LETTERS

Mountain pine beetle and forest carbon feedback toclimate changeW. A. Kurz1, C. C. Dymond1, G. Stinson1, G. J. Rampley1, E. T. Neilson1, A. L. Carroll1, T. Ebata2 & L. Safranyik1

The mountain pine beetle (Dendroctonus ponderosae Hopkins,Coleoptera: Curculionidae, Scolytinae) is a native insect of thepine forests of western North America, and its populations peri-odically erupt into large-scale outbreaks1–3. During outbreaks, theresulting widespread tree mortality reduces forest carbon uptakeand increases future emissions from the decay of killed trees. Theimpacts of insects on forest carbon dynamics, however, are gene-rally ignored in large-scale modelling analyses. The current out-break in British Columbia, Canada, is an order of magnitudelarger in area and severity than all previous recorded outbreaks4.Here we estimate that the cumulative impact of the beetle out-break in the affected region during 2000–2020 will be 270 mega-tonnes (Mt) carbon (or 36 g carbon m22 yr21 on average over374,000 km2 of forest). This impact converted the forest from asmall net carbon sink to a large net carbon source both during andimmediately after the outbreak. In the worst year, the impactsresulting from the beetle outbreak in British Columbia were equi-valent to 75% of the average annual direct forest fire emissionsfrom all of Canada during 1959–1999. The resulting reduction innet primary production was of similar magnitude to increasesobserved during the 1980s and 1990s as a result of global change5.Climate change has contributed to the unprecedented extent andseverity of this outbreak6. Insect outbreaks such as this representan important mechanism by which climate change may under-mine the ability of northern forests to take up and store atmo-spheric carbon, and such impacts should be accounted for inlarge-scale modelling analyses.

Forest insect epidemics can have severe impacts on ecosystemdynamics by causing mortality and reducing the growth of millionsof trees over extensive areas7. Native insects and alien invasive speciesaffect both managed and natural forests. Beyond the ecologicalimpacts are the associated economic (for example, disrupted timbersupply to mills) and social (for example, unemployment, crime rates)effects8. The impact of insects on carbon (C) dynamics and globalclimate are not well documented9.

The current outbreak of mountain pine beetle in western Canada isan order of magnitude greater in area than previous outbreaks owingto the increased area of susceptible host (mature pine stands) andfavourable climate4 (see also Supplementary Fig. 3). An expansion inclimatically suitable habitat for the mountain pine beetle, includingreduced minimum winter temperature, increased summer tempera-tures and reduced summer precipitation, during recent decades hasfacilitated expansion of the outbreak northward and into higherelevation forests4,10. This range expansion, combined with an increasein the extent of the host, has resulted in an outbreak of unprece-dented scale and severity. By the end of 2006, the cumulative out-break area was 130,000 km2 (many stands being attacked in multipleyears), with tree mortality ranging from single trees to most of a

stand in a single year11. Timber losses are estimated to be morethan 435 million m3, with additional losses outside the commercialforest12. The forest sector has responded by increasing harvest ratesand reallocating some harvest, increasing the pine portion of theprovincial total volume harvested from 31% to 45% over four years(2001–2004).

We estimated the combined impact of the beetle, forest fires andharvesting on forest productivity and carbon balance from 2000 until2020 for the south-central region of British Columbia (Fig. 1). Thisarea includes 374,000 km2 of productive forest, largely dominated bypine (Pinus) and spruce (Picea) species. We used a Monte Carlodesign for simulating future net biome production (NBP) using aforest ecosystem model (the Carbon Budget Model of the CanadianForest Sector, CBM-CFS3). This model accounts for annual treegrowth, litterfall, turnover and decay, and it explicitly simulatesharvest, beetle-caused mortality, and fire-caused mortality and fuelconsumption. We developed regional probability distribution func-tions of the annual area burned and projected future beetle dynamicson the basis of the characteristics of the remaining host (that is, pinestands of suitable age) and the judgement of regional entomologists.We conducted 100 Monte Carlo simulations with different randomdraws from the probability distributions for the annual area of beetleoutbreak and the annual area burned.

For the period 2000–2020, the average annual NBP was 215.8 6

7.9 Mt C yr21 (or 242.4 6 21 g C m22 yr21; Fig. 2). Carbon lossesresult from emissions from decomposition and fires and fromthe transfer of timber to the forest product sector. In a separateanalysis13, we estimated that the study area was a net sink from1990 to 2002. The first two years of this study also reported a netsink (0.59 Mt C yr21), but with increasing beetle impact (Fig. 3), theforest converted to a source of 17.6 Mt C yr21from 2003 to 2020.With decreasing beetle impact (Fig. 3), NBP began to recover, butby 2020, the estimated NBP had not yet returned to pre-outbreaklevels. Although we can expect that forests will eventually recoverfrom the beetle outbreak, we are reluctant to extend projectionsbeyond 2020 or to speculate on the rate of recovery beyond 2020given uncertainties about non-host responses, rates of regeneration,and future fires in a region in which major climate change impacts areforecast14.

One component of the uncertainty in future NBP is that we do notknow the future area that will be infested by the beetle. We projectedthe area infested during 2007–2020 using random draws fromregionally calibrated probability distributions of outbreak area andduration that were based on: the 2000–2006 area; mortality and hoststatistics; historical, spatial and temporal dynamics; remaining hostpopulation; and judgment from entomologists. The outbreak wasprojected to peak between 2006 and 2008, with the maximum areainfested ranging from 74,000 km2 to 94,000 km2 (Fig. 3).

1Natural Resources Canada, Canadian Forest Service, Pacific Forestry Centre, Victoria, British Columbia, V8Z 1M5, Canada. 2British Columbia Ministry of Forests and Range, Victoria,British Columbia, V8W 9C2, Canada.

Vol 452 | 24 April 2008 | doi:10.1038/nature06777

987Nature Publishing Group©2008

We partitioned the impact of the beetle outbreak on NBP using asingle projection of beetle (Fig. 3) and fire area and by simulatingthree forest carbon dynamics scenarios: without beetle (control),with beetle but no additional harvest, and with beetle and additionalharvest. The additional harvest is considered to be part of the beetleimpact because it is a management response to the vast quantities ofbeetle-killed trees. The impact of beetles on NBP was projected to

reach 220 Mt C yr21 (253 g C m22 yr21) in 2009 (Fig. 4). Cumu-lative beetle impact over the 21-yr simulation period was projectedto be 2270 Mt C, and the additional salvage harvest and removal ofboth beetle-killed and living trees contributed a further 250 Mt C.Cumulative transfer of beetle-killed timber to the forest productsector was projected to be 31 Mt C (13% of total harvest).Harvesting results in a loss of carbon from the ecosystem, but onlysome of that carbon is emitted to the atmosphere; the remainder isstored in wood products and landfills.

