Copyright copy 2012 by the author(s) Published here under license by the Resilience AlliancePaterson S and B A Bryan 2012 Food-carbon trade-offs between agriculture and reforestation land usesunder alternate market-based policies Ecology and Society 17(3) 21httpdxdoiorg105751ES-04959-170321

Research part of a Special Feature on Applying Landscape Science to Natural Resource Management

Food-Carbon Trade-offs between Agriculture and Reforestation LandUses under Alternate Market-based PoliciesStacey Paterson 12 and Brett Anthony Bryan 1

ABSTRACT Understanding the effects of payments on the adoption of reforestation in agricultural areas and the associatedfood-carbon trade-offs is necessary to inform climate change policy Economic viability of reforestation under payment perhectare and payment per tonne schemes for carbon sequestration was assessed in a region in southern Australia supporting 61Mha of rain-fed agriculture The results show that under the median scenario a carbon price of 27 A$tCO2-e could make one-third of the study area (nearly 2 Mha) more profitable for reforestation than agriculture and at 58 A$tCO2-e all of the studyarea could become more profitable The results were sensitive to variation in carbon risk factor establishment costs and discountrates Pareto-optimal land allocation could realize one-third of the potential carbon sequestration from reforestation (1635MtCO2-eyr at a carbon risk factor of 08) with a loss of less than one-tenth (10789 A$Myr) of the agricultural productionBoth payment schemes resulted in efficiencies within 1 of the Pareto-optimum Understanding food-carbon trade-offs andpolicy efficiencies can inform carbon policy design

Key Words agriculture agroecosystem carbon sequestration ecosystem services food security land use payment policyreforestation

INTRODUCTIONLand clearance and agricultural production have increasedemissions of climate-changing atmospheric greenhouse gasesfrom agroecosystems (Foley et al 2005) Howeveragricultural landscapes can also have an important role inabating greenhouse gas emissions and mitigating climatechange (Palm et al 2010) Reforestation of agricultural landcan mitigate climate change through sequestration of carbonin biomass and soils or through the production of bioenergy(Obersteiner et al 2010) The trade-off from reforestation isthe reduced availability of land for production of food andfiber (Smith et al 2010) Direct competition for finite landresources creates a tight link between carbon sequestrationand agricultural commodities (Golub et al 2009 Smith et al2010) Hence policies designed to mitigate climate change bypromoting reforestation can put upward pressure upon foodprices (Wise et al 2009) with subsequent implications forglobal human nutrition To avoid these adverse outcomesclimate policy needs to consider potential food-carbon trade-offs between agricultural and reforestation land uses in rurallandscapes

The potential for carbon and ecosystem service market policiesto alter the relative profitability of land uses and motivate largescale reforestation has been recognized (Alig et al 2010Crossman et al 2011) The impact of market-based climatepolicy on reforestation of agricultural land depends on manyfactors with expected profitability critical among these (Irwinand Geohegan 2001 Lubowski et al 2008) Severalassessments of the impact of market policies on reforestationand carbon sequestration have been made (Plantinga et al1999 Stavins 1999 Lubowski et al 2006 Harper et al 2007

Hunt 2008 Lawson et al 2008 Polglase et al 2008 2011)As opportunity costs of alternative land uses have often beenconsidered these studies have implicitly assessed trade-offsbetween land uses Although much attention has been paid tofood production impacts from bioenergy policy (Bryan et al2010b Tilman et al 2009) there have been few explicitassessments of the potential impact on food production ofpolicies encouraging reforestation for carbon sequestration

Food-carbon trade-offs resulting from reforestation ofagricultural land will depend upon the spatially varying abilityof land areas to sequester carbon relative to their agriculturalproductivity (Crossman et al 2011) The design of market-based policy specifically whether payments are heterogeneous(spatially-targeted payment per tonne of carbon sequestered)or homogeneous (payment per unit area reforested) can affectpolicy efficiency and the nature of the food-carbon trade-offsrealized by carbon payments It is well established that spatialtargeting based on both costs and benefits results in moreefficient outcomes than targeting based upon costs or benefitsalone when considering single contracts and the extent of thisefficiency gain is dependent upon the relative variance of costsand benefits (Babcock et al 1997 Wu and Bogess 1999Ferraro 2003 Newburn et al 2005 Waumltzold and Drechsler2005 Crossman and Bryan 2009 Chen et al 2010 Stoms etal 2011)

West et al (2010) highlighted the influence of spatialarrangement of land use on efficiency gains for managingfood-carbon trade-offs at a global scale The potentialinfluence of land use allocation on efficiency at finer spatialscales eg landscape scale may also be substantial (DeFrieset al 2004) because spatial heterogeneity has been found at

1CSIRO Ecosystem Sciences 2University of Adelaide

Ecology and Society 17(3) 21httpwwwecologyandsocietyorgvol17iss3art21

this scale in carbon sequestration (Cantarello et al 2011)agricultural production (Bryan et al 2009) and land useprofitability (Bryan et al 2011a) Improving the scale andresolution of biophysical and economic modeling includingthe accurate representation of spatial heterogeneity can betterestimate the food-carbon trade-offs likely to result fromchanges in economic viability of reforestation caused bymarket-based climate policy (Van der Werf and Peterson2009) This can inform the design of more efficient policyinstruments which are better able to minimize adverse impactsof reforestation on food production

