Cost estimates for carbon sequestration from fast growing poplar plantations in Canada

www.elsevier.com/locate/forpol

Forest Policy and Economics 6 (2004) 345–358

Cost estimates for carbon sequestration from fast growing poplar

plantations in Canada

Daniel W. McKenneya,*, Denys Yemshanova, Glenn Foxb, Elizabeth Ramlalb

aCanadian Forest Service, Great Lakes Forestry Centre, 1219 Queen Street East, Sault Ste. Marie, ON, Canada P6A 2E5bDepartment of Agricultural Economics and Business, University of Guelph, Guelph, ON, Canada N1G 2W1

Abstract

With concern over human activities affecting the Earth’s climate, the potential role of forests to sequester carbon is of

growing interest to national policy-makers. Countries like Canada may be able to use afforestation of marginal agricultural lands

to sequester carbon in a cost-effective manner. A spatial simulation study that links the biology and economics of afforestation

of marginal agricultural lands in Canada using a modified Hartman-type model is presented. The model recognizes wood

production and carbon sequestration and calculates ‘break-even’ carbon prices inclusive of an opportunity cost for agricultural

production values. A simplified carbon budget-tracking algorithm is used that predicts accumulation of carbon in soil, litter,

standing aboveground and root biomass, carbon flows among ecosystem components and CO2 release from biomass and forest

products decay. Variables are represented as probability distribution functions. Monte-Carlo simulation and sensitivity analysis

techniques are used to help assess both biological and economic uncertainty. Some results are presented for Canada and issues

identified to improve model results (e.g. spatially varying estimates of productivity). Substantively more land is attractive for

afforestation in Western Canada than Eastern Canada but results are highly sensitive to growth and yield assumptions and

spatial variation in agricultural production opportunity costs.

D 2004 Elsevier B.V. All rights reserved.

Keywords: Carbon; Cost-benefit analysis; Geographic information system; Afforestation; Parameter uncertainty; Spatial analysis; Break-even

carbon prices

1. Introduction

Climate change from anthropogenic emissions of

greenhouse gases, such as CO2, is believed by some

to be one of the most significant environmental

concerns of the 21st century (IPCC, 2001; see Essex

and McKitrick, 2002 for an alternative view). In

1389-9341/$ - see front matter D 2004 Elsevier B.V. All rights reserved.

doi:10.1016/j.forpol.2004.03.010

* Corresponding author. Tel.: +1-705-541-5569; fax: +1-705-

541-5700.

E-mail addresses: [email protected] (D.W. McKenney),

[email protected] (D. Yemshanov), [email protected]

(G. Fox), [email protected] (E. Ramlal).

Canada, annual emissions of CO2 were estimated

as 5.96�108 tons of CO2-equivalent (1.62�108 tons

of C equivalent) in 1990 (Jacques, 1998) and

7.26�108 tons of CO2-equivalent (1.97�108 tons

of C equivalent) in 2000 (Environment Canada,

2001). A specific target of a 25% reduction of

CO2 emissions below 1990 levels by the year 2012

has been set through the so-called Kyoto protocol.

Canada’s federal government has recently ratified the

Kyoto protocol. To meet the Kyoto target, Canadian

emissions must be 5.60�108 tons of CO2 equivalent,

which is 29% below the emission estimates expected

for the commitment period.

D.W. McKenney et al. / Forest Policy and Economics 6 (2004) 345–358346

The potential for using forests as a short-term

method for reducing atmospheric concentrations of

carbon dioxide is well recognized (van Kooten et

al., 1992; Sedjo et al., 1995; Cannell, 1999). The

Kyoto protocol allows countries to claim as a credit

against their emissions any verifiable amount of

carbon sequestered as a result of afforestation

undertaken since 1990. All carbon assimilated by

newly planted, and importantly, unanticipated prior

to Kyoto, sites over the period 2008–2012 may be

included. Given Canada’s large agricultural land-

base (greater than 67.5 million ha—StatCan, 2001),

one possible option for credits is clearly afforesta-

tion of agricultural lands (van Kooten et al., 1999;

Stevens et al., 2002). Although the physical aspects

of the carbon cycle are generally well understood,

policy development requires integration of this

knowledge in spatial models that link biological

processes with economic models. This will help

assess the cost effectiveness of afforestation relative

to other policy choices.

In this article, we present a spatial model of

possible afforestation activities for Canada. The

primary metric of assessment is mapped break-even

carbon prices. It is currently unclear what the price

of carbon credits will be. Included in the calculations

are plantation establishment and maintenance costs, a

spatially varying opportunity cost for agricultural

production, timber and bioenergy production values.

Thus, our model calculates what the price of carbon

would have to be to justify afforestation activities on

economic efficiency grounds if the timber and bio-

energy values do not generate positive net present

values on their own. For countries like Canada that

have relatively slow growth rates, joint products

obtained from afforested land may create interest

for shared investments in plantations. Lastly, there

is a need for models that integrate biological and

economic uncertainties (e.g. Paoli and Bass, 1997;

Krcmar et al., 2001). In our model, the main bio-

logical and economic parameters and variables (e.g.

prices, biomass estimates, product conversion rates)

are represented as probability distributions. Sensitiv-

ity analysis is also used to characterize the relative

impacts of variation in model parameters on the

break-even carbon price.