NBP was reduced in the affected landscape by the beetle and theassociated additional harvest response because of reduced netprimary production (NPP) and increased heterotrophic respiration.Annual NPP dropped from 440 to 400 g C m22 in 2000–2009. Thelowest NPP values (391 g C m22) were estimated for 2015 to 2018,at which point NPP started to recover. NPP was reduced in theaffected landscape during and after the beetle outbreak because ofthe reduction in photosynthetic capacity caused by widespread treemortality. Conversely, heterotrophic respiration increased from 408to 424 g C m22 in the 2003–2007 period because of the large transfersof biomass to dead organic matter pools (471 Mt C transferred by thebeetle during the outbreak) and subsequent decomposition.

Impacts of the current beetle outbreak are important because theyare of comparable magnitude to other global change factors. Fireemissions vary widely between years, but within the study area theworst year for fires in the simulation period produced direct emis-sions of 13 Mt C, an amount exceeded by annual beetle-causedimpacts from 2005 to 2014. The maximum annual beetle impact(20 Mt C yr21) for the relatively small study area is of similar mag-nitude to the direct forest fire emissions from all of Canada during1959–1999 (27 Mt C yr21)15. Climate warming, elevated atmosphericCO2 concentrations, and atmospheric nitrogen deposition may

ab

c

0 75 150 300 km

N

Figure 1 | Geographic extent of mountain pine beetle outbreak in NorthAmerica. a, Extent (dark red) of mountain pine beetle. b, The study areaincludes 98% of the current outbreak area. c, A photograph taken in 2006

showing an example of recent mortality: pine trees turn red in the first yearafter beetle kill, and grey in subsequent years. Photo credit: Joan Westfall,Entopath Management Ltd.

Percentiles5

0

–5

–10

–15

–20

–25

–30

–352000 2002 2004 2006 2008 2010

Year

Net

bio

me

pro

duc

tion

(Mt

C y

r–1 )

2012 2014 2016 2018 2020

40–6030–40 or 60–7020–30 or 70–8010–20 or 80–901–10 or 90–99

Figure 2 | Annual NBP by percentile from the Monte Carlo simulations.The model results are based on statistics from 2000 to 2006 and projectionsfrom 2007 to 2020. Negative values represent fluxes from the forest to theatmosphere (a net source of carbon). Asymmetry of the range of estimates ofNBP in any single year is a function of the area burned and the associateddirect carbon emissions.

LETTERS NATURE | Vol 452 | 24 April 2008

988Nature Publishing Group©2008

enhance forest carbon sinks16,17. For example, it was estimated in ref.5 that NPP increased 6% during the 1980s and 1990s as a result ofglobal change. The impact of the beetle outbreak would negate thesegains within the affected region because NPP was reduced by 10%and heterotrophic respiration increased by 6%. Here we examinedjust one insect outbreak and estimated that its net greenhouse gasimpacts over 21 yr (990 Mt carbon dioxide equivalents, CO2e) arecomparable to about 5 yr of emissions from Canada’s transportationsector (200 Mt CO2e in 2005)13.

Climate change will have impacts on insect distribution andabundance either directly via effects on their life cycles or indirectlyvia host–plant defences, the abundance of natural enemies or inter-actions with competitors18. For some forest insects, the outcome maybe changes in: outbreak frequencies or duration; rates of herbivoryand damage; ranges, and associations with host species 11,19. Evidencefor climate-change-related increases in the extent or severity of forestinsect disturbance has begun to accumulate. In north-western NorthAmerica, warmer temperatures have halved the time required toreproduce for the spruce beetle (Dendroctonus rufipennis Kirby,Coleoptera: Curculionidae, Scolytinae) and have contributed tounprecedented damage to spruce forests20. In the western UnitedStates, warmer annual temperatures have caused an altitudinal shiftin the range of habitats, allowing the mountain pine beetle to invadehigh-elevation pine forests21. Insect impacts have, however, generallybeen ignored in large-scale carbon budget modelling studies22,23

because data on insect impacts covering large areas are expensive

to collect and are therefore not widely available24. Moreover, complexmodelling approaches are required to adequately represent theimpacts of insects on stand- and landscape-level forest ecosystemdynamics. This failure to account for insect impacts may haveresulted in overestimation in previous studies of the potential forforests to offset anthropogenic CO2.

Other insects, such as eastern spruce budworm (Choristoneurafumiferana Clemens, Lepidoptera: Tortricidae) and forest tent cater-pillar (Malacosoma disstria Hubner, Lepidoptera: Lasiocampidae)also exhibit outbreaks during which they affect forest carbondynamics by reducing growth and increasing tree mortality over largeareas25. Other forms of disturbance, such as fires, also markedlyinfluence terrestrial carbon budgets26. Significant climate warminghas already allowed the mountain pine beetle to expand its rangeinto formerly climatically unsuitable habitats4,21. Future projectedwarming could facilitate further range expansion. Insect outbreaks,together with fires, could put North American forest carbon sinks atrisk27. Disturbances are one of the principal drivers of the carbonbudget in northern forests26, and this study shows but one exampleof how climate change can affect disturbance regimes for whichimpacts on the global carbon cycle then provide strong positive feed-back to the global climate system.

METHODS SUMMARYThe CBM-CFS3 uses forest inventory as well as growth and yield data to simulate

stand- and landscape-level forest carbon dynamics28–30. It tracks carbon stocks,

transfers between pools and emissions. The model accounts for tree growth,

litterfall, turnover and decay as annual components of the forest carbon cycle.

Fires were simulated as stand-replacing events where a portion of foliage and

above-ground dead organic matter pools were consumed; all remaining biomass

was killed, and standing snags were transferred to litter and coarse woody debris.

Harvests were also simulated as stand-replacing clear-cut events where stem-wood carbon was transferred out of the ecosystem to the forest products sector,

and the rest of the carbon in biomass pools was transferred to pools of dead

organic matter. Beetle infestation areas were simulated as partial-mortality

events killing a portion of softwood biomass pools. The remaining stand con-

tinues to grow and recover from beetle impacts. The stand mortality classes were

5%, 10%, 30% or 50% per year (see Supplementary Information for more

details).

The British Columbia Ministry of Forests and Range supplied forest inventory

and growth curve data. Almost 700,000 forest stands were assigned to 86 spatial

analysis units that represented the administrative and ecological variability

in the study area. Spatially and temporally explicit beetle outbreak area and

a

b

Percentiles100

Bee

tle a

rea

(thou

sand

s of

km

2 )B

eetle

are

a (th

ousa

nds

of k

m2 )

80

60

40

20

0

80

60

40

20

0

Mortality class

2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020

2000 2002 2004 2006 2008 2010Year

2012 2014 2016 2018 2020

LightModerateSevereVery severe

40–6030–40 or 60–7020–30 or 70–8010–20 or 80–901–10 or 90–99

Figure 3 | Area infested with the mountain pine beetle during thesimulation period. Statistics are used from 2000 to 2006 and projections areused from 2007 to 2020. a, Percentiles describing the parameter spaceresulting from 100 Monte Carlo projections of the beetle-infested area. Ourprojections of decreasing area after 2009 are largely based on the decline inavailable live host (pine) area. b, Area statistics for the single Monte Carlosimulation used for scenario analysis broken down by host mortality class.

