maybe these were all coincidences, but it seemed like rolling dice and always seeing fours or fives come up. We now know that the Pacific climate regime shift from 1976 to 1977 was one of the most powerful influences on the condition and functioning of ecosystems across a vast area of western North America (Ebbesmeyer et al. 1990) and that these effects were particularly noticeable in Alaska.

At the time, although I considered global warming from human-caused increases in greenhouse gas as an explanation, something as dramatic as global-scale climate and ecosystem change seemed like a distant prospect, not something likely actually to be important in my career. Maybe, I thought, the odd weather events were just cyclic variability. But if these events really were an early expression of global greenhouse warming, I wanted the areas chosen for the ecological reserves network to have baseline observations and data in case—or whenever—global climate change did occur. That's one of the main reasons we were establishing them.

Beyond Coincidence?

On one hand, the physics of how increases in greenhouse gasses retain more heat in a system seemed virtually certain. On the other hand, maybe some process such as enhanced carbon sequestration would operate to dampen any warming effect to the point it would be negligible. In any event, how would we recognize global warming effects if we saw them? A good answer to that question was not available at the time.

Mean daily low temperatures during the warm season at Fairbanks.

Mean daily low temperatures during the warm season at Fairbanks. Summer temperatures have remained at elevated levels since the mid-1970s. A greater magnitude of warming in the daily low vs. high temperatures is pronounced and consistent with the mechanism by which greenhouse gasses work. Figure courtesy of G.P. Juday.

To address that question, a group of faculty at the University of Alaska Fairbanks (UAF) planned a scientific meeting to evaluate the evidence from the atmosphere, cryosphere, oceans, and land systems and gain a bigger perspective about environmental change in Alaska. I joined the conference organizing committee, which included Jenifer McBeath (Agricultural and Forestry Experiment Station), Gunter Weller and Tom Osterkamp (Geophysical Institute), and Richard Neve (Marine Science). We agreed to consider both what the science could tell us and what the implications of a warmer Alaska would be for society in general.

With support from the UAF School of Agriculture and Land Resources Management and funding from the Alaska Humanities Forum, one of the first national meetings to consider climate change evidence in a specific region, including human implications, convened in April 1982 (McBeath et al. 1984). Charles Keeling of the Scripps Institution of Oceanography presented his atmospheric CO2 concentration measurements (at that time about 341 ppm, in 2008 386 ppm). Climatologist Will Kellogg of the National Center for Atmospheric Research noted that models based on simple representations of heat flux showed that if CO2 concentrations continued to increase at anticipated rates, "…the Arctic Ocean will become ice-free with a relative modest warming, one that could occur very early in the next century…." Current evidence suggests that his prediction was accurate.

My paper analyzed temperature trends in the Alaska climate record. My office at the Institute of Northern Forestry (INF) had two tools that helped greatly. First, INF's small library held the nearly complete National Weather Service Climatological Data and Local Climatological Data publication series for Alaska. Second, INF had just obtained a computer with a pen plotter. Today, access to data or the ability to manipulate and display information on an affordable device seems trivial, but they were big challenges at the time.

Boreal forest is found south of arctic treeline (dark green line).

Boreal forest is found south of arctic treeline (dark green line). The orange line indicates the Arctic, as defined by the Arctic Council's Arctic Climate Impact Assessment (ACIA) and Arctic Monitoring and Assessment Programme (AMAP). Graphic from UNEP/GRID Arendal (2002).

As I analyzed the Alaska temperature data, the pile of squiggly lined graphs grew higher and higher, and nearly all displayed a sharp upswing at the far right of the page, representing the high temperatures of the most recent years. Again, this sounds elementary today, but at the time it was a noteworthy trend—seeing the hard data at so many stations going up to such high levels was compelling.