Two main techniques have been used to quantify the trade-offs between competing land uses at a landscape scaleScenario analysis has been used to quantify spatially-explicittrade-offs between food production and greenhouse gasmitigation through bioenergy land uses (Bryan et al 2010aThomson et al 2010) but only for a limited number of spatialand policy options Alternatively Polasky et al (2008)searched for efficient land use combinations to maximizespecies conservation for a range of levels of economic returnsand vice versa resulting in a Pareto-optimal frontier ofspatially explicit land use arrangements Higgins et al (2008)and Nelson et al (2008 2009) employed similar methods tohighlight the trade-offs between carbon sequestration andother ecosystem services across a spectrum of policy optionseg a range of target levels under alternative targetingstrategies However no studies have explicitly assessed thefood-carbon trade-offs of reforestation under alternativemarket-based policies on a landscape scale for informingefficient policy design

In this study we quantified food-carbon trade-off curves at alandscape scale under two commonly used market-basedcarbon policy instruments payment per tonne and paymentper hectare Using modeled spatially explicit (2km x 2km landunits) estimates of agricultural production and carbonsequestration potential we calculated layers of profit fromagriculture and from reforestation under each payment schemefor carbon prices ranging from 1 to 200 $tCO2-e (tonnes ofcarbon dioxide equivalent all dollar figures in Australiandollars) and identified the most profitable land use Wecalculated Pareto-optimal production frontiers that quantifychanges in carbon sequestration and food production forproduction-maximizing land allocation for all possible levelsof carbon sequestration We compared the efficiency of food-carbon trade-off curves against the production frontiers Wetested the sensitivity of economic viability of reforestationpolicy efficiency and food-carbon trade-offs to variation incarbon risk discounting establishment costs transaction andmaintenance costs and economic discount rates We discussthe implications of the results for carbon policy design

METHODS

Study area descriptionThe 119 million ha Lower Murray case study area includesthe South Australian Murray-Darling Basin Natural ResourcesManagement Board region (SAMDB) and the Mallee andWimmera Catchment Management Authority areas (Fig 1)Climate ranges from semiarid in the north of the study areaexperiencing just 300mmyear rainfall through Mediterraneanto cool temperate in the south (up to 900mmyr rainfall) Soilsare typically nutrient-deficient The two dominant land usesare rain-fed agriculture and nature conservation with small(16) but economically important areas of irrigatedagriculture along the River Murray Rain-fed agriculturalproduction ranges from continuous sheep grazing for meat andwool in the drier northern and central SAMDB and northernMallee regions to continuous cereal cropping in the wetterWimmera region with cropping-grazing rotations employedin between Wheat yields range from 068 tha to 218 tha(average 076 tha) Net economic returns from agriculturerange from 6 to 479 $hayr (average 179 $hayr Bryan et al2011a) We confined our assessment of food-carbon trade-offs to the 61 million ha of land currently used for rain-fedagriculture (Fig 1) We tessellated this area into 2km x 2kmgrid cells for analysis thereby creating 15241 individual landunits each 400 ha in area

Fig 1 Location and land use in the Lower Murray studyarea

Ecology and Society 17(3) 21httpwwwecologyandsocietyorgvol17iss3art21

Table 1 Summary of model parameters assessed in this study (- used in low cost scenario dagger used in median scenario + used inhigh cost scenario)

Parameter Units Symbol ValuesCarbon price $tCO2-e p 1 2 200Carbon risk discounting factor Factor R 10 08dagger 06Establishment costs $ha ec 1150- 2000dagger 6000+

Transaction and maintenance costs $hayr tmc 40- 70dagger 100+

Economic discount rates r 3 7dagger 11 15Total number of years in the analysis Years T 64

Agricultural productionWe used net economic returns to represent the value ofagricultural production The advantage is that it provides asingle measure of the production of different agriculturalproducts ie wheat and other crops sheep meat and woolthat reflects the value society places upon them

Spatial estimates of net economic returns to agriculture werecompiled in two stages First the frequency of rotation ofdifferent farming system phases ie wheat lupins and sheepwere derived from agricultural census data and catchmentscale land use mapping and used to characterize the spatialdistribution of farming system rotations Second agriculturalprofit was calculated using a profit function The profitfunction includes information on yield price and costs ofproduction by Statistical Local Area (SLA) derived fromagricultural census and state government Gross MarginHandbooks The SLA-based estimates were then smoothedusing pycnophylactic (mass-preserving) interpolation andcombined with the spatial distribution of farming rotationsystems to create a layer of annual expected profit fromfarming systems (Bryan et al 2009 2011a) This wasconverted to Net Present Value (NPV) over the analysis periodof 64 years (2006-2070) The period to 2070 was selected toaccount for the long time period required for carbon to besequestered