Five growth and yield scenarios for fast growing

hybrid poplar species are presented using a model

version that examines wood, bioenergy and carbon

sequestration values. These scenarios are based on a

policy goal of 20-year rotation periods. Some policy

and research implications of our findings and areas of

future research are discussed.

2. Methods and data

2.1. Model overview

The model calculates costs and benefits of wood

production and carbon sequestration inclusive of an

opportunity cost for agricultural production values,

and then determines the price of carbon that would

make afforestation financially attractive. This car-

bon price (‘break-even’ price) is portrayed as a

raster (grid) map. The spatial portrayal of cost-

benefit results allows for identification and com-

parison of afforestation suitability across large

areas. Output maps are compatible with Geographic

Information System (GIS) programs, and can be

further summarized and analyzed using standard

GIS techniques.

The model integrates a simplified carbon budget

tracking algorithm that predicts accumulation of

carbon in soil, litter, standing aboveground and root

biomass, carbon flows among ecosystem components

and CO2 release from biomass and forest products

decay using raster spatial input data to perform

simulations. The minimum data element size (‘grid

cell’ in GIS terms) determines the spatial resolution

of the model. This adds convenience, as our aim is

to apply the model at various spatial scales ranging

from whole-country simulations to fine-scale case

studies.

Net present value calculations (NPVAfforestation)

are based on present values for wood production

(PVTimber), carbon sequestration benefits (PVCarbon

sequest.), benefits associated with fossil fuel substitution

by burning wood as a bioenergy source (PVBioenergy)

less the opportunity costs when converting the land

use from crops or pasture (PV Ag land value):

NPVAfforestation ¼ PVTimber þ PVCarbon seq:

þ PVBioenergy�PVAg land value: ð1Þ

PVTimber þ PVCarbon seq: ¼ ptbvðtÞ � CEst½ e�rT þZ T

0

pcs/cvVðtÞe�rtdt �Z T

0

CTendingðtÞe�rtdt

� CEst; ð2Þ1� e�rT

The present values of timber harvests (PVTimber)

and the flow of carbon sequestration benefits

(PVCarbon seq.) were calculated using a modified

Hartman model (Hartman, 1976; van Kooten et al.,

1995, see also Bowes and Krutilla, 1989; Pearce,

1994):

D.W. McKenney et al. / Forest Policy and Economics 6 (2004) 345–358 347

where t is time, T is the rotation age, r is a discount

rate, /c is a conversion factor from biomass volume to

carbon, vV(t) is an annual tree stand biomass volume

increment, ptb is the price paid for standing timber,

and pcs is the ‘carbon price’. Costs include plantation

establishment, CEst (the second CEst term being the

initial plantation establishment cost), and plantation

tending, CTending(t) (simulated as a function of stand

age). Instead of using fixed carbon price, the model

calculates the value at which NPVAfforestation=0. This

break-even carbon price is calculated in terms of

discounted dollars per metric ton of carbon.

The Hartman model can be used to identify the

rotation age that maximizes timber values and flows

of the specified non-wood (amenity) benefits (Hart-

man, 1976). In our case the rotation age is fixed to 20

years because we are assessing a specific proposed

program—the establishment of fast growing, short

rotation poplar plantations. The model provides maps

of mean, maximum, minimum break-even carbon

prices and present values. Maps can also be provided

for the mean, minimum, maximum and standard

deviation of each output variable and be further

summarized using conventional GIS programs (e.g.

identifying areas available at given carbon prices in

certain regions, etc.).

2.2. Agricultural land values

Afforestation potentially competes with agricul-

tural production. Sedjo et al. (1995) notes the

importance of economic assessments including the

full costs of land and labour for carbon sequestra-

tion projects and that some studies oversimplify this

(e.g. Richards, 1992; de Jong et al., 2000). Agri-

cultural land values vary widely across Canada. To

ensure some consistency of opportunity cost esti-

mates, we used the 2001 Canadian Census of

Agriculture (StatCan, 2001) present value of annual

land rental expenses (PVR):

PVR ¼ 1

r� BR

SR; ð3Þ

where BR is total rental and lease expenses for

agricultural land, SR is the area of agricultural land

rented annually, and r is the discount rate.

The 2001 Census of Agriculture database provides

total rental/lease expenses and the areas of agricultural

land rented annually (i.e. BR and SR) at the level of

Consolidated Census Subdivisions (CCS units, Stat-

Can, 2001). These subdivisions can be plotted as

polygons with areas from 25 km2 (in highly populated

urban areas) to several million hectare (in remote

northern areas). Thus, each CCS unit has its own

PVR value. Some CCS units did not have rental/lease

expenses estimates due to confidentiality reasons

(StatCan, 2001). In that case, rental/lease expenses

summarized at coarser levels were used (StatCan,

2001).