Sink

10

Eco

syst

em c

arb

on s

tock

cha

nge

(Mt

C y

r–1 )

5

0

–5

–10

–15

–20

–25

–30

–352000 2005

Control scenarioBeetle scenarioBeetle and additional harvest scenario

2010

Year

2015 2020

Source

Figure 4 | Total ecosystem carbon stock change for three scenarios. Thecontrol simulation was run with no beetle outbreak, and with base harvestand fires. The beetle simulation added insect impacts to the control scenario.The additional harvest simulation added the management response ofincreased harvest levels from 2006 to 2016 to the beetle simulation. Negativeecosystem carbon stock change values represent fluxes from the forest to theatmosphere (net source of carbon). The source in 2003 was, in part, theresult of the large area burned (2,440 km2 in the study area) that wasincluded in all three scenarios.

NATURE | Vol 452 | 24 April 2008 LETTERS

989Nature Publishing Group©2008

area-burned data sets were used for 2000–2006 and projected for 2007–2020using random draws from historically calibrated probability distributions.

Harvest rates for each administrative unit were available for 2000–2005, and

forecasts for 2006–2020 were based on local management plans (see

Supplementary Information for more details).

Received 9 December 2007; accepted 29 January 2008.

1. Amman, G. D. & Cole, W. E. in Mountain Pine Beetle Dynamics in Lodgepole PineForests Part II: Population Dynamics. General Technical Report No. GTR-INT-145(United States Department of Agriculture Forest Service, Intermountain Forestand Range Experiment Station, Ogden, 1983).

2. Safranyik, L., Shrimpton, D. & Whitney, H. in Management of Lodgepole Pine toReduce Losses from the Mountain Pine Beetle. Technical Report 1 (EnvironmentCanada, Canadian Forestry Service, Victoria, 1974).

3. Taylor, S. W. & Carroll, A. L. in Mountain Pine Beetle Symposium: Challenges andSolutions BC-X-399 (eds Shore, T. L., Brooks, J. E. & Stone, J. E.) 41–51 (NaturalResources Canada, Canadian Forest Service, Victoria, 2004).

4. Taylor, S. W., Carroll, A. L., Alfaro, R. I. & Safranyik, L. in The Mountain Pine Beetle: ASynthesis of Biology, Management and Impacts in Lodgepole Pine (eds Safranyik, L. &Wilson, B.) 67–94 (Natural Resources Canada, Canadian Forest Service, Victoria,2006).

5. Nemani, R. R. et al. Climate-driven increases in global terrestrial net primaryproduction from 1982 to 1999. Science 300, 1560–1563 (2003).

6. Carroll, A. L., Taylor, S. W., Regniere, J. & Safranyik, L. in Mountain Pine BeetleSymposium: Challenges and Solutions. Report number BC-X-399 (eds Shore, T. L.,Brooks, J. E. & Stone, J. E.) 223–232 (Natural Resources Canada, Canadian ForestService, Victoria, 2004).

7. Mattson, W. J. & Addy, N. D. Phytophagous insects as regulators of forest primaryproductivity. Science 190, 515–522 (1975).

8. Parkins, J. R. & McKendrick, N. A. Assessing community vulnerability: a study ofthe mountain pine beetle outbreak in British Columbia, Canada. Glob. Environ.Chang. 17, 460–471 (2007).

9. Volney, W. J. A. & Fleming, R. A. Climate change and impacts of boreal forestinsects. Agr. Ecosys. Env. 82, 283–294 (2000).

10. Williams, D. W. & Liebhold, A. M. Climate change and the outbreak ranges of twoNorth American bark beetles. Agr. For. Ent. 4, 87–99 (2002).

11. Westfall, J. 2006 Summary of Forest Health Conditions in British Columbia (BritishColumbia Ministry of Forests and Range, Victoria, 2007).

12. Walton, A. et al. Provincial-Level Projection of the Current Mountain Pine BeetleOutbreak Æhttp://www.for.gov.bc.ca/hre/bcmpb/Year4.htmæ (British Columbia,Ministry of Forest and Range, Victoria, 2007).

13. Environment Canada. National Inventory Report, 1990–2005: Greenhouse GasSources and Sinks in Canada. Æhttp://www.ec.gc.ca/pdb/ghg/inventory_report/2005_report/tdm-toc_eng.cfmæ. (Environment Canada, Gatineau, 2007).

14. Hamann, A. & Wang, T. Potential effects of climate change on ecosystem and treespecies distribution in British Columbia. Ecology 87, 2773–2786 (2006).

15. Amiro, B. D. et al. Direct carbon emissions from Canadian forest fires, 1959–1999.Can. J. For. Res. 31, 512–525 (2001).

16. Canadell, J. G. et al. in Terrestrial Ecosystems in a Changing World (eds Canadell,J. G., Pataki, D. E. & Pitkelka, L. F.) 59–78 (Springer, Berlin, 2007).

17. Denman, K. L. et al. in Climate Change 2007: The Physical Science Basis. Contributionof Working Group I to the Fourth Assessment Report of the Intergovernmental Panel onClimate Change (eds Solomon, S. et al.) 499–587 (Cambridge Univ. Press,Cambridge, 2007).

18. Ayres, M. P. & Lombardero, M. J. Assessing the consequences of global changefor forest disturbance from herbivores and pathogens. Sci. Total Environ. 262,263–286 (2000).

19. Kurz, W. A., Apps, M. J., Stocks, B. J. & Volney, W. J. A. in Biotic Feedbacks in theGlobal Climatic System: Will the Warming Feed the Warming? (eds Woodwell, G. M.& Mackenzie, F. T.) 119–133 (Oxford Univ. Press, Oxford, 1995).

20. Berg, E. E., Henry, J. D., Fastie, C. L., De Volder, A. D. & Matsuoka, S. M.Spruce beetle outbreaks on the Kenai Peninsula, Alaska, and Kluane NationalPark and Reserve, Yukon Territory: relationship to summer temperatures andregional differences in disturbance regimes. For. Ecol. Manage. 227, 219–232(2006).

21. Logan, J. A. & Powell, J. A. Ghost forests, global warming, and the mountain pinebeetle (Coleoptera: Scolytidae). Am. Ent. 47, 160–173 (2001).

22. McGuire, A. D. et al. Carbon balance of the terrestrial biosphere in the twentiethcentury: analyses of CO2, climate and land use effects with four process-basedecosystem models. G. Biogeoch. Cycles 15, 183–206 (2001).