Spatial prediction of forest growth and carbonsequestrationThe Physiological Principles to Predict Growth (3-PG) model(Landsberg and Waring 1997) was used to estimatereforestation rates over the study area 3-PG takes soil andclimatic data inputs and knowledge of the physiology of treespecies to predict stand biomass water use and available soilwater on a monthly basis To model tree growth and carbonsequestration in 3-PG we used a species growth parameter setfor a mallee tree species (Eucalyptus kochii CSIROunpublished data) well adapted to the climate in our regionbut not native to the study area

We assembled spatial information on broad soil class fromstate government databases We used representative soil pitdata to derive maximum available soil water (ASW) for each

soil class We also specified fertility rating = 08 minimumASW = 0 mm initial ASW = 60 mm seedling mass = 5 gplanting density of 1000 stemsha We created spatial layersof mean monthly rainfall temperature (maximum andminimum) and solar radiation using the ESOCLIM climatemodel We used 3-PG spatial to model the spatial distributionof stand growth over a time period of 64 years Forestproductivity was modeled at a spatial resolution of 200m x200m grid cells then converted to the 2km x 2km resolutionusing bilinear resampling

Carbon sequestration is dependent upon the growth of biomassthat varies over time We used the von Bertalanffy-Chapman-Richards (vBCR) growth function (Richards 1959 Zhao-gangand Feng-ri 2003) to capture nonlinear (sigmoidal) rates ofbiomass growth and carbon sequestration over time asmodeled by 3-PG The growth curve has been commonly usedto predict growth patterns in forestry (Alexandrov 2008) Wefit a vBCR growth function to the 3-PG modeled growth usinga genetic algorithm Stand carbon sequestration was derivedfrom stand biomass in carbon dioxide equivalent terms (tonnesCO2-eha) and the annual carbon increment Yt (tCO2-ehayr)was calculated

Economic scenarios and sensitivity analysisWe assessed food-carbon trade-offs associated with policyalternatives using a range of carbon prices under median valuesfor carbon sequestration risk discounting establishment coststransaction and maintenance costs and economic discountrates (Table 1) We also tested the sensitivity of the results tovariation in these parameters

Carbon sequestration risk discountingCacho et al (2005) highlighted the importance of discountingcarbon biosequestration compared to other more permanentmethods of reducing atmospheric greenhouse gasconcentrations When carbon is sequestered in organicmaterial there is some risk that it will be released back intothe atmosphere through a range of processes eg fireharvesting land clearance In contrast avoided emissions areeffectively avoided forever In addition a number of otherfactors also contribute to uncertainty in carbon sequestrationincluding modeling error and moral hazard We accounted for

Ecology and Society 17(3) 21httpwwwecologyandsocietyorgvol17iss3art21

this suite of risk factors by discounting the 3-PG-modeledcarbon sequestration rates We used a median carbon riskdiscount factor of 08 such that for each tonne of CO2-esequestered (as modeled by 3-PG) only 08 tCO2-e could besold in a carbon market We also tested the sensitivity of themodel at risk discount factors of 06 and 10 (Table 1)

Carbon pricesThe range of carbon prices in this study (Table 1) was designedto enable analysis of the full trade-off curve ranging from noeconomically viable areas for reforestation to all areasbecoming viable Increasing carbon price ($tCO2-e) in $1increments captured the shape of the supply curve and theproduction frontier

Establishment costsEstimates of establishment costs of reforestation (Table 1)using two different methods ie direct seeding and tube stockplanting were obtained from Greening Australia (JMcGregor personal communication January 17 2011) Theestimates included maintenance costs for two yearspostplanting and are conservative being based upon costs ina higher rainfall (600+ mmyr) area with good soil At a costof 1150 $ha direct seeding was significantly cheaper thanplanting seedling tube stock at 6000 $ha However plantingwith tube stock increases survivorship and control of thecanopy layers We also used a median estimate ofestablishment costs of 2000 $ha estimated from the SouthAustralian Governmentrsquos large scale River Murray Forestrestoration program (Crossman et al 2011)

Ongoing transaction and maintenance costsAnnual transaction and maintenance costs also accrue for themanagement of reforested areas (Table 1) Transaction costsinclude a range of administrative activities associated withcarbon sequestration trading and accounting Maintenancecosts include feral pest and disease management fire riskmanagement and a range of other activities required tomaintain reforested stands

Economic discount ratesWe assessed five discount rates (Table 1) The rates r = 3and 11 encompass the minimum and maximum cash rate setby the Reserve Bank of Australia since September 1991 (RBA2010) Variation in discount rate within this range is likelyThe rate r = 15 was also included as a more extreme caseAlthough higher than commercial interest rates this interestrate reflects the case in which landholders have a strongpreference for income in the present over income in the future

Policy alternativesWe calculated food-carbon trade-offs under two alternativepayment schemes payment per tonne and payment perhectare The trade-off models were implemented in theGeneral Algebraic Modeling System (GAMS) We solved themodels for each combination of carbon risk factors carbon

prices discount rates transaction and maintenance costs andestablishment costs (Table 1) In both policy scenariospayments were made annually using the lsquoideal paymentsystemrsquo of Cacho et al (2003) which accounts for the variationin rates of tree growthcarbon sequestration over time Wehave not considered the costs of carbon monitoring andverification in either policy alternative The magnitude ofthese costs is sensitive to policy design (see Discussion)