To define a distribution of potential agricultural

production opportunity costs we estimated variation in

PVR from adjacent areas (50 km radius for southern

agricultural regions areas with high population densi-

ty, and 100 km for northern areas, where agricultural

land covers less than 30%). Mean, minimum, maxi-

mum and standard deviation estimates were used to

define the distribution of agricultural land opportunity

costs (mean and standard deviation were used to

define the shape of the distribution, minimum and

maximum—to define variation range). For some areas

in northern parts of the country, little variation was

detected due to lack of data. For these locations, we

D.W. McKenney et al. / Forest Policy and Economics 6 (2004) 345–358348

assumed F20% range. Because no assumption about

the shape of the distribution was made, we used the

simplest, a triangular distribution (see Stevens et al.,

2002).

As would be expected, the highest agricultural land

values were observed close to urban centers and

populated areas and exceeded $15 000 ha�1. The

lowest estimates were located in the northern part of

the Prairie provinces, close to the south-boreal forest

transition zone and were below $300 ha�1.

2.3. Carbon sequestration calculations

Three aboveground C pools (merchantable timber

biomass, non-merchantable aboveground biomass,

and forest floor biomass) and two belowground C

pools (soil carbon and root system carbon) are repre-

sented using the algorithm described in van Kooten et

al. (1999). Our model does not simulate fine-scale

carbon flows between intermediate ecosystem pools

such as slow- and fast-decaying soil pools or operate

on a daily step like some models (e.g. Peng et al.,

2002). In fact most carbon sequestration models used

in economic studies have relatively simple carbon

tracking algorithms and use only major biophysical

parameters, e.g. growth rates and conversion ratios

into the carbon equivalent (van Kooten et al., 1999;

Bateman and Lovett, 2000; Creedy and Worzbacher,

2001). Accumulation of carbon in soil, litter, biomass

and CO2 release is simulated via redistributing carbon

among main ecosystem pools (see also Kurz et al.,

1992).

Stand biomass was derived from growth and

yield curves. Above ground biomass equivalent

was estimated using yield curve estimates multiplied

by 1.5 (Kurz et al., 1992). Carbon in root biomass

was calculated as a proportion of living stand

biomass (0.43 for hybrid poplar from Guy and

Benowicz, 1998). Conversion factors for carbon

transfers between pools were represented as ratios.

This approach is also used in other biophysical

carbon budget models (e.g. Kurz and Apps, 1999).

Note, however, that these ratios are represented as

distributions.

The amount of carbon in the litter and forest floor

follows Smith and Heath (2002) as a species-specific

function of stand age and site conditions. Decompo-

sition of forest floor biomass existing prior to harvest

is an exponential function of stand age (i.e. time since

last clear-cut) and a forest floor residence time (see

Smith and Heath (2002) for details).

Conversion of agricultural land to forest planta-

tions should lead to an increase in soil carbon content.

This is assumed to occur over a specified period after

converting agricultural land to plantations and then

remain stable. During that period, the annual rate of

carbon accumulation was assumed to be linear. For

hybrid poplar plantations, we used a period of 50

years and rates for the Prairies 0.96 and 1.35 ton/ha

for British Columbia and Eastern Canada (van Kooten

et al., 1999).

Five growth and yield scenarios for hybrid poplar

clones were represented: 20, 16, 14, 12 and 10 m3

ha�1 year�1 for 20-year rotations (scenarios 1–5,

respectively). Scenarios 4 and 5 have growth rates

close to growth expectations reported in Guy and

Benowicz (1998) and other Canadian studies (van

Kooten et al., in press). The other three scenarios

represent more optimistic growth rates of policy

interest.

2.4. Carbon redistribution among forest products

Carbon stored in harvested forest products is

released into the atmosphere through decay. The

model distinguishes three types of forest products

based on specific decay rates: lumber, pulp/paper

fast- and slow-decaying products. Fast-decaying pa-

per products usually have half-life periods of 1–5

years (Skog and Nicholson, 1998), lumber: 80–100

years (Winjum et al., 1998). The decay rate for fast-

decaying paper products was 0.5 (van Kooten et al.,

1999). Following van Kooten et al. (1999); we

assumed that 66% of paper products end up in

landfills (slow-decaying pool) with a decay rate

0.005. The remainder goes into the fast-decaying C

pool with a mean decay rate 0.5. A decay rate of

0.01 was assumed for lumber and other wood

products (Table 1). Carbon emissions from forest

products were considered as costs (van Kooten et al.,

1995) using the simulated carbon prices.

2.5. Substitution of fossil fuels

The model includes an assumption that some

harvested wood is used as fuel to offset the release

Table 1

Proportions of forest products and carbon decay rates for hybrid poplar

Parameters Forest products

Bioenergy use Lumber Paper products

(wood fuel), Fast-decaying Slow-decaying

log residuals pools pools

Decay rates 1 0.01 0.5 0.005

Forest product 0.2 0.2 0.6 0.15 0.45

ratios

D.W. McKenney et al. / Forest Policy and Economics 6 (2004) 345–358 349

of carbon by burning fossil fuels. Again following van

Kooten et al. (1999), we assumed that burning 1 m3 of

wood prevents the release of 0.190 ton of carbon from

burning coal.

The substitution of fossil fuels of course also

directly provides benefits from the sales of the resid-

ual material for bioenergy, i.e.:

PVBioenergy ¼pbe/bev tð Þ½ e�rT

1� e�rT; ð4Þ

where /be is a proportion of residual wood used for

bioenergy in the total harvest, pbe is the price paid for

residual wood, other symbols defined above.