My results also showed a strong cyclic feature in the record, which was partly related to the solar cycle and to El Niño, as a few others had suggested earlier. In addition to giving a summary perspective on about 80 years of climate data, I was looking for a reasonable and specific test that would address the question of greenhouse gas warming. I concluded that "if, as expected, CO2 begins to overwhelm the natural range of climate variability between now [1982] and the end of the century, Alaska would experience a stairstep increase in temperatures, the peaks of which would reach unprecedented highs." That basically describes what happened, but, of course, I wasn't certain at the time.

I had unilaterally defined my work on the conference topic as part of my ecological reserve duties. Fortunately, my boss, Ken Wright (associate director of the U.S. Forest Service Pacific Northwest Research Station), was very understanding. He thought that this global warming issue might eventually be important (a brave admission at the time), but probably only in the long-term future. My administrators and funding sources were anxious for me to get back to "real work," so I returned to selecting, documenting, and establishing ecological reserves and starting monitoring to enable us to detect important ecological changes if they occurred.

Burning Questions

In 1983, an external review panel from the National Science Foundation (NSF) met in Fairbanks in late May to assess the accomplishments of the Taiga Biome project and identify priorities for future boreal forest research. Recognizing the good progress the project had made in understanding fire ecology and the black spruce ecosystem, the panel recommended that the next phase of research focus on higher productivity white spruce forests. As the group was meeting, the fast-moving Rosie Creek wildfire burned across 8,600 acres just west of Fairbanks, including about one third of the Bonanza Creek Experimental Forest, one of my ecological reserve sites. The fire burned a significant amount of productive white spruce forest and displayed particularly severe fire behavior because of the warm, dry conditions. Another coincidence, it seemed.

Figure courtesy of G.P. Juday

Record high temperatures in 2004 and 2005 led to severe reduction of the 2005 ring from drought stress; warm weather favorable to the spruce budworm led to growth reduction from defoliation in 1993 and 1995. Figure courtesy of G.P. Juday.

As the panel continued its work, I talked with the local investigators attending the review to define the research topics they thought were important now that the fire had occurred. We quickly developed a research plan, which we sent to local members of the state legislature as the legislative session in Juneau was ending and final agreements on appropriations were being made. The Fairbanks delegation arranged an immediate appropriation for the Rosie Creek Fire Research plan. The contributing scientists and I were amazed that it happened at all, let alone so quickly. The fire effects studies added to the considerable research history in the Bonanza Creek Experimental Forest, and in 1987 the area became one of the early sites in the NSF-supported Long Term Ecological Research (LTER) network.

During a sabbatical leave, the strong El Niño of 1987–88 kept climate anomalies before me as I visited five Canadian provinces in addition to 20 U.S. states and began to see firsthand the practical challenges of managing areas for biodiversity in a changing environment. I spent time with Gary Davis, biologist for Channel Islands National Park in southern California, who had developed what was universally recognized as the model environmental monitoring program for parks or nature reserves. I helped him record data in intertidal plots, documenting huge ecological effects cascading through the marine ecosystem—ultimately triggered by an exceptional warm water anomaly. Another coincidence?

As I returned from my sabbatical, a flurry of events related to global warming culminated in the Yellowstone fires of 1988 and James Hansen's testimony to Congress, which news media saw as the first unequivocal statement by an eminent scientist that ongoing temperature anomalies could be interpreted as human-caused global warming. But record warmth did not continue uninterrupted, and by the early 1990s it seemed that climate change had faded on the national agenda. I had to consider whether I would continue doing climate change work at that stage in my career. For me personally, the decision came down to this: how would I feel if the biggest change to affect northern forests in the past several thousand years occurred, and I was too busy to notice? I decided that even if the time scale of change put the confirmation of global warming effects past my retirement, I would go ahead and focus on the potential effects of warming on boreal tree growth and forest health.