Payment per tonneThe payment per tonne scheme incorporates spatialheterogeneity in carbon sequestration consistent with theKyoto Protocol and the Clean Development Mechanism Oneof the major drawbacks of this scheme is the transaction costsassociated with carbon monitoring verification andaccreditation Typically carbon sequestration monitoring isundertaken through a combination of remote sensingmodeling and on-ground measurement As an example of thepotential impact of these costs most participants in lsquoActivitiesImplemented Jointlyrsquo of the United Nations FrameworkConvention on Climate Change were not willing to monitorcarbon sequestration because of the high costs (Cacho et al2005) Here we assume landholders are paid annually for theamount of carbon they sequester (Yt) as modeled using the 3-PG spatial estimates and the vBCR growth curve describedabove Under the payment per tonne scheme the economicreturn to landholders was calculated in NPV terms as

(1)

p is the price per tonne of carbon sequestration and pYtR represents the cash inflow from reforestation in each period tgiven a carbon risk discounting factor R tmc represents theannual transaction and maintenance costs r is the discountrate and T denotes the total number of years (64) in theanalysis

Payment per hectareUnder the payment per hectare scheme payment for carbonsequestration was paid at uniform production levels equal tothe regional average sequestration rate per hectare (1006tCO2-eyr) This is similar to other schemes such as theCanadian Permanent Cover Program This style of paymenthas potential to reduce transaction costs and enhance paymentequity between landholders (Wu and Boggess 1999) To arriveat the regional average carbon sequestration rate (Y) wecalculated the average annual rate of carbon sequestration overtime based on the 3-PG model estimates over the 64-year timehorizon then averaged this across all rain-fed agriculture gridcells Under this scheme economic returns to landholderswere calculated as

Ecology and Society 17(3) 21httpwwwecologyandsocietyorgvol17iss3art21

(2)

Pareto-optimal benchmarkTo determine the efficiency of the two payment schemes andquantify the food-carbon trade-offs we calculated Pareto-optimal production frontiers The Pareto frontiers showpotential production bundles resulting from all possibleefficient combinations of land use An efficient land useallocation is one in which no units of land could be reallocatedto increase production of one good without decreasingproduction of the other The following steps were used tocalculate each Pareto frontier

1 For each land unit calculate the ratio of potentialagricultural production measured in economic profit ($ha) to potential carbon sequestration (tCO2-eha) underreforestation

2 Rank land units in increasing order of food-carbon ratio

3 In rank order calculate and graph the total agriculturalproduction ($) and carbon sequestration (tCO2-eha) inthe study area as each land unit is sequentially convertedto reforestation

Economic viability of reforestation food-carbon trade-offs and efficiencyWe quantified and mapped the net economic returns toreforestation ie returns to reforestation minus returns toagricultural production under the full range of carbon pricesThe economically viable area of reforestation ie wherereturns to reforestation exceed those from agriculture andcarbon sequestration supply were graphed against carbonprice Trade-off curves track the aggregate change in foodproduction and carbon sequestration from economicallyviable areas with carbon price To quantify policy efficiencythe area under the trade-off curves for both the payment pertonne and payment per hectare schemes was calculated andcompared against the area under the Pareto-optimalproduction frontier Sensitivity analysis was then conductedto assess the influence of model parameters on (a)economically viable area and (b) the efficiency of paymentschemes and the nature of food-carbon trade-offs

RESULTS

Carbon sequestration and economically viable areasModeled carbon sequestration rates ranged from annualaverages of 731 thayr to 1648 thayr (mean 1006 thayrFig 2) Payment per hectare and payment per tonne systemsproduced similar economically viable areas For each paymentsystem economically viable areas and carbon supply curveshave four inflection points that occur at similar prices andlevels of sequestration under the median scenario (Fig 3)

These points mark shifts in the economic viability ofreforestation in response to changes in price The greatest shiftin economic viability occurs as the price for carbon increasesfrom 20 $tCO2-e to 27 $tCO2-e Nearly one-third (3263)of the land in the study areas becomes more profitable underreforestation than under agriculture at 27 $tCO2-e A furtherincrease in carbon price to 32 $tCO2-e has much less of animpact on the relative profitability of reforestation andagriculture For prices greater than 32 $tCO2-e the rate ofchange in economic viability of reforestation is higher andremains constant up to around 52 $tCO2-e where nearly allof the study area becomes more profitable for reforestationthan for agriculture The supply curves for carbonsequestration follow similar patterns to the economicallyviable area curves (Fig 3)

Fig 2 Total carbon sequestration in the study area over 64years as modeled by 3-PG

The spatial distribution of net economic returns toreforestation areas under the two payment schemes ispresented for the median scenario in Figure 4 Although thegeneral pattern in net economic returns is similar differencesoccur at all carbon prices Differences arise because under thepayment per hectare scheme the least profitable agriculturalland becomes economically viable first Under a payment pertonne scheme the first areas that become economically viablehave both low returns to agriculture and high carbonsequestration capacity