23. Myneni, R. B. et al. A large carbon sink in the woody biomass of northern forests.Proc. Nat. Acad. Sci. 98, 14784–14789 (2001).

24. Wulder, M. A., Dymond, C. C., White, J. C., Leckie, D. G. & Carroll, A. L. Surveyingmountain pine beetle damage of forests: a review of remote sensingopportunities. For. Ecol. Manage. 221, 27–41 (2001).

25. Berryman, A. A. Dynamics of Forest Insect Populations: Patterns, Causes, Implications(Plenum, New York, 1988).

26. Bond-Lamberty, B., Peckham, S. D., Ahl, D. E. & Gower, S. T. Fire as the dominantdriver of central Canadian boreal forest carbon balance. Nature 450, 89–92(2007).

27. Kurz, W. A., Stinson, G. & Rampley, G. Could increased boreal forest ecosystemproductivity offset carbon losses from increased disturbances? Phil. Trans. R. Soc.B.. doi:10.1098/rstb.2007.2198 (2007).

28. Kurz, W. A., Apps, M. J., Webb, T. M. & McNamee, P. J. The Carbon Budget of theCanadian Forest Sector: Phase I. Information Report NOR-X-326 (Forestry Canada,Northwest Region, Northern Forestry Centre, Edmonton, 1992).

29. Kurz, W. A. & Apps, M. J. A 70-year retrospective analysis of carbon fluxes in theCanadian forest sector. Ecol. Appl. 9, 526–547 (1999).

30. Kull, S. et al. Operational-Scale Carbon Budget Model of the Canadian Forest Sector(CBM-CFS3) Version 1.0: User’s Guide (Natural Resources Canada, CanadianForest Service, Edmonton, 2006).

Supplementary Information is linked to the online version of the paper atwww.nature.com/nature.

Acknowledgements Funding for this study was provided by Canada’s ClimateChange Action Plan. We thank British Columbia Ministry of Forests and Range(MoFR) collaborators for providing data and detailed reviews: D. Draper,G. Lawrence, M. Boyce and D. Spittlehouse. S. Beukema provided assistance informatting MoFR data for use in CBM-CFS3. Fire and beetle projectionmethodologies were developed with the assistance of J. Metsaranta, S. Beukema,B. Stocks, B. Amiro, M. Flannigan, R. Landry, B. de Groot, K. Anderson, S. Taylor andT. Shore. Climate data were provided by D. McKenney. CBM-CFS3 modeldevelopment was assisted by T. White, G. Zhang, M. Magnan, E. Banfield, C. Shawand B. Simpson. Finally, we thank V. Nealis, D. Draper, D. Spittlehouse andA. Nussbaum for comments on an earlier draft.

Author Contributions All authors contributed to study design, modelparameterizations, and development of future projections; C.C.D., G.J.R., G.S. andE.T.N. conducted model simulations; W.A.K., C.C.D., G.S. and E.T.N. analysedresults; and all authors contributed to the manuscript.

Author Information Reprints and permissions information is available atwww.nature.com/reprints. Correspondence and requests for materials should beaddressed to W.A.K. (wkurz@nrcan.gc.ca).

LETTERS NATURE | Vol 452 | 24 April 2008

990Nature Publishing Group©2008

Mountain pine beetle and forest carbon feedback to climate change

W.A. Kurz, C.C. Dymond, G. Stinson, G. J. Rampley, E.T. Neilson, A. Carroll, T. Ebata, and L.

Safranyik.

Supplementary Information

Forest ecosystem model description and parameters

The Carbon Budget Model of the Canadian Forest Sector (CBM-CFS3) is an empirically driven,

stand- and landscape-level model for the simulation of forest carbon (C) dynamics. Earlier

versions of the model were described by Kurz et al.1 and by Kurz and Apps2. The current version

of the model builds upon the functionality of CBM-CFS2, and is the core model used in

Canada’s National Forest Carbon Monitoring, Accounting and Reporting System (NFCMARS)3.

Stand-level C dynamics are represented in the CBM-CFS3 using a system of pools that allow the

model to represent key ecological processes and that allow users to compare estimates of stocks

with field measurements. Each forest stand is represented by 21 C pools: 5 softwood biomass

pools, 5 hardwood biomass pools, 2 softwood snag pools, 2 hardwood snag pools, 4 additional

aboveground dead organic matter (DOM) pools, and 3 belowground DOM and soil C pools.

Aboveground biomass pools are estimated from merchantable volume using a system of regional

and species-specific volume to biomass conversion equations4. Belowground biomass is

predicted from aboveground biomass using stand-level regression equations for softwood and

hardwood species5. Tree bark is accounted for in each structural component. Each biomass C

pool is assigned forest-type and ecozone-specific litterfall and turnover rates.

CBM-CFS3 simulates the annual changes in the C stocks of each pool that occur due to growth,

litterfall and turnover, decomposition, natural disturbances, and forest management. The growth

module simulates net uptake of C from the atmosphere, i.e. net primary production (NPP), but

does not estimate gross primary production or autotrophic respiration. C is transferred to DOM

through annual turnover and litterfall rates or through event-driven disturbances. The DOM

pools are distinguished by the type of biomass they receive, and by unique decay and C transfer

rates. Decomposition is simulated using temperature-dependent, pool-specific decay processes

which transfer most of the decay losses to the atmosphere and a portion to a slow C pool.

SUPPLEMENTARY INFORMATION

doi: 10.1038/nature06777

www.nature.com/nature 1

Disturbances can contribute to C transfers and the loss of C from the forest ecosystem.

Harvesting causes transfers to the forest product sector that are reported as ecosystem losses in

the model output. Gaseous emissions, such as from a fire, are transferred to the atmosphere as

CO2, CH4 and CO and are represented as ecosystem losses in the model output. Disturbances

can affect stand age and the subsequent biomass and DOM C dynamics in the disturbed stand.

The simulation of initial conditions for soil and DOM C pools is generated by a spin-up

procedure that establishes DOM pool sizes based on disturbance history before the beginning of

the simulation2. Parameters have been updated for decomposition, fire impacts6, insect impacts

and climate. The model and all ecological parameters are freely available on the internet at

carbon.cfs.nrcan.gc.ca.

CBM-CFS3 was designed to meet the requirements for an operational tool7. While the CBM-

CFS2 was a research tool that provided no tools or support for simulation design or input/output

file processing, the CBM-CFS3 has a graphical user interface and includes data pre- and post-

processing tools and a detailed User’s Guide8. Furthermore, the spatial resolution of the model

has been greatly increased, allowing the explicit simulation of millions of stands in annual time

steps. When used in a monitoring role to assess past changes in C stocks, the model can be

spatially explicit and represent each polygon in a forest inventory with a corresponding database

record. Flexibility for modelling growth, management activities, disturbances and post-

disturbance dynamics has also been improved relative to CBM-CFS2.