In the present study, the expected mean price for

residual wood was assumed to be a triangular

distribution with $7.5 mode (Stevens et al., 2002).

We assumed a price variation of plus and minus

30%.

2.6. Modelling uncertainty

Uncertainty in the parameters and variables is

addressed using Monte–Carlo simulation techniques

(Law and Kelton, 2000). Most model variables such

as growth rates, biophysical constants, can be associ-

ated with the range of variation. Thus numeric vari-

ables were defined as statistical distributions rather

than fixed values with a probability-density function

type, expected range, mean and variance specified for

each model parameter. Algorithms described in Sauc-

ier (2000) were used to represent the three types of

distributions: triangular, constrained Gaussian and

Weibull (Table 2). Biological parameters were repre-

sented by a constrained Gaussian distribution. Most

economic parameters (i.e. agricultural land values,

prices and discount rates) were represented by the

triangular distribution with given minimum, maxi-

mum and mode (Stevens et al., 2002). The Weibull

probability-density function was used to simulate the

distribution of age-dependent parameters such as

stand senescence age and the period of initial carbon

accumulation in soil after beginning afforestation.

Thus, spatially explicit estimates of break-even carbon

price variation can be achieved for specific given

scenarios, but inclusive of a standard deviation and

range estimates. All results presented here are based

on simulations of 50 model runs for each particular

scenario.

2.7. Sensitivity analysis

To further assess the response of the break-even

carbon price to parameter uncertainty, sensitivity

analyses were performed. For each model parameter,

model simulations were repeated with the parameter

shifted plus and minus 20% from its original values

(but still inclusive of distribution limits). Finally, the

resulting distribution of break-even carbon price was

obtained for each parameter and plotted in one graph

for further comparison. This shows the relative sensi-

tivity of break-even carbon prices to the decrease or

increase of the particular parameter value.

2.8. Data

A spatial database of agricultural land was

generated from a 1-km AVHRR landcover classifi-

cation of Canada (Cihlar and Beaubien, 1998). The

following classes were used to delineate the area of

land potentially available for afforestation: grassland

(with shrub cover less than 10%), high, medium

and low biomass croplands, cropland-woodland

(mosaic land with prevalence of croplands, mixtures

Table 2

Basic uncertainty assumptions for the model parameters and variables

Definition Units Uncertaintya

Annual soil carbon accumulation rate in soil ton C ha�1 year�1 3

Annual tending/maintenance costs $ ha�1 year�1 1

Area of land rented annually ha Fixed value

(including land rented from Government)b

Average mature forest floor carbon mass tons ha�1 1

Carbon content of timber ton C m�3 3

Carbon price, social value of sequestering C $ ton�1 C 1, 3

Conversion factor from stem volume into carbon – 3

Conversion of above-ground biomass to carbon – 3

Conversion of above-ground biomass to – 3

non-merchantable timber volume

Discount rate % 1

Carbon offset from fossil fuel substitution tons ha�1 year�1 4c

Farm areab ha Fixed value

Forest products decay rates for fast-decaying, year�1 1

slow-decaying carbon pools and

Gross farm cash receiptsb $ year�1 1

Mean residence time of carbon in forest litter years 1

Period of soil carbon accumulation since the years 2

beginning of the afforestation

Plantation establishment costs $ ha�1 1

Proportion of forest products that goes into forest – 3

products decaying pool (fast- slow- decay and lumber)

Regression coefficients (rate of carbon accumulation in – 1

litter as function of age)

Soil carbon content in agricultural land ton C ha�1

Stand age at the rotation time years 1

Stand senescence age years 2

Standing timber price $ m�3 1

The amount of timber growing on a stand at the time t m3 ha�1 3

Timber price used for bioenergy $ m�3 1

Total farm business expensesb $ year�1 1

Total market value of the landb $ ha�1 1

Total rental/lease expenses for agricultural landb $ year�1 1

a Statistical distributions: 1—triangular; 2—constrained Weibull; 3—constrained Gaussian, see Saucier (2000).b Data from 2001 Census of Agriculture Consolidated Subdivisions (CCS), see StatCan (2001).c This output calculated from the mixture of randomized variables has no assumptions about the type of the distribution.

D.W. McKenney et al. / Forest Policy and Economics 6 (2004) 345–358350

with woodlands and other forest cover types). This

classification results in an estimate of agricultural

land area of 72 million ha including grasslands (67

million ha not including grasslands). This compares

to 67.5 million ha estimated in the Canadian

Census of Agriculture (StatCan, 2001). Grasslands

are included because hybrid poplar may be grown

in these areas. To outline a possible range of hybrid

poplar, the plant hardiness zone database was used

(McKenney et al., 2001). Zone 1a roughly outlines

a northerly limit for hybrid poplar. We assumed that

the appropriate clones of hybrid poplar would be

found to cover the existing variety of ecological

conditions within the study area.

As noted previously, agricultural land values and

their locations were taken from the 2001 Census of

Agriculture at the level of Census Consolidate Sub-

division units (StatCan, 2001).