Ecology and Society 17(3) 21httpwwwecologyandsocietyorgvol17iss3art21

Fig 3 Economically viable area for reforestation andcarbon sequestration supply curve under a range of carbonprices for the median scenario (R = 08 ec = A$2000hatmc = A$70hayr r = 7) Carbon sequestration is anannual average value over 64 years

Policy efficiency and food-carbon trade-offsFood-carbon trade-off curves were convex in shape As thecarbon price increases and more carbon is sequestered inreforested land the marginal cost in terms of lost agriculturalproduction increases Relatively larger amounts of carbon canbe sequestered for minimal impact on food production Forthe Pareto-optimum production frontier one-third of totalaverage annual carbon sequestration potential fromreforestation (1635 MtCO2-eyr R = 08) can be achievedwith a loss of 983 of annual agricultural profits (10789 $Myr) The food-carbon trade-off curves of both paymentschemes are very similar and both are extremely close to thePareto-optimum suggesting high efficiency in land useallocation under both payment systems The efficiency of thepayment per tonne and payment per hectare schemes underthe median scenario compared to the Pareto-optimal was9903 and 9956 respectively (Fig 5)

Sensitivity analysisVariation in establishment costs transaction and maintenancecosts and discount rates affects the profitability ofreforestation compared with agriculture (Fig 6) Increasedcosts significantly increase the carbon price required to makereforestation profitable and higher discount rates exaggeratethis effect Of these costs establishment costs have a muchstronger influence than transaction and maintenance costs onthe carbon price required to make reforestation a profitableoption for any land unit High discount rates (11 15)decrease the relative profitability of reforestation because ofthe lag time associated with tree growth and discounted returnsfrom carbon sequestration (Fig 6)

Increasing the establishment costs and discount rate decreasedthe efficiency of reforestation of economically viable areasbut only under the payment per tonne scheme (Fig 7) Thegreatest deviation from the Pareto-optimal occurs when highestablishment costs (6000 $ha) high transaction andmaintenance costs (100 $hayear) and high discount rates(15) were considered simultaneously This combination ofparameters changes the spatial distribution of economicallyviable areas and results in less efficient land allocation andassociated production outcomes The high cost and highdiscount rate scenario achieved 9279 efficiency relative tothe Pareto-optimal production frontier The low cost (ec =1150 $ha tmc = 40 $ha) and low discount rate (3) scenarioachieved 9982 efficiency The efficiency of the per hectarepayment system was unaffected by changes in these economicparameters (Fig 7)

DISCUSSION

Economic viability of reforestation and food-carbontrade-offsThe calculation of economic returns shows that large areas ofagricultural land could become more profitable as carbon sinksat relatively modest carbon prices For example under themedian scenario a carbon price of 27 $tCO2-e makes nearlyone-third of the study area (199 Mha) more profitable forcarbon sequestration At 58 $tCO2-e all of the study area ismore profitable However these results are very sensitive tothe choice of model parameters with the carbon risk factorestablishment costs and discount rates all having significanteffects on the economically viable area for reforestation andthe resultant carbon supply curves Although the results arestriking they should not be taken as a prediction ofreforestation extent under a carbon market Rather theseinsights are most effectively used in informing policy designand in providing direction for future research

Ecology and Society 17(3) 21httpwwwecologyandsocietyorgvol17iss3art21

Fig 4 Spatial distribution of net economic returns (annualized) to reforestation under the payment per tonne and paymentper hectare schemes for a range of carbon prices using the median scenario parameter values (R = 08 ec = A$2000ha tmc =A$70hayr r = 7)

Fig 5 Food-carbon trade-offs for the payment per tonneand payment per hectare schemes Trade-offs werecalculated using the median scenario parameter values (R =08 ec = A$2000ha tmc = A$70hayr r = 7) and valuesof carbon sequestration are an annual average over a 64-yearperiod

The land use system analyzed in this paper is land-limited suchthat increased production of one good (carbon) reduces

production of the other (food) Land units show spatialheterogeneity in the productivity of agriculture and carbonsequestration Convex trade-off curves presented anopportunity through efficient land allocation to sequesterroughly one-third of the total carbon at a cost of less than one-tenth of the total agricultural production The converse risk isthat maximally inefficient land allocation could result in theother two-thirds of the total carbon coming at a cost of nine-tenths of the agricultural production The opportunity to usereforestation to sequester significant amounts of carbon withminimal impacts on agricultural output can be realized andthe risk of reforestation significantly reducing agriculturalproduction can be avoided through efficient policy design

Policy efficiencyA priori we expected the payment per tonne scheme to be themore efficient payment scheme because it links bothheterogeneous costs (foregone agricultural profits) andheterogeneous benefits (tonnes of carbon sequestered) to amonetary incentive for the landholder Conversely thepayment per hectare scheme links heterogeneous costs and ahomogeneous proxy for benefits (regional mean carbonsequestration) to the landholderrsquos monetary incentives Theuse of a homogeneous proxy to determine payment rates fora heterogeneous benefit introduces the potential for inefficientland use allocation However under median parameters bothpayment schemes led to production outcomes with efficiencies