In the CBM-CFS3, disturbances and forest management activities cause transfers of C between

pools and removals from the ecosystem. In this study, fires were simulated as stand-replacing

events. A portion of foliage and aboveground DOM pools were consumed, all remaining

biomass was killed, and standing snags were transferred to litter and coarse woody debris pools.

Fire C transfer coefficients are provided in Table S1. Harvest events were simulated as stand-

replacing clearcut events, with 85% of merchantable stemwood and 72% of snag stemwood C

transferred out of the ecosystem to the forest products sector. The rest of the C in biomass pools

was transferred to DOM pools. Harvest C transfer coefficients are provided in Table S2.

Transfers of C from the ecosystem to the forest products sector were reported by the model as

losses from the ecosystem, in accordance with current international C accounting guidelines9 and

doi: 10.1038/nature06777 SUPPLEMENTARY INFORMATION

www.nature.com/nature 2

good practice guidance provided by the Intergovernmental Panel on Climate Change (IPCC)10.

Following stand-replacing disturbance events, disturbed stands were regrown from age zero on

the original growth curve.

Mountain pine beetle impacts were simulated as partial-mortality events which killed a portion

of softwood (host) biomass pools. The beetle impacts were parameterized by translating the

damage classes reported in the aerial overview survey into percent of crown killed and then into

percent of softwood biomass killed for use in the CBM-CFS3 (Table S3). The stand mortality

classes were 5%, 10%, 30% or 50% per year (Table S4). These assignments take into

consideration the nature of the aerial survey data, the stand structure of the infested stands, and

the tendency of the beetle to infest larger diameter host trees first when attacking a stand. First,

the area distribution of damage within a damage class tends to be greater towards the lower end

of the range. Second, lighter damage tends to precede more severe damage during multi-year

infestations, with larger diameter trees being attacked first. Finally, we calibrated the damage

classes such that multi-year outbreaks in the model would result in the same cumulative

mortality levels that are observed in the field following beetle outbreak. Even in the most severe

cases, it is rare for mountain pine beetle to kill 100% of the pine component of infested stands.

We validated our calibrations of beetle impact for the portion of the forest subject to commercial

timber harvest against estimates produced by the British Columbia Ministry of Forests and

Range (MoFR) using a different modelling approach11. MoFR estimates of the loss of

merchantable wood volume resulting from the beetle outbreak were very similar to our estimates

of wood volume loss (Figure S1). MoFR estimates were generated by converting area infested

(by severity class) directly into merchantable volume impacts while our estimates were generated

by converting area infested into biomass C impacts. Our biomass C impact estimates were

translated into volume for comparison purposes by assuming a specific gravity of 0.397 g m-3

and 0.5 t C t-1 biomass for the merchantable stemwood.

Decomposition processes were simulated as the loss of C as gaseous emissions.and the transfer

of C between DOM pools. Decay rates determine the amount of organic matter that decomposes

in the DOM pools every year. Proportions determine the amount of C in decayed material that is

released to the atmosphere (Patm) or transferred to a more stable DOM pool (Ptrans) with (Patm+

doi: 10.1038/nature06777 SUPPLEMENTARY INFORMATION

www.nature.com/nature 3

Ptrans =1) (Table S5). The decay rates vary with mean annual temperature (MAT). We determined

the average MAT for each analysis unit by geographic intersection with 30 arc-second gridded

1961-1990 climate normals.

Input data

The British Columbia Ministry of Forests and Range (MoFR) supplied the forest inventory and

growth curve data used in this study. We assigned almost 700,000 forest stands to 86 spatial

analysis units that represented the administrative and ecological variability in the study area.

These were the same data that the MoFR uses for timber supply reviews, but were supplemented

with additional information describing forest lands outside of the timber harvesting landbase

(e.g. parks, dedicated wildlife habitat areas, riparian buffer zones, physically inoperable lands,

etc.) to complete the inventory of all forest lands in the study area.

Spatially explicit area burned data were extracted from the Canadian Large Fire Database12 and

the Canadian Wildland Fire Information System13 for 2000-2005. Spatially explicit data on areas

infested by mountain pine beetle were extracted from MoFR annual forest health surveys14 for

2000-2006. Fire and beetle disturbance datasets were geographically intersected with our 86

sptial analysis units in a Geographic Information System and disturbance target areas were

calculated for each disturbance type by spatial analysis unit and year for use in the CBM-CFS3

simulations. The MoFR also provided statistics on historical harvest, base projected harvest and

additional projected harvest response for each management unit in the study area (Figure S2). A

series of disturbance eligibility rules and sorting algorithms were used in the CBM-CFS3 to

select stands for disturbance from the forest inventory. Stands affected by fire were selected at

random. Mountain pine beetle host stands were defined as pine-leading between 40 and 250

years old, with the oldest stands selected first during the infestation. In the most severely infested

part of the study area, spruce-leading stands (which contain a significant pine component in the

study area) and stands as young as 20 years were also eligible for beetle disturbance. For areas

with severe or very severe mortality classes, the stand selection rules further specified that only

stands with infestation in prior years were eligible for beetle disturbance. The host characteristics

and stand selection algorithms were developed based on observed and published beetle

doi: 10.1038/nature06777 SUPPLEMENTARY INFORMATION

www.nature.com/nature 4

behaviour15,16. CBM-CFS3 selected stands for base harvest by sorting for the highest amount of

C in the merchantable stemwood pool. For the additional harvest, the model selected stands

based on the greatest amount of available C in the snag (standing dead tree) pools. These

simulation rules were developed in consultation with MoFR experts to ensure they adequately

reflected current forestry operations.

We projected the area infested by mountain pine beetle from 2007-2020 using random draws

from regionally calibrated probability distributions of outbreak area and duration. The

probability distributions were defined using interpretations of (i) spatial and temporal outbreak

dynamics during 1960-2006 (Figure S3), (ii) mortality and host statistics for the early part of the

current outbreak, (iii) status of remaining host, and (iv) expert judgment. The annual area

infested and severity of impact were modelled as temporally and spatially autocorrelated; area

infested in a given year affected the probability and size of the area infested in the next year.