The model also requires four other user-specified

spatial databases to run: rotation age, minimum,

modal and maximum expected plantation establish-

ment costs. These layers are ‘scenario-specific’. We

assumed a rotation age of 20 years for all areas

(F20%) and establishment costs with minimum

Table 3

Case study assumptions

Variables/Processes Value/Source/Comments

Tree species Hybrid poplar No assumptions about clonal variability

Discount rate Triangular distribution with mode=4%

Forest biomass Growth and yield curves Range of scenarios from 10 to 20 m3 per

ha year�1

Carbon accumulation rate Age-dependent

Above-below-ground biomass ratio Distribution (constrained Gaussian)

Harvested/residual post- Distribution (constrained Gaussian)

harvest biomass ratio

Carbon Carbon price Break-even

sequestration Time-dependent price fluctuations Not assumed

Carbon accumulation by soil Age-dependent for the first 50 years

Long-term carbon storage in forest Assumed (decay rate is specified as distribution)

products

Forest plantation Establishment costs $2000 ha�1F30% (triangular distribution)

Tending/treatment costs $5 ha�1 year�1F30% (triangular distribution)

Timber price $12 m�3F30% (triangular distribution)

Rotation age 20 years

Forest products Wood fuel (bioenergy) price $7.5 m�3F10% (triangular distribution)

Forest product rations and decay rates van Kooten et al., 1999

D.W. McKenney et al. / Forest Policy and Economics 6 (2004) 345–358 351

Cdn $1400, mode $2000 and maximum $2600 ha�1.

These ranges reflect reported estimates from the

available literature for hybrid poplar (see for exam-

ple, Stevens et al., 2002).

We also included annual plantation tending and

maintenance costs with a time-independent cost var-

iation of 30% (represented by a triangular distribu-

tion). Timber prices were represented by a triangular

distribution of plus/minus 30% around a mode of $12/

m3. Other assumptions and variables are summarized

in Table 3.

3. Results

3.1. Afforestation feasibility

Maps of break-even carbon prices are shown for

scenarios 2 and 4 (16, and 12 m3/ha year�1) in Fig. 1

for Western Canada and Fig. 2 for Eastern Canada.

The results clearly illustrate the effect of spatial

variation in agricultural land values. Break-even car-

bon prices are higher in Eastern Canada compared to

Western Canada (excepting parts of British Colum-

bia). At any given carbon price the Prairies show the

largest amount of the land potentially available for

afforestation. As might be expected, the northern and

central Prairies (incorporating the transition zone

between agriculture and boreal forest) appear to be

the most atractive. These areas have some of the

lowest agricultural land values in the country. Ontario

and Quebec generally have higher land values, and,

therefore increased break-even carbon prices. There

are, however, north–south and east–west gradients in

break-even prices in Eastern Canada. For example,

some parts of Northern Ontario where agriculture

occurs have break-even prices comparable to parts

of Western Canada.

Values between $10 and $50 ton�1 of CO2 have

been discussed as a possible price range for market-

able carbon credits. Fig. 3 provides a cumulative

portrayal of break-even carbon prices vs. agricultural

land available with the $15 and $50 break-even price

lines noted for all growth scenarios including the

range of values calculated from the Monte Carlo

analyses. Thus, for a given price, Fig. 3 indicates

the land area that might be available and converted to

plantation if landowners reacted purely to these price

Fig. 1. Mean break-even carbon prices, $ ton�1 CO2: scenarios 2 (16 m3 year�1) and 4 (12 m3 year�1), Western Canada.

D.W. McKenney et al. / Forest Policy and Economics 6 (2004) 345–358352

signals for their land-use activities. With growth rates

at 12 m3/ha year�1, locations with break-even carbon

prices less than $15 ton�1 CO2 are almost non-

existent in Eastern Canada.

However, with scenario 2 (16 m3/ha year�1),

more than 64% of agricultural land would potentially

be available for hybrid poplar afforestation in the

Western Canada at CO2 prices less than $15 ton�1.

Note the bound of variation in this scenario ranges

from f20% to greater than 90% land availability

(shaded areas in Fig. 3). In Eastern Canada results

indicate approximately 13% of agricultural land with

a range of 0–40%. Scenario 4 shows substantially

less land potentially available: less than 1% of

agricultural land in Western Canada and almost none

(0.07%) available for afforestation in the Eastern

Canada. Even the range of values generated by the

Monte Carlo analyses is small between $15 and $50

ton�1 of CO2 in this scenario. Not until prices are

greater than $75 ton�1 of CO2 do plantations appear

Fig. 2. Mean break-even carbon prices, $ ton�1 CO2: scenarios 2

(16 m3 year�1), and 4 (12 m3 year�1), Eastern Canada.

D.W. McKenney et al. / Forest Policy and Economics 6 (2004) 345–358 353

attractive relative to agriculture when growth rates

are 12 m3/ha year�1 or less in Eastern Canada. This

drastic change in potential land availability across

these scenarios is an indication of the importance of

valid growth and yield estimates for afforestation

policy assessments. The relationship between the

area and biomass expectations is non-linear and

shows an abrupt threshold between the scenarios of

14 and 16 m3/ha year�1.

