The Natural Disturbance Regime: The Natural Disturbance Regime: Implications for Forest ManagementImplications for Forest Management

Glen W. Armstrong

University of Alberta

CIFFC Science Forum4 November 1999

Presentation OutlinePresentation Outline

ContextTwo studies

– A characterization of the natural disturbance regime

– Planning for timber and wildlife under uncertainty

Concluding comments

ContextContext

Sustainable forest management– Biodiversity

Coarse filter– Manage for a natural (natural-appearing)

forest structure: this may accommodate the majority of species

Extractive uses are important

Context (cont’d) Context (cont’d)

Malcolm Hunter. 1993. Natural fire regimes as spatial models for managing boreal forests. – Frequency of harvest– Size and distribution of openings– Residual organic matter

Characterization of the Natural Characterization of the Natural Disturbance Rate of Alberta’s Disturbance Rate of Alberta’s Boreal Mixedwood ForestBoreal Mixedwood Forest

Glen Armstrong

Natural Disturbance RegimesNatural Disturbance Regimes

Murphy– Based on analysis of age class data– 2%/year– Independent of stand characteristics

Cumming– Based on analysis of fire history– ~0.5%/year– Dependent on softwood content

Alternative Equilibrium Age Alternative Equilibrium Age Class DistributionsClass Distributions

0

0.005

0.01

0.015

0.02

0.025

0 100 200 300 400 500 600

Age (years)

Pro

port

iona

l Are

a

2.0%0.5%

OutlineOutline

Statistical characterization of the natural disturbance rate

Monte Carlo simulations– Confidence intervals– Age classes

Conclusions and implications

Study Area and ParametersStudy Area and Parameters

8.6 million ha Fire history data

1961-1995 Annual area of

lightning caused fires that started in the study area

Annual Area BurnedAnnual Area Burned

0

100

200

300

400

500

1961

1963

1965

1967

1969

1971

1973

1975

1977

1979

1981

1983

1985

1987

1989

1991

1993

1995

Year

Are

a (`

000

ha)

Annual Area Burned (log scale)Annual Area Burned (log scale)

0.001

0.01

0.1

1

10

100

100019

61

1963

1965

1967

1969

1971

1973

1975

1977

1979

1981

1983

1985

1987

1989

1991

1993

1995

Year

Are

a (`

000

ha)

CDF for Burn RateCDF for Burn Rate

0

0.2

0.4

0.6

0.8

1

1.00E-07 1.00E-05 1.00E-03 1.00E-01

Burn Rate

F(r

ate)

Monte Carlo SimulationsMonte Carlo Simulations

Mean Rate Confidence LimitsMean Rate Confidence Limits

95% confidence limits for estimates mean disturbance rates

10 000 sets of draws for each sample period of interest

Calculate the mean for each setDetermine the 2.5 and 97.5

percentiles

Confidence Interval ResultsConfidence Interval Results

Sample Period Lower limit Upper limit 50 years 0.18 % 3.4 % 230 years 0.44 % 2.1 %

The 2 % and 1/2 % rates are within both confidence intervals

Age Class Distributions Age Class Distributions

Starting age class distribution1000 years of disturbance

– Random draw of annual burn rate– Area burned in each age class

proportional to area100 replications

Four Age Class OutcomesFour Age Class Outcomes

00.20.40.60.8

0 40 80 120

160

200 0

0.20.40.60.8

0 40 80 120

160

200

00.20.40.60.8

0 40 80 120

160

200 0

0.20.40.60.8

0 40 80 120

160

200

ConclusionsConclusions

The annual burn rate for the study area can be characterized by a simple two-parameter distribution

The burn rate is highly variable: the mean rate of disturbance cannot be determined precisely

There is no equilibrium age-class structure

ImplicationsImplications

No “ecologically correct” disturbance rate or age class distribution exists for my study area

Determination of burn rates based on age class distributions is highly questionable

Planning for Timber and Wildlife Planning for Timber and Wildlife Habitat Under UncertaintyHabitat Under Uncertainty

Glen Armstrong, Jim Beck,Vic Adamowicz, Fiona Schmiegelow, and Steve Cumming

Modeling StrategyModeling Strategy

Describe the existing forest using state variables

Relate this description to habitat Use Monte Carlo simulation to project the

range of natural variability in habitat area Use simulation results to guide constraint

selection for optimization model Quantify trade-offs between timber and

habitat in the context of RNV

Forest State DescriptorsForest State Descriptors

Cover type1) pine

2) white spruce

3) aspen

4) mixed

5) black spruce

Habitat stage1) establishment

2) max density

3) max crown closure

4) max basal area

5) mature

6) overmature

See...

Cumming, S.G. et al. 1994. Potential conflicts between timber supply and habitat protection... For. Ecol. Manage. (68) 281-302.

Selected Wildlife SpeciesSelected Wildlife Species

Pine martenMeadow voleBroad-winged hawkThree-toed woodpeckerBlack-throated green warbler

(BTGW)

Pine Marten Habitat PreferencesPine Marten Habitat Preferences

Habitat stageCover type 1 2 3 4 5 6Pine 2 2 2 2White Spruce 2 3 4 6Mixed 2 3 4

Starting Forest InventoryStarting Forest Inventory

0

50

100

150

200

250

300

350

Age

Are

a ('

000

ha)

SbPMxAwSw

A portion of the DMI FMA

888 000 ha of the net merchantable land base

Large spike at the 60 year age class

Simulated Habitat ProjectionsSimulated Habitat Projections

Pine Marten Habitat (5+)

0

20

40

60

80

100

0 50 100

Years

Are

a

BTGW Habitat (5+)

0

40

80

120

160

0 50 100

Years

Are

a

Optimization RunsOptimization Runs

Maximize net present value of timber harvest

s.t. non-declining yield constraintss.t. habitat constraint levels

– None (business as usual)– Habitat area for all species at different

percentiles

Harvest With Percentile Harvest With Percentile ConstraintsConstraints

Pine marten

0

25

50

75

100

0 25 50 75

Year

Are

a

BTGW

0

50

100

150

200

0 25 50 75

Year

Are

a

Trade-off AnalysisTrade-off Analysis

Habitat Constraint Level

None 2.5% 40%

NPV (106 $) 1 363 1 164 622

Swd AAC (103 m3) 829 707 143

Hwd AAC (103 m3) 1 239 1 045 129

What Have We Done?What Have We Done?

Developed a system that– Projects probability distributions of wildlife

habitat and/or forest structure through time– Incorporates natural disturbance– Allows for comparisons between managed

outcomes and the range of natural variability– Explicitly quantifies trade-offs between

financial values and wildlife habitat

What Should We Do?What Should We Do?

Incorporate successionRefine the disturbance modelRefine the timber cost modelStochastic optimizationExplore the use of the system in a

public consultation context

Final SummaryFinal Summary

Natural Disturbance Management is likely to have a large impact on timber supply

The “natural” rate of disturbance and age class structure do not exist

Optimization approaches that consider variablity may be useful tools

AcknowledgementsAcknowledgements

Sustainable Forest Management Network

Alberta-Pacific Forest Industries Inc.Daishowa-Marubeni International

Ltd.

Acknowledgements (cont’d)Acknowledgements (cont’d)

Vic Adamowicz, Jim Beck,Steve Cumming, Rick Pelletier, and Fiona Schmiegelow

Stan Boutin, Darrell Errico, Daryll Hebert, Ellen Macdonald, Peter Murphy, Bill Reed, and Brad Stelfox