Contradictory Tree-Ring Results

To complement the forest monitoring (looking forward in time) I had been doing, I wanted to study the history of forest growth and development (looking backward in time). So I began to learn tree ring analysis, with great help from Gordon Jacoby and Rosanne D'Arrigo of the Lamont-Doherty Tree Ring Laboratory of Columbia University. At that time, I wasn't interested in dendroclimatology, which involves using tree rings to reconstruct past climates, because reconstructions based on Alaska tree rings had been published for a number of years, and those questions seemed fairly settled.

Figure courtesy of G.P. Juday.

Increased temperatures reduce growth of productive white spruce stands both directly and indirectly. Growth of a monitored stand at the Bonanza Creek LTER is directly proportional to summer temperatures, and highest temperatures are the least favorable. Figure courtesy of G.P. Juday.

For my first big tree ring sample, I used a chain saw to cut off stump sections from 100 large white spruce trees killed in the Rosie Creek Fire, a stand of some of the biggest, fastest growing trees in interior Alaska, and definitely not the kind of cold treeline site typically used for a tree ring-based climate reconstruction. Les Viereck's work at Bonanza Creek LTER had previously shown that, as you might expect, total productivity (ability to grow plant matter) was greatest on sites with warm soils and least on sites with cold soils (Viereck et al. 1986). Just for due diligence, I plotted my ring-width sample data against Fairbanks climate data, expecting to see no relationship. I was quite wrong, however, because the year-to-year change in temperatures and growth of my white spruce sample showed a strong relationship—only it was a negative relationship.

This meant that as summer temperatures increased the trees grew less, and as summer temperatures decreased the trees grew more. That just seemed the wrong result in Alaska, where all the published papers from "properly" collected tree samples at cold treeline sites show a positive relationship. I was concerned I might have made a mistake. A negative relationship between growing season temperatures and tree ring width, I knew, mainly happened on hot, dry sites as a result of drought stress limiting tree growth. If I had a valid result, the implications were very great for the boreal forest.

I increased my white spruce tree ring sampling effort with the same result. The trees were definitely not growing well in the regime of increasing summer temperatures that had begun in 1977 and strengthened since. But I wanted evidence about the mechanism causing the common growth signal in the trees.

About that time Valerie Barber, now at the UAF-Palmer Center for Sustainable Living, started a Ph.D. program, working with me and Bruce Finney, who ran a paleoecological lab in the Institute of Marine Science with expertise in stable isotopes. Val thought we might be able to use the carbon-13 (13C) in the wood to measure moisture stress. From my samples, she painstakingly harvested the wood from each year's ring from several trees. Her lab was soon filled with little jars holding ground-up wood samples for later cellulose digestion and extraction.

I distinctly remember the moment Val brought all the data, and we sat down to plot isotopes versus ring width and temperature. If we were right about drought stress, 13C would be related to temperature and ring width, and if we were wrong they would be unrelated. When the graphs appeared on the screen, the relationship was so strong that we laughed.

We knew that we had to get this story right, so Val went to the Lamont-Doherty Tree Ring Lab to measure wood density with x-ray. In most conifers if a tree has experienced high temperatures and/or drought, more of the year's growth is produced as dense latewood. Val did another meticulous job of preparing wood slices for analysis, and we found that the density data agreed with our ring width and 13C results. Finally, I compared the isotope and density results to ring width during the 20th century in 269 white spruce trees from 20 stands across central Alaska. The relationship, which was the same for all the stands and nearly all the trees, was as strong in the first half of the 20th century as in the second half. Temperature-induced drought stress controlled the growth of these trees, which were representative of stands with the greatest value for timber production and the most active in taking up atmospheric CO2. Many, if not most, ecosystem or general circulation models had boreal trees grow more as temperatures increased, but we showed that these trees would do the opposite (Barber et al. 2000).