We constructed 100 Monte Carlo projections of future years of the beetle outbreak using a

number of parameters defined for five modelling regions (Figure S4). The areas annually

infested by beetle in each region were single time series, drawn at random using regional

parameters (Table S4). Future area infested by low, moderate, severe and very severe mortality

levels was derived by describing for each region: (i) the range of possible outbreak durations in

years, (ii) the total area available as host, (iii) the area infested to date during the current

outbreak and during historical outbreaks, and (iv) the rate of expansion and contraction of area

infested (outbreak shape). The outbreak duration is the period of time in the outbreak cycle

during which the beetle population is high enough to cause visible damage to host trees. This

damage is usually assessed during forest health aerial overview surveys and can be used to

estimate host mortality. Spatially correlated regions had similar parameters for projecting the

remaining duration of the ongoing outbreak. The total area affected by an outbreak was the sum

of annual area affected by an outbreak. The duration and total area parameters are described

using probability density functions. The distributions to model these parameters were restricted

to simple distributions that require few data: uniform, triangular and normal17. For the mountain

pine beetle outbreak, the parameters were extended beyond the historical data to reflect the

unprecedented infestation that is currently underway and the entomologists’ best estimates of the

future trajectories of this outbreak.

doi: 10.1038/nature06777 SUPPLEMENTARY INFORMATION

www.nature.com/nature 5

Fire projections of future area burned were derived from regionally calibrated probability density

functions parameterised using historical fire statistics for the period 1959-1999. The fire

modelling regions were defined based on previous analysis of fire occurrence, area burned and

severity. The historical record of annual area burned in each stratum was extracted from the

Canadian Large Fire Database, which contains spatially explicit information on all fires greater

than 200 ha in size from 1959 to 1999 (Figure S5).

To format the time series of annual area burned into an observed cumulative probability

distribution, we defined 100 equally sized bins from zero to the maximum annual area burned in

the region. The frequency (i.e. number of years) was assigned to each bin and these were then

converted into the proportion of all years. We used the software BestFitTM version 4.5 (Palisade

Corporation, Newfield, New York) to fit probability density functions to the empirical

distributions. BestFitTM chooses from up to twenty-eight probability distribution functions, and

selects the appropriate parameters for that distribution using maximum-likelihood estimators. We

chose probability density functions region by region based on evaluation of the representation of

large, low probability events and the overall RMS error (Table S7). The probability of an annual

area burned greater than the largest area in the historical record in the region under consideration

was constrained to between 1% and 2%. That is, if the distribution had a greater than 2%

probability of generating an area burned greater than the maximum observed, it was rejected

(even if it was a valid fit statistically and had a low RMS error). Similarly, if a distribution had a

less than 1% probability of generating an area burned greater than the maximum observed it was

also rejected. We also truncated the projected maximum area burned to twice the historical

maximum. Our preliminary investigation indicated that the assumptions made about the

magnitude of projected maximum had significant impacts on the fire cycle (i.e. the number of

years that it takes to burn an area equal in size to the reference area). Truncating the maximum

annual area burned at twice the maximum observed in the historical record maintained the

historical fire cycle in repeated, randomly generated projections.

doi: 10.1038/nature06777 SUPPLEMENTARY INFORMATION

www.nature.com/nature 6

Table S1. Fire impacts as defined for this study by the proportion of carbon pool transferred to a new pool or emitted directly to the atmosphere as CO2, CO or CH4. Pre-disturbance pool Post-disturbance pool Proportion Hardwood foliage CH4 0.010 CO 0.090 CO2 0.900 Hardwood merchantable Hardwood stem snag 1.000 Hardwood other wood Hardwood branch snag 1.000 Hardwood fine roots Aboveground very fast DOM 0.036 CH4 0.005 CO 0.042 CO2 0.422 Belowground very fast soil 0.500 Hardwood coarse roots Aboveground fast soil 0.500 Belowground fast soil 0.500 Hardwood stem snag Medium DOM 1.000 Hardwood branch snag Aboveground fast soil 1.000 Softwood foliage CH4 0.010 CO 0.090 CO2 0.900 Softwood merchantable Softwood stem snag 1.000 Softwood other wood CO 0.025 CO2 0.225 Softwood branch snag 0.750 Softwood fine roots Aboveground very fast DOM 0.041 CH4 0.005 CO 0.041 CO2 0.413 Belowground very fast DOM 0.500 Softwood coarse roots Aboveground fast soil 0.500 Belowground fast soil 0.500 Softwood stem snag Medium DOM 1.000 Softwood branch snag Aboveground fast soil 1.000 Medium DOM CH4 0.009 CO 0.085 CO2 0.853 Unaffected portion 0.052 Aboveground very fast DOM CH4 0.010 CO 0.090 CO2 0.900 Aboveground fast soil CH4 0.010 CO 0.088 CO2 0.882 Unaffected portion 0.020 Aboveground slow soil CH4 0.009 CO 0.083 CO2 0.826 Unaffected portion 0.082 Belowground very fast DOM Unaffected portion 1.000 Belowground fast soil Unaffected portion 1.000 Belowground slow soil Unaffected portion 1.000

doi: 10.1038/nature06777 SUPPLEMENTARY INFORMATION

www.nature.com/nature 7

Table S2. Harvest impacts defined for this study by the proportion of carbon pool transferred to a new pool or removed to the forest product sector. Pools not explicitly included in this table are unaffected by harvest events. pre-disturbance pool post-disturbance pool Proportion Hardwood foliage Aboveground very fast DOM 1.000 Hardwood merchantable Forest products sector 0.850 Medium DOM 0.150 Hardwood other wood Aboveground fast DOM 1.000 Hardwood fine roots Aboveground very fast DOM 0.500 Belowground very fast soil 0.500 Hardwood coarse roots Aboveground very fast DOM 0.500 Belowground very fast soil 0.500 Hardwood stem snag Forest products sector 0.720 Medium DOM 0.280 Hardwood branch snag Aboveground fast DOM 1.000 Softwood foliage Aboveground very fast DOM 1.000 Softwood merchantable Forest products sector 0.850 Medium DOM 0.150 Softwood other wood Aboveground fast DOM 1.000 Softwood fine roots Aboveground very fast DOM 0.500 Belowground very fast soil 0.500 Softwood coarse roots Aboveground fast DOM 0.500 Belowground fast soil 0.500 Softwood stem snag Forest products sector 0.720 Medium DOM 0.280 Softwood branch snag Aboveground fast DOM 1.000

Table S3. Mountain pine beetle aerial overview survey damage classes are defined by the percent of trees recently killed*. The CBM-CFS3 simulates beetle impacts for each damage class as percent softwood biomass killed per year.

Damage Class Percent of trees Killed

Percent softwood biomass killed

Trace <1 not simulated Light 1-10 5

Moderate 11-30 10 Severe 31-50 30

Very Severe >50 50

* Procedures for Landscape-level Forest Health Factor Surveys http://www.for.gov.bc.ca/tasb/legsregs/fpc/fpcguide/health/gfhs0004.htm#E10E4

doi: 10.1038/nature06777 SUPPLEMENTARY INFORMATION

www.nature.com/nature 8

Table S4. Mountain pine beetle impacts defined for this study by the proportion of carbon pool transferred to a new pool in CBM-CFS3 for light (L), moderate (M), severe (S) and very severe (VS) damage classes. Pools not explicitly included in this table are unaffected by beetle infestation.