Ecology and Society 17(3) 21httpwwwecologyandsocietyorgvol17iss3art21

Fig 6 Sensitivity of economically viable area for reforestation to variation in carbon risk discounting factor economicdiscount rate establishment costs and transaction and maintenance costs under the payment per tonne scheme (High cost ec= A$6000yr tmc = A$100yr Median cost ec = A$2000yr tmc = A$70yr Low cost ec = A$1150yr tmc = A$40yr)

Ecology and Society 17(3) 21httpwwwecologyandsocietyorgvol17iss3art21

very near Pareto-optimal (section 32) This is consistent withtheory (Babcock et al 1997 Wu and Boggess 1999) as thebenefits have a very low variance (COV = 14) comparedwith costs (COV = 64) Hence replacing the heterogeneousbenefit with the homogeneous proxy does not significantlyaffect the economic returns to reforestation

Fig 7 Sensitivity of food-carbon trade-offs and efficiencyof the payment schemes to changes in economic parameters(High cost scenario ec = A$6000ha tmc = A$100hayr r= 15 Median cost scenario ec = A$2000ha tmc = A$70hayr r = 007 Low cost scenario ec = A$1150ha tmc =A$40hayr r = 003)

Increasing costs and discount rates decreased the efficiencyof the payment per tonne scheme but did not affect theefficiency of the payment per hectare scheme Under thepayment per hectare scheme profits from carbon sequestrationare homogeneous so absolute advantage in agriculturalproduction determines the relative profitability of carbonsequestration for each land unit Increasing establishmentcosts and discount rates reduced the attractiveness of carbonsequestration and shifted its supply but did so in the sameway for each land unit Agricultural profits were not affectedHence the relative profitability of carbon sequestrationbetween land units was not affected by changes in theseparameters and neither was efficiency Conversely under thepayment per tonne scheme increasing establishment costs and

discount rates affected the profitability of carbon sequestrationdifferently for each land unit High establishment costs anddiscount rates can distort the relative profitability ofreforestation leading to allocative inefficiency in land useThis is reflected by the inward shift in the trade-off curve (Fig7)

To create a tradable commodity the amount of carbonsequestered must be verified thereby incurring monitoringcosts (these costs are not considered in this study) Monitoringcosts affect the overall efficiency of a payment scheme Themagnitude of these costs will be highly dependent upon policydesign with the payment per hectare scheme potentially havingless need for rigorous measurement of carbon sequestrationand hence lower administration costs Previous findings havevaried with some studies (Parks and Hardie 1995 Stavins1999) suggesting that the extra costs associated withmeasurement of carbon sequestration were likely to outweighefficiency gains from using a policy that accounts for bothcosts and benefits On the other hand others have found thatefficiency gains from accounting for farm scale variation insoil carbon sequestration often outweighed increasedtransaction costs (Antle 2003) These equivocal findingshighlight the importance of context-specific assessment forinforming efficient policy design

The high efficiency of the payment per hectare scheme itsrobustness to variation in economic parameters as well as thepotential for lower administration costs may make this policyalternative an attractive option However moral hazardassociated with the payment per hectare scheme poses asignificant risk The payment per tonne scheme provideslandholders with incentives to ensure that carbon sequestrationis validated and maintained It may promote technologicalinnovation by landholders leading to increased efficiency(Antle 2003) Additional contractual requirements may benecessary under the payment per hectare scheme to ensure thatdue care is taken in the establishment and maintenance phasesof reforestation Such an input-based contract combined withregular compliance checks may incur lower costs thanmeasurement of carbon sequestered in forests required underthe payment per tonne scheme while maintaining a high levelof compliance The effectiveness of input standards dependsupon the clear specification of inputs although higher levelsof specificity will increase costs of ensuring that inputs are met

Limitations and future directions

Refinement and validationRelative variance of carbon sequestration and agriculturalproduction are key determinants of food-carbon trade-offsunder alternative payment schemes We modeled the spatialdistribution of these processes by integrating a range ofeconomic and biophysical data that vary in their spatial scaleand detail affecting relative variance in costs and benefitsFuture work should focus on refining and validating the factors

Ecology and Society 17(3) 21httpwwwecologyandsocietyorgvol17iss3art21

driving this relative variability Landholder surveys can be aneffective approach for validating landscape-scale modeling(Bryan et al 2011a) A key parameter to reassess is theassumption of homogeneous establishment costs The 3-PGmodeling used also has limitations To date there have beenfew opportunities for validation of mallee species parameterfiles in the study area Ongoing refinements are being madeto both the model and species growth parameters Increasingthe spatial refinement of the model is a useful direction forfuture research Higher resolution studies can capture thepotential of small scale reforestation eg agroforestry stripplantings which can also have positive feedbacks foragricultural productivity eg through provision of livestockshelter water table maintenance erosion control Significantpotential exists for mixed production to increase carbonsequestration with minimal losses to agricultural production(Monjardino et al 2010)