In follow-up work, Val and I isolated the oldest trees in our data set, extended the climate sensitivity analysis back another century, and performed a formal temperature reconstruction for central Alaska during the 1800s (Barber et al. 2004). We found signals in the 19th century of the Pacific Decadal Oscillation (PDO), as well as two unexpected periods of warm summers that were likely accompanied by large-scale fires (Juday et al. 2003). Overall, we could say that summer temperatures since the 1976–77 regime shift were the warmest in the past 200 years. In fact, given that mature spruce trees are often about 200 years old, the great majority of these trees probably had not experienced temperatures as warm in their lifetimes.

These results raised other issues as well. As I mentioned earlier, dendroclimatologists typically collect samples from "limiting stands" such as treeline, where trees are at their margin of cold tolerance and temperature effects on ring width should be less confounded by other factors; presumably, warming should mitigate, if not completely overcome, temperature limitation to growth. Yet accumulating evidence suggested that the relationship between site-based tree-ring chronologies and temperature predictions of growth became weaker around the mid-20th century (Briffa et al. 1998). If tree-rings don't respond consistently to climate forcing functions, then the entire field of reconstructing past climates from tree-rings might need to be re-evaluated.

Doctoral student Martin Wilmking, now at the University of Greifsvald in Germany, did a comprehensive assessment of the relationship of white spruce trees to environmental characteristics at treeline. The published literature actually used relatively small samples, typically a few dozen trees carefully selected by the investigator based on a judgment that they were the most likely to contain a climate signal. The samples Martin and I analyzed ultimately totaled about 2,600 trees from 15 treeline and near-treeline locations in the Brooks Range and Alaska Range. A bit less than 40% of our treeline trees had a positive growth response to warming, as expected. In all but the coolest years, however, the growth of over 40% of the sampled trees was negatively related to midsummer temperatures (Wilmking et al. 2004). Once a threshold temperature was reached, additional warming reduced growth in these negative responders. So the apparent weakening of treeline response to recent temperature increases came from mixing samples in which growth responses to temperature varied, with some increasing and some decreasing (Wilmking et al. 2005). By using only one consistent responder type, tree ring temperature reconstructions could be reliable.

A spruce budworm feeding on a tree on the University of Alaska Fairbanks campus.

A spruce budworm feeding on a tree on the University of Alaska Fairbanks campus in 2008. Outbreaks of this insect were rare or unknown in Alaska until temperature increases created favorable conditions. Figure courtesy of G.P. Juday.

Obviously if global climate change was in fact occurring, its effects should also become evident on a larger scale. In 2004, the Arctic Council sponsored the Arctic Climate Impact Assessment, a major international collaborative study and synthesis of climate change and its effects across the circumpolar north. I was given the task of pulling together information on forests, land management, and agriculture with a large author team (Juday et al. 2005). It became clear that a period of major, sustained temperature increases was, in fact, underway in the North. In parts of the boreal forest with greater precipitation, such as eastern Canada, western Russia, and the Nordic countries, tree growth generally increased with increasing temperatures, but in the Russian Far East and central and western North American boreal region, temperature increases were often (but not exclusively) decreasing tree growth and increasing fire and insect outbreaks. I reported that in addition to white spruce, growth of some black spruce and Alaska birch populations responds negatively to warming.

The Changing Future

In assessing boreal forest response to warming, it had been assumed that forest fires and tree-damaging insect outbreaks, which are warm temperature phenomena, would increase as well. By the 1990s, the extent of fire in the global boreal forest had increased, but it was difficult to see the trend in the Alaska wildland fire record until 2004—then the fires of 2004 and 2005 burned over 4.2 million hectares in Alaska, equivalent in size to Sri Lanka. The unexpectedly large fire season of 2009 burned an additional 1.2 million hectares, resulting in a cluster of record or near-record fire years closely spaced over a mere six years. This rapid transformation of the landscape appears to be beyond previous disturbance regimes, taking us into an unknown future boreal forest.