Proportion Pre-disturbance pool Post-disturbance pool L M S VS

Softwood foliage Aboveground very fast DOM 0.050 0.100 0.300 0.500 Unaffected portion 0.950 0.900 0.700 0.500 Softwood merchantable Softwood stem snag 0.050 0.100 0.300 0.500 Unaffected portion 0.950 0.900 0.700 0.500 Softwood other wood Softwood branch snag 0.050 0.100 0.300 0.500 Unaffected portion 0.950 0.900 0.700 0.500 Softwood fine roots Aboveground very fast DOM 0.025 0.050 0.150 0.250 Belowground very fast soil 0.025 0.050 0.150 0.250 Unaffected portion 0.950 0.900 0.700 0.500 Softwood coarse roots Aboveground fast DOM 0.025 0.050 0.150 0.250 Belowground fast soil 0.025 0.050 0.150 0.250 Unaffected portion 0.950 0.900 0.700 0.500

Table S5. Parameters used to simulate dead organic matter (DOM) dynamics in this study. Decay rates listed are base decay rates for 10oC; all except slow pools have a Q10 of 2; slow pools have a Q10 of 0.9; AG = aboveground, BG = belowground.

DOM Pool

Turnover rate

(% of C)

Turnover receiving

pool

Decay rate (% of C)

Patm

Decay receiving

pool (Ptrans = 1-Patm)

Stem snags 0.032 Medium 0.0187 0.830 AG Slow Snag branches 0.100 AG Fast 0.0718 0.830 AG Slow Medium 0.0374 0.830 AG Slow AG Fast 0.1435 0.830 AG Slow AG Very fast 0.5000 0.830 AG Slow AG Slow 0.006 BG Slow 0.0032 1.000 BG Fast 0.1435 0.830 BG Slow BG Very fast 0.5000 0.830 BG Slow BG Slow 0.0032 1.000

doi: 10.1038/nature06777 SUPPLEMENTARY INFORMATION

www.nature.com/nature 9

Table S6. Parameters for probability density functions (PDFs) describing different aspects of the mountain pine beetle projections. Modelling region (see Figure S4)

PDF for Length of Insect Outbreak (years)

PDF for Total Area of Insect Outbreak (percent of host area)

Host area (kha)

Start Year of Current

Outbreak

Sum of Area from start

year to 2006 (kha)

Contiguous host f(x)=1/(25-14) ^ f(x)=1/(600-560) ^ 5,295 1996 14,978

Discontiguous host f(x)=1/(25-18) ^ f(x)=1/(750-400) ^ 5,186 1999 9,527

Patchy host (NW) f(x)=2(x-14)/(33)*

when x<17 f(x)=2(25-x)/(88) * when x>17

f(x)=1/(100-50) ^ 132 2001 17

Patchy host (SE) f(x)=2(x-14)/(22)* when x<16 f(x)=2(25-x)/(99)* when x>16

f(x)=1/(250-150) ^ 1,264 2000 475

Climate limited f(x)=2(x-8)/(36)* when x<11 f(x)=2(20-x)/(108)* when x>11

f(x)=1/(200-130) ^ 1,158 2002 358

* Triangular distribution f(x)=2(x-min)/((mode-min)(max-min)) when x<mode f(x)=2(max-x)/((max-mode)(max-min)) when x>mode ^ Uniform distribution f(x)=1/(max-min)

Table S7. Parameters for annual burned area projections. Modelling region (see Fig. S4)

PDF for annual area burned (ha) Max. annual area burned (ha)

Probability of maximum occurring in a random draw

1^ f(x)=[1/((0.60652)(17857)Γ)] (x/17857) 0.60652-1 (e-x/17857) 97,687 0.14

2^ f(x)=[1/((0.1892)(45673)Γ)] (x/45673)0.1892−1 (e-x/45673) 226,710 0.03

3* f(x)= [(0.2985x0.2985-1)/ 1070.30.2985] (e-(x/1070.3)^0.2985) 130,494 1.51

* Weibull distribution

f(x)= [(αxα−1

)/βα] (e-(x/β)^α) ^ Gamma distribution, Γ= Gamma function f(x)=[1/(αβΓ)] (x/β)

α−1 (e-x/β)

doi: 10.1038/nature06777 SUPPLEMENTARY INFORMATION

www.nature.com/nature 10

Figure S1. Comparison of our model estimates of cumulative volume killed by the mountain pine beetle against timber volume killed estimates produced using different methods by the British Columbia Ministry of Forests and Range. This comparison is only for the portion of the study area subject to commercial harvest.

doi: 10.1038/nature06777 SUPPLEMENTARY INFORMATION

www.nature.com/nature 11

Figure S2. Base harvest statistics used in all simulations and additional harvest in response to the beetle outbreak. Solid squares (2000 – 2005) denote statistics; circles and triangles mark projections (2006-2020) for base harvest and additional harvest, respectively. After 2015, the additional harvest scenario returns to the base harvest level.

doi: 10.1038/nature06777 SUPPLEMENTARY INFORMATION

www.nature.com/nature 12

Figure S3. Historical record of area annually burned from Stocks et al. (a) and area infested by mountain pine beetle (b) in British Columbia from Taylor et al.18

doi: 10.1038/nature06777 SUPPLEMENTARY INFORMATION

www.nature.com/nature 13

Figure S4. Mountain pine beetle modelling regions (shaded areas) were defined based on spatial distribution of host on the landscape and on climate characteristics. Contiguous host resulted in most severe infestations with patchy host providing less than optimal forest conditions for infestation. Climate limited host area is characterized by colder winters and less than optimal climatic conditions. Discontiguous host area had suitable climatic conditions, but with less than optimal host configuration on the landscape, interspersed with areas of non-host species. The study area also contained three discrete fire modelling regions (1, 2, and 3), delineated by thick black borders. The northwest corner of fire region (1) (white) contained unsuitable host and climate conditions for beetle. Fire area 1 overlapped with three beetle modelling regions, while each of the other two fire regions overlapped with only one beetle modelling region.

doi: 10.1038/nature06777 SUPPLEMENTARY INFORMATION

www.nature.com/nature 14

Figure S5. (a) percentiles from 100 random draws from three fire probability density functions based on historic data from 1959 - 1999. Solid circles (2000 – 2005) denote statistics. (b) Area disturbed by fire used in beetle scenario analysis.

doi: 10.1038/nature06777 SUPPLEMENTARY INFORMATION

www.nature.com/nature 15

References for Supplementary Information

1 Kurz, W. A., Apps, M. J., Webb, T. M. & McNamee, P. J. The Carbon Budget of the Canadian Forest Sector: Phase I. Information Report NOR-X-326 (Forestry Canada, Northwest Region, Northern Forestry Centre, Edmonton, 1992).

2 Kurz, W. A. & Apps, M. J. A 70-year retrospective analysis of carbon fluxes in the Canadian forest sector. Ecol. Appl. 9, 526-547 (1999).