Landholder decision makingWe assessed carbon sequestration based on economicprofitability Although this is a common approach for policyassessment (Antle and Valdivia 2006 Bryan et al 20082010a 2010b Hunt 2008 Dymond et al 2012) in realitymany other factors also affect actual land use decision making(Pannell et al 2011) Many of these factors that includepersonal values preferences attitudes and situations eghealth are difficult to quantify over large areas althoughattempts are being made (eg Raymond et al 2009) Empiricalanalyses using land use change observations to estimate costsof reforestation have found that landholders often retainexisting practices after it is profitable to change land use(Plantinga et al 2001 Lubowski et al 2008) The rationaldesire of landholders to take time to investigate and trial newtechnologies (Pannell et al 2011) may account for delayeduptake of new land uses Uncertainty about future pricesattaches an option value to the status quo (Plantinga et al2002) Once changed costs of converting land use are sunkand a return to the original land use will impose extra costsand for reforestation these are likely to be high Delaying landuse conversion retains the option of future land use conversionAn empirical analysis of agricultural plots in Georgia USAfound that the option value of a land asset can range from 7to 81 of its expected value (Schatzki 2003) Extending thestatic net present value framework used in this paper toconsider price and production volatility within an option valueframework is likely to increase the predictive power of theresults There are also significant limits to physical capacityeg labor seed stock machinery capital that can affect ratesof reforestation Expansion of eucalypt and pine plantationsover the past 20 years was restricted by availability of theseresources despite incentives from state and federalgovernments and managed investment schemes to increaseplantations (Polglase et al 2011) Future rates of reforestationare highly uncertain and depend upon the design of policy andinstitutional arrangements

Other cobenefits and trade-offsAlthough we focus here on food-carbon trade-offs othercobenefits and trade-offs associated with the reforestation ofagricultural land under a carbon market also existReforestation can have substantial cobenefits for economicdevelopment through increased landholder incomes (Bryanet al 2011b) biodiversity through environmental plantings(Crossman et al 2011) soils through reduced erosion andsalinization (Bartle et al 2007 Harper et al 2007) and energysecurity through bioenergy processing (Bryan et al 20082010a 2010b) However reforestation may also have othersocioeconomic and environmental costs such as reduced wateravailability Integrated assessment of reforestation impacts isrequired to fully account for interactions between complexcobenefits and trade-offs which vary with location type andarea of reforestation and the preferences and values of society(Nelson et al 2008 2009 Bryan et al 2011b) Considerationof the value of cobenefits and trade-offs for ecosystem servicesfor both reforestation and agriculture will change the relativeeconomic viability of these land uses Understanding thesetrade-offs and cobenefits especially how they vary underclimate change is the focus of ongoing work

CONCLUSIONWe have combined detailed spatio-temporal biophysical andeconomic information to assess food-carbon trade-offs andpolicy under two common carbon payment schemes Therelative ability of land units to sequester carbon and producefood was heterogeneous across the study area leading toconvex production frontiers reflecting food-carbon trade-offsThe nature of these trade-offs highlight the potential tosequester carbon in the landscape with a limited impact onagricultural production Our analysis of both payment pertonne and payment per hectare schemes suggests that bothpolicy instruments may achieve very efficient outcomesVariation in model parameters had a strong effect on the priceof carbon at which reforestation became more profitable thanagriculture and affected the efficiency of the payment pertonne scheme but not the payment per hectare scheme Thisrobustness and potential for reduced transaction costs areattractive qualities of the payment per hectare scheme althoughthe risks of moral hazard under this scheme need to be carefullymanaged The results can inform the design of policy and haveilluminated factors that may be important for futureconsideration of policy efficiency Further work to betterdevelop the land use decision model will further improve ourunderstanding of food-carbon trade-offs

Responses to this article can be read online athttpwwwecologyandsocietyorgvol17iss3art21responses

Ecology and Society 17(3) 21httpwwwecologyandsocietyorgvol17iss3art21

Acknowledgments

We are grateful for the support of CSIROs Integrated CarbonPathways project and Sustainable Agriculture Flagship andthe Land Technologies Alliance We thank all the agencies andindividuals who supplied us with data especially Jenny Carter(CSIRO) for the 3-PG species parameter file StephanieMcWhinnie helped guide the research and comments from JeffConnor and Michael Battaglia improved the manuscript

LITERATURE CITEDAlexandrov G A 2008 Forest growth in the light of thethermodynamic theory of ecological systems EcologicalModelling 216102-106

Alig R G Latta D Adams and B McCarl 2010 Mitigatinggreenhouse gases the importance of land base interactionsbetween forests agriculture and residential development inthe face of changes in bioenergy and carbon prices ForestPolicy and Economics 1267-75 httpdxdoiorg101016jforpol200909012

Antle J 2003 Spatial heterogeneity contract design and theefficiency of carbon sequestration policies for agricultureJournal of Environmental Economics and Management 46231-250 httpdxdoiorg101016S0095-0696(02)00038-4

Antle J M and R O Valdivia 2006 Modelling the supplyof ecosystem services from agriculture a minimum-dataapproach Australian Journal of Agricultural and ResourceEconomics 501-15 httpdxdoiorg101111j1467-8489200600315x