Multiple and simultaneous outbreaks of forest damaging insects have occurred to greater extents as temperatures increased in the past 20 to 30 years in Alaska. In some cases, such as spruce budworm in central Alaska or the spruce bark beetle in southcentral Alaska, outbreaks are clearly related to increasing temperatures. In others, such as aspen leaf miner, the cause is not known. There seems little doubt that continued temperature increases would allow the survival and successful reproduction of a greater variety of potentially forest-damaging insects, while the process of forest tree adaptation or addition of species is likely to be markedly slower.

The strong trend of increasing temperature and the variety and vast scale of major effects of warming on Alaska boreal forests are so obvious today that the continuing change is now impossible to ignore. But when did I become convinced?

I found myself facing that question a couple of years ago in an interview with the Finnish newspaper Helsingin Sanomat, and I realized that no single piece of evidence was responsible. From the beginning of my work, I knew that the temperature anomalies in Alaska might be an effect of global warming, but also that I might be wrong. I felt the need to test my interpretations and use the objections of those who disagreed to come back with more convincing evidence. After enough specific effects that were anticipated—if not predicted—had occurred, I just got to the point that it was unreasonable to me to believe that the global warming explanation was wrong. During the opportunities we have had to explain our results, my colleagues and I have always tried to convey that sense of how science works. We constantly have to test our ideas and look for consistency in our explanations. And when we find it, we need to draw the conclusions that are the most reasonable.

Glenn Patrick Juday is a Professor of Forest Ecology at the School of Natural Resources and Agricultural Sciences, University of Alaska Fairbanks.

Further Reading

Barber, VA, GP Juday, BP Finney. 2000. Reduced growth of Alaska white spruce in the twentieth century from temperature-induced drought stress. Nature 405: 668-673.

Barber, VA, GP Juday, BP Finney. 2004. Reconstruction of summer temperatures in interior Alaska: Evidence for changing synoptic climate regimes. Climatic Change 63: 91-120.

Briffa, KR, PD Jones, FH Schweingruber, TJ Osborn. 1998. Influence of volcanic eruptions on northern hemisphere summer temperature over the past 600 years. Nature 393: 450-455.

Ebbesmeyer, CC, DR Cayan, DR McLain, FH Nichols, DH Peterson, KT Redmond. 1990. 1976 Step in the Pacific climate: Forty environmental changes between 1968-1975 and 1977-1984. In Proceedings of the Seventh Annual Pacific Climate (PACLIM) Workshop. California Department of Water Resources. Sacramento, California.

Juday, GP, V Barber, E Vaganov, S Rupp, S Sparrow, J Yarie, H Linderholm. 2005. Forests, Land Management, Agriculture. In Arctic Climate Impact Assessment. Arctic Council. Cambridge University Press. Pages 781-862.

Juday, GP, V Barber, S Rupp, J Zasada, MW Wilmking. 2003. A 200-year perspective of climate variability and the response of white spruce in interior Alaska. In Climate Variability and Ecosystem Response at Long-Term Ecological Research (LTER) Sites. Oxford University Press. Pages 226-250.

McBeath, JH, GP Juday, G Weller, M Murray, eds. 1984. The Potential Effects of Carbon Dioxide-Induced Climatic Changes in Alaska, The Proceedings of a Conference. School of Agriculture and Land Management, University of Alaska, Misc. Publication 83-1.

Viereck, LA, KV Cleve, CT Dyrness. 1986. Forest ecosystem distribution in the taiga environment. Forest Ecosystems in the Alaskan Taiga: A Synthesis of Structure and Function. Springer-Verlag, New York. Pages 22-43.

Wilmking, M, R D'Arrigo, GC Jacoby, GP Juday. 2005. Increased temperature sensitivity and divergent growth trends in circumpolar boreal forests. Geophysical Research Letters 32(15): L15715. doi:10.1029/2005GL023331.

Wilmking, M, GP Juday, V Barber, H Zald. 2004. Recent climate warming forces contrasting growth responses of white spruce at treeline in Alaska through temperature thresholds. Global Change Biology 10: 1-13.