For growth rates less than 14 m3/ha year�1 (sce-

narios 4 and 5), the results indicate somewhat mar-

ginally attractive conditions for afforestation at CO2

prices less that $50/ton, e.g. approximately 2.3% of

the total agricultural land in Canada for Scenario 4. 14

m3/ha year�1 appears to be an important threshold

affecting the economic attractiveness of fast growing

plantations. The overall results and apparent land

availability at $10, $15, $25 and $50 ton�1 of CO2

are summarized in Table 4.

Our estimates are more conservative than reported

in van Kooten et al., (1999). For example, at $20

ton�1 of carbon (i.e. $73.3 ton�1 CO2), our Scenario 4

indicates 12.7% of agricultural land available com-

pared to 2.3�106 ha (approx. 32%) in the van Kooten

et al. study (1999). The difference can probably be

explained by two factors: an assumption of higher

plantation establishment costs and higher land agri-

cultural values (opportunity costs). However, our

scenarios 1 and 2 (20 and 16 m3/ha year�1) give an

estimate of over 90% of the agricultural land in

Western Canada. This seems unrealistic if for no other

reason than the model does not include any changes in

prices that may arise if afforestation was adopted on

such a scale, putting pressure on regional labour and

other factor markets.

3.2. Relative importance of the model parameters

Spatially explicit sensitivity analyses were per-

formed on a region that includes the Ottawa to

Montreal corridor and the Eastern Ontario Model

Forest (approx. 40 000 km2) to provide more insights

as to the most important variables driving the results.

Large changes in break-even carbon prices and large

standard deviations identify the most important var-

iables. Figs. 4 and 5 summarize these results for

scenarios 2 and 4 (i.e. 16 and 12 m3/ha year�1) and

a 4% discount rate. Variables that had the most

impact on break-even carbon prices include the

discount rate, agricultural land values, merchantable

timber prices, aboveground yields, the conversion

factor between stand biomass and timber volume

and, the percent of timber used for forest products.

Belowground carbon accumulation is relatively less

important as plantation growth rates increase.

Importantly, these results indicate where more de-

tailed knowledge can help refine results. The impor-

tance of agricultural opportunity costs is notable. This

is a challenging model parameter partly because such

costs are arguably quite subjective andmay reflect non-

financial considerations (Buchanan, 1969). The Cana-

dian Forest Service is undertaking landowner surveys

Fig. 3. The percentage of agricultural land available for afforestation as function of break-even carbon price.

D.W. McKenney et al. / Forest Policy and Economics 6 (2004) 345–358354

to provide insights as to attitudes and inducements

required to convert agricultural land to plantations.

Some landowners may require plantation establish-

ment and some maintenance costs to be covered to

induce them to plant trees. Given the importance of

growth and yield assumptions on the break-even car-

bon price, spatially explicit syntheses of hybrid poplar

clone performance related to site conditions would be

valuable. Although, field studies may indicate growth

and yield expectations for some clones of above 12 m3

ha�1 year�1, disease outbreaks for mass plantings may

be a problem that could add to management costs and

lower yields (e.g. Challen et al., 2002).

There was little sensitivity to the forest product

decay rates, the redistribution of carbon among the

main types of forest products (e.g. lumber, pulp and

Table 4

Apparent land availability (millions of ha) for Western and Eastern Canada at $10, $15, $25 and $50 per ton CO2 for growth and yield scenarios

10, 12, 14, 16 and 20 m3 per ha * year�1

Scenarios

$10 per ton of CO2,

millions ha

$15 per ton of CO2,

millions ha

$25 per ton of CO2,

millions ha

$50 per ton of CO2,

millions ha

Western Eastern Total Western Eastern Total Western Eastern Total Western Eastern Total

10 m3/ha * yr.�1 0.15 <0.01 0.15 0.26 <0.01 0.26 0.31 0.02 0.33 0.48 0.05 0.53

12 m3/ha * yr.�1 0.29 <0.01 0.29 0.50 <0.01 0.51 0.72 0.07 0.79 1.03 0.17 1.20

14 m3/ha * yr.�1 1.23 0.08 1.31 2.34 0.13 2.47 6.67 0.87 7.54 41.67 2.55 44.22

16 m3/ha * yr.�1 26.62 0.92 27.54 32.87 1.42 34.29 45.08 6.93 52.01 49.41 11.98 61.39

20 m3/ha * yr.�1 39.74 1.91 41.65 41.79 2.36 44.15 48.17 8.34 56.51 50.74 13.19 63.93

D.W. McKenney et al. / Forest Policy and Economics 6 (2004) 345–358 355

paper), or the plantation establishment costs. The

latter can be explained by the relatively short rotation

age (20 years in present study). Large changes in

establishment costs would likely be more important

in longer rotation scenarios. As harvest yields in-

crease the importance of timber prices become more

critical to the break-even carbon price estimates.

Estimates of future values for merchantable timber

in Canada remains a challenge because prices are

generally not determined through market transac-

Fig. 4. Sensitivity of break-even carbon price to the changes in model param

change in percentage from mean break-even carbon price.

tions. $12/m3 is less than the value used by van

Kooten et al. (in press), however, is still more than

stumpage prices in parts of the country. The sensi-

tivity analysis results have implications for research

activities on carbon sequestration in Canada. For

example one implication is that research on below-

ground carbon sequestration processes may not be as

critical in helping determine the most cost-effective

afforestation programs for fast-growing tree species.