Babcock B A P G Lakshminarayan J J Wu and DZilberman 1997 Targeting tools for the purchase ofenvironmental amenities Land Economics 73325-339 httpdxdoiorg1023073147171

Bartle J G Olsen D Cooper and T Hobbs 2007 Scale ofbiomass production from new woody crops for salinity controlin dryland agriculture in Australia International Journal ofGlobal Energy Issues 27115-137 httpdxdoiorg101504IJGEI2007013652

Bryan B A N D Crossman D King and W S Meyer2011b Landscape futures analysis assessing the impacts ofenvironmental targets under alternative spatial policy optionsand future scenarios Environmental Modelling amp Software 2683-91 httpdxdoiorg101016jenvsoft201003034

Bryan B A S Hajkowicz S Marvanek and M D Young2009 Mapping economic returns to agriculture for informingenvironmental policy in the Murray-Darling Basin AustraliaEnvironmental Modeling amp Assessment 14375-390 httpdxdoiorg101007s10666-008-9144-8

Bryan B A D King and E Wang 2010a Potential of woodybiomass production for motivating widespread naturalresource management under climate change Land Use Policy 27713-725 httpdxdoiorg101016jlandusepol200909012

Bryan B A D King and E Wang 2010b Biofuelsagriculture landscape-scale trade-offs between fueleconomics carbon energy food and fiber Global ChangeBiology Bioenergy 2330-345 httpdxdoiorg101111j1757-1707201001056x

Bryan B A D King and J R Ward 2011a Modelling andmapping agricultural opportunity costs to guide landscapeplanning for natural resource management EcologicalIndicators 11199-208 httpdxdoiorg101016jecolind200902005

Bryan B A J Ward and T Hobbs 2008 An assessment ofthe economic and environmental potential of biomassproduction in an agricultural region Land Use Policy 25533-549 httpdxdoiorg101016jlandusepol200711003

Cacho O J R L Hean and R M Wise 2003 Carbon-accounting methods and reforestation incentives AustralianJournal of Agricultural and Resource Economics 47153-179httpdxdoiorg1011111467-848900208

Cacho O J G R Marshall and M Milne 2005 Transactionand abatement costs of carbon-sink projects in developingcountries Environment and Development Economics 10597-614 httpdxdoiorg101017S1355770X05002056

Cantarello E A C Newton and R A Hill 2011 Potentialeffects of future land-use change on regional carbon stocks inthe UK Environmental Science amp Policy 1440-52 httpdxdoiorg101016jenvsci201010001

Chen X F Lupi A Vintildea G He and J Liu 2010 Usingcost effective targeting to enhance the efficiency ofconservation investments in payments for ecosystem servicesConservation Biology 241469-1478 httpdxdoiorg101111j1523-1739201001551x

Crossman N D and B A Bryan 2009 Identifying cost-effective hotspots for restoring natural capital and enhancinglandscape multifunctionality Ecological Economics 68654-668httpdxdoiorg101016jecolecon200805003

Crossman N D B A Bryan and D M Summers 2011Carbon payments and low-cost conservation ConservationBiology 25835-845 httpdxdoiorg101111j1523-1739201101649x

DeFries R S J A Foley and G P Asner 2004 Land-usechoices balancing human needs and ecosystem functionFrontiers in Ecology and the Environment 2249-257 httpd

Ecology and Society 17(3) 21httpwwwecologyandsocietyorgvol17iss3art21

xdoiorg1018901540-9295(2004)002[0249LCBHNA]20CO2

Dymond J R A-G E Ausseil J C Ekanayake and M UF Kirschbaum 2012 Tradeoffs between soil water andcarbon - a national scale analysis from New Zealand Journalof Environmental Management 95124-131 httpdxdoiorg101016jjenvman201109019

Ferraro P J 2003 Assigning priority to environmental policyinterventions in a heterogeneous world Journal of PolicyAnalysis and Management 2227-43 httpdxdoiorg101002pam10094

Foley J A R DeFries G P Asner C Barford G BonanS R Carpenter F S Chapin M T Coe G C Daily H KGibbs J H Helkowski T Holloway E A Howard C JKucharik C Monfreda J A Patz I C Prentice NRamankutty and P K Snyder 2005 Global consequences ofland use Science 309570-574 httpdxdoiorg101126science1111772

Golub A T Hertel H-L Lee S Rose and B Sohngen2009 The opportunity cost of land use and the global potentialfor greenhouse gas mitigation in agriculture and forestryResource and Energy Economics 31299-319 httpdxdoiorg101016jreseneeco200904007

Harper R J A C Beck P Ritson M J Hill C D MitchellD J Barrett K R J Smettem and S S Mann 2007 Thepotential of greenhouse sinks to underwrite improved landmanagement Ecological Engineering 29329-341 httpdxdoiorg101016jecoleng200609025

Higgins A J S Hajkowicz and E Bui 2008 A multi-objective model for environmental investment decisionmaking Computers amp Operations Research 35253-266 httpdxdoiorg101016jcor200602027

Hunt C 2008 Economy and ecology of emerging marketsand credits for bio-sequestered carbon