3 Kurz. W. A. & Apps, M. J. Developing Canada’s National Forest Carbon Monitoring, Accounting and Reporting System to meet the reporting requirements of the Kyoto Protocol. Mitig. Adapt. Strat. Global Change 11, 33-43 (2006). 4 Boudewyn, P., Song, A. & Gillis, M.A. Model-based, volume-to-biomass conversion for forested and vegetated land in Canada. Information Report BC-X-411. (Natural Resources Canada, Canadian Forest Service, Victoria, 2007). 5 Li, Z., Kurz, W. A., Apps, M. J. & Beukema, S. J. Belowground biomass dynamics in the Carbon Budget Model of the Canadian Forest Sector: recent improvements and implications for the estimation of NPP and NEP. Can. J. For. Res. 33, 126-136 (2001). 6 de Groot, W. J., et al. Estimating direct carbon emissions from Canadian wildland fires. Int. J. Wildl. Fire 16, 593-606 (2007). 7 Kurz, W. A., Apps, M., Banfield, E. & Stinson, G. Forest carbon accounting at the operational scale. For. Chron. 78, 672-679 (2002). 8 Kull, S., et al. Operational-Scale Carbon Budget Model off the Canadian Forest Sector (CBM-CFS3) Version 1.0: USER’S GUIDE. (Natural Resources Canada, Canadian Forest Service, Edmonton, 2006). 9 IPCC. Revised 1996 Guidelines for National Greenhouse Inventories. (IPCC/OECD/IEA, Bracknell, UK, 1997). 10 IPCC. Good Practice Guidance for Land Use, Land-Use Change, and Forestry. (Cambridge University Press, Cambridge, UK, 2000). ? 2003? 11 Walton, A., et al. Update of the Infestation projection Based on the 2006 Provincial Aerial Overview of Forest Health and Revisions to the Model (BCMPB.v4). (British Columbia Ministry of Forests and Range, Victoria, 2007). http://www.for.gov.bc.ca/hre/bcmpb/Year4.htm 12 Stocks, B.J., et al. Large forest fires in Canada, 1959-1997. J. Geophys. Res. Atmos. 108, FFR5.1-FFR5.12 (2003). 13 Lee, B.S., et al. Information systems in support of wildland fire management decision making in Canada. Comput. Electr. Agr. 37, 185-198 (2002).

doi: 10.1038/nature06777 SUPPLEMENTARY INFORMATION

www.nature.com/nature 16

14 Westfall, J. 2006 Summary of Forest Health Conditions in British Columbia. (British Columbia Ministry of Forests and Range, Forest Practices Branch, Victoria, BC, 2006).

15 Maclauchlan, L. Status of Mountain Pine Beetle Attack in Young Lodgepole Pine in the Central Interior of British Columbia. (British Columbia Ministry of Forests and Range, Victoria, BC, 2006). http://www.for.gov.bc.ca/rsi/foresthealth/PDF/Young%20Pine%20Report_CF_final.pdf 16 Safranyik, L. & Carroll, A.L. The biology and epidemiology of the mountain pine beetle in lodgepole pine forests. In: The Mountain Pine Beetle: A Synthesis of Biology, Management and Impacts in Lodgepole Pine. (Natural Resources Canada, Canadian Forest Service, Victoria, BC, 2006). 17 Evans, M., Hastings, N. & Peacock, B. Statistical Distributions. Third Edition. (Wiley Series in Probability and Statistics, John Wiley & Sons, Toronto, ON, 2000). 18 Taylor, S. W. & Carroll, A. L. in Mountain Pine Beetle Symposium: Challenges and Solutions BC-X-399 (eds Shore, T. L., Brooks, J. E. & Stone, J. E.) 41-51 (Natural Resources Canada, Canadian Forest Service, Victoria, 2004).

doi: 10.1038/nature06777 SUPPLEMENTARY INFORMATION

www.nature.com/nature 17

02/05/08 10:58 Forest carbon switch : Nature

Page 1 of 1http://www.nature.com/nature/journal/v452/n7190/edsumm/e080424-12.html

Editor's Summary24 April 2008

Forest carbon switch

The forests of British Columbia are suffering a severe infestation with the mountain pine beetle (Dendroctonusponderosae). Climate change is thought to have contributed to the severity of this outbreak by allowing it toexpand its range into formerly inhospitable areas. An analysis of the likely impact of the outbreak during theperiod 2000 to 2020 suggests that it will convert the forest from a small net carbon (C) sink to a large net Csource. This change — and similar effects caused by other insect pests and forest fires — could put NorthAmerican forest carbon sinks at risk, and should be taken into account when modelling the impact of climatechange on carbon cycling.

AUTHORSMaking the paper: Werner KurzMountain pine beetles contribute to carbon release and climate change.doi:10.1038/7190xa

Full Text (/nature/journal/v452/n7190/full/7190xa.html) | PDF (105K) (/nature/journal/v452/n7190/pdf/7190xa.pdf)

LETTERMountain pine beetle and forest carbon feedback to climate changeW. A. Kurz, C. C. Dymond, G. Stinson, G. J. Rampley, E. T. Neilson, A. L. Carroll, T. Ebata & L. Safranyikdoi:10.1038/nature06777

First paragraph (/nature/journal/v452/n7190/abs/nature06777.html) | Full Text (/nature/journal/v452/n7190/full/nature06777.html) |PDF (917K) (/nature/journal/v452/n7190/pdf/nature06777.pdf) | Supplementary information(/nature/journal/v452/n7190/suppinfo/nature06777.html)

Search This journal go Advanced search

02/05/08 11:04 Tages-Anzeiger

Page 1 of 1http://epaper2.tagesanzeiger.ch/ee/tages/default.php?pSetup=tages

Käferplage macht Klimaschutz zunichte Victoria. – Die Attacke eines einheimischen Baumkäfers wird die kanadischen Wälder aller Voraussicht nach zueiner bedeutenden Kohlendioxid-Quelle machen. Die Bergkieferkäfer haben sich bereits infolge des Klimawandelsweiter in den Wäldern ausgebreitet als zuvor. Sie raffen viele Millionen Bäume dahin. Das CO2 aus den zer-fallenden BaumleichenverstärktdenTreibhauseffekt. Diese Analyse stammt von Werner Kurz, Ökologe bei denkanadischen Forstbehörden am Pacific Forestry Center in Victoria (Kanada). Den Rechnungen seiner Gruppezufolge werden die Käferplage und seine Folgen, etwa Waldbrände, in den Jahren von 2000 bis 2020 insgesamt270 Millionen Tonnen CO2 auf 374 000 Quadratkilometern freisetzen. Die sei jene Treibhausgas- Menge, dieKanada gemäss KyotoProtokolls bis 2012 sparen will. ( DPA/FWT) «Nature», Bd. 452, S. 987

Powered by TIOWS Copyright (c) Tages-Anzeiger