It is, however, possible to conceive of other situa-

eters of the scenario 2 (16 m3 year�1). Results are shown as relative

Fig. 5. Sensitivity of break-even carbon price to the changes in model parameters of the scenario 4 (12 m3 year�1). Results are shown as relative

change in percentage from mean break-even carbon price.

D.W. McKenney et al. / Forest Policy and Economics 6 (2004) 345–358356

tions where belowground carbon pools may be more

critical to break-even carbon prices (e.g. longer

rotation periods for higher value hardwood species

or slow-growing species). This type of analysis,

along with representations of other types of benefits

will be included in future studies.

4. Conclusions

A spatial model of the economics of the affor-

estation was presented for several growth and yield

scenarios of fast-growing hybrid poplar plantations

in Canada. The availability and growth potential of

fast growing clones is still a matter of research and

debate hence we attempted to represent a wide

range of possible growth rates. Relatively few

studies have developed spatially explicit models

(Bateman and Lovett, 2000; Turner et al., 2002).

Our model explicitly incorporates wood and bioen-

ergy values and plantation establishment and main-

tenance costs. Because carbon prices are unknown,

break-even carbon prices were used as the metric

for presentation. The results are akin to a marginal

cost curve that indicates potential land available

under different carbon price assumptions.

For all growth and yield scenarios the Prairies offer

the most promise in terms of cost-effectiveness but

results are sensitive to the assumed growth rate. For

growth rates less than 14m3/ha year�1, the results

currently indicate marginal feasibility for afforestation

at CO2 prices less that $50/ton. 14 m3/ha year�1

appears to be an important threshold affecting the

economic attractiveness of fast growing plantations.

Higher growth rates significantly increase the amount

of land potentially suitable for afforestation (at CO2

prices less than $50/ton). While the difference be-

tween 14 and 12 m3/ha year�1 may seem small, this is

a substantive change in biological productivity in

Canadian conditions.

The Monte Carlo approach and sensitivity analy-

ses provide important insights on further research

D.W. McKenney et al. / Forest Policy and Economics 6 (2004) 345–358 357

needs and relative priorities. However, note the

priority of refining particular parameter estimates is

sometimes dependent on the nature of the scenario.

When the rate of carbon accumulation by above-

ground biomass is high, and the period of forest

rotation is short, accurate estimates of above ground

yields become one of the most important factors in

determining areas feasible for carbon sequestration.

Growth and yield estimates that include spatial

variability due to site conditions and climate would

improve results. Further refinements in spatial esti-

mates of economic parameters such as timber and

bioenergy prices and plantation establishment and

maintenance costs may also improve accuracy. Fu-

ture model runs will include these enhancements and

representations of selected non-wood, non-carbon

values.

References

Bateman, I.J., Lovett, A.A., 2000. Estimating and valuing the car-

bon sequestered in softwood and hardwood trees, timber prod-

ucts and forest soils in Wales. Journal of Environmental

Management 60, 301–323.

Bowes, M.D., Krutilla, J.V., 1989. Multiple-Use Management: the

Economics of Public Forestlands. Resources for the Future,

Washington, DC, p. 357.

Buchanan, J.M., 1969. Cost and Choice: an Inquiry in Economic

Theory. The University of Chicago Press, Chicago, p. 104.

Cannell, M.G.R., 1999. Growing trees to sequester carbon in the

UK: answers to some questions. Forestry 72, 237–247.

Challen, L., Watt., K., Prezio, J., 2002. Assessment of hybrid poplar

plantations for fibre supply in Northwestern Ontario, 3rd year

results. LUSTR Co-op Res. Rep. LUSTR Co-op, Thunder Bay,

ON. p. 44.

Cihlar, J., Beaubien, J., 1998. Land cover of Canada Version 1.1.

Special Publication. NBIOME Project. Produced by the Canada

Centre for Remote Sensing and the Canadian Forest Service,

Natural Resource s Canada. Available on CD ROM from the

Canada Centre for Remote Sensing, Ottawa, ON.

Creedy, J., Worzbacher, A.D., 2001. The economic value of a for-

ested catchment with timber, water and carbon sequestration

benefits. Ecological Economics 38, 71–83.

de Jong, B.H.J., Tipper, R., Montoya-Gomez, G., 2000. An eco-

nomic analysis of the potential for carbon sequestration by for-

ests: evidence from Southern Mexico. Ecological Economics

33, 313–327.

Environment Canada, 2001. 1990–2000 National and Provincial

GHG Emissions (Official website). 1990–2000 Greenhouse

Gas Emission Estimates for Canada. Available on Internet on

http://www.ec.gc.ca/pdb/ghg/documents/tables/Canada2000.pdf.

Essex, C., McKitrick, R., 2002. Taken by Storm: the troubled sci-

ence, policy and politics of global warming. Key Porter Books,

Toronto, ON.

Guy, R.D., Benowicz, A., 1998. Can afforestation contribute to a

reduction in Canada’s net CO2 emissions? Report prepared for

Canadian Pulp and Paper Association. Mimeograph, March.

Department of Forest Sciences, University of British Columbia,

Vancouver, BC.

Hartman, R., 1976. The harvesting decision when a standing forest

has value. Economic Inquiry 14, 52–58.

IPCC (Intergovernmental Panel on Climate Change), 2001. Climate

Change 2001: Impacts, Adaptation and Vulnerability. Cam-

bridge University Press, Cambridge, UK, p. 1033.

Jacques, A., 1998. Revised 1990 and 1996 greenhouse gas emis-

sions estimates. Environment Canada, Pollution Data Branch,

ON, Ottawa.

Krcmar, E., Stennes, B., van Kooten, G.C., Vertinsky, I., 2001.

Carbon sequestration and land management under uncertainty.

European Journal of Operational Research 135, 616–629.

Kurz, W.A., Apps, M.J., Webb, T.M., McNamee, P.J., 1992. The

carbon budget of the Canadian forest sector: phase 1. For. Can.,

NW Region, Northern Forest Centre. Edmonton, AB. Inf. Rep.

NOR-X-326. p. 93.

Kurz, W.A., Apps, M.J., 1999. A 70-year retrospective analysis of

carbon fluxes in the Canadian forest sector. Ecological Appli-

cations 9, 526–547.

Law, A.M., Kelton, W.D., 2000. . Simulation Modelling and Anal-

ysis, 3rd ed. McGraw-Hill, Burr Ridge, IL, p. 784.

McKenney, D.W., Kesteven, J.L., Hutchinson, M.F., Venier, L.A.,

2001. Canada’s Plant Hardiness zones revisited using modern

climate interpolation techniques. Canadian Journal of Plant Sci-

ence 81, 117–129.

Paoli, G., Bass, B., 1997. Editorial: climate change and variability,

uncertainty and decision making. Journal of Environmental

Management 49, 1–6.

Pearce, D.W., 1994. The environment: assessing the social rate of

return from investment in temperate zone forestry. In: Layard,

R., Glaister, S. (Eds.), Cost-Benefit Analysis. Cambridge Uni-

versity Press, Cambridge, UK, pp. 464–490.

Peng, C., Liu, J., Dang, Q., Apps, M.J., Jiang, H., 2002. TRI-

PLEX: a generic hybrid model for predicting forest growth

and carbon and nitrogen dynamics. Ecological Modelling 153,

109–130.

Richards, K.R., 1992. Policy and research implications of recent

carbon-sequestering analysis. In: Reilly, J.M., Anderson, M.

(Eds.), Economic Issues in Global Climate Change. Westview

Press, Boulder, CO, pp. 288–308.

Saucier, R., 2000. Computer generation of statistical distributions.

US Army Research Laboratory. Adelphi, MD. Rep. ARL-TR-

2168. p. 106.

Sedjo, R.A., Wisniewski, J., Sample, A.V., Kinsman, J.D., 1995.

The economics of managing carbon via forestry: assessment of

existing studies. Environmental and Resource Economics 6,

139–165.

Skog, K.E., Nicholson, G.H., 1998. Carbon cycling through wood

product: the role of wood and paper products in carbon seques-

tration. Forest Products Journal 4, 75–83.

Smith, J.E., Heath, L.S., 2002. A model of forest floor carbon mass

D.W. McKenney et al. / Forest Policy and Economics 6 (2004) 345–358358

for United States forest types. USDA Forest Service, NE Res.

Stn., Newtown Square, PA. Res. Pap NE-722. p. 37

StatCan (Statistics Canada), 2001. 2001 Census of Agriculture. On-

line database. Accessed Jan 2003. http://www.statcan.ca/

english/freepub/95F0301XIE/.

Stevens, M.L., McKenney, D.W., Campbell, K., 2002. Afforesta-

tion potential in Canada: a spatial analysis of economic land

suitability with carbon sequestration benefits. In: Shaw, C.H.,

Apps, M.J. (Eds.), Proceedings of the International Science

Confernce ‘The Role of Boreal Forests and Forestry in the

Global Carbon Budget’, May 8–12, 2000. Edmonton, AB. Ca-

nadian Forest Services, Edmonton, AB, pp. 201–215.

Turner, B.J., Chikumbo, O., Davey, S.M., 2002. Optimization

modelling of sustainable forest management at the regional

level: an Australian example. Ecological Modelling 153,

157–179.

van Kooten, G.C., Arthur, L.M., Wilson, W.R., 1992. Potential to

sequester carbon in Canadian forests: some economic consider-

ations. Canadian Public Policy 18, 127–138.

van Kooten, G.C., Binkley, C.S., Delcourt, G., 1995. Effect of

carbon taxes and subsidies on optimal forest rotation age and

supply of carbon services. American Journal of Agricultural

Economics 77, 365–374.

van Kooten, G.C., Krcmar-Nozic, E., Stennes, B., van Gorkom, R.,

1999. Economics of fossil fuel substitution and wood product

sinks when trees are planted to sequester carbon on agricultural

lands in Western Canada. Canadian Journal of Forest Research

29, 1669–1678.

van Kooten, G.C., Shaikh, S.L., Suchanek, P. in press. Mitigating

climate change by planting trees: the transaction costs trap. Land

Economics.

Winjum, J.K., Brown, S., Schlamadinger, B., 1998. Forest harvests

and wood products: sources and sinks of atmospheric carbon

dioxide. Forest Science 44, 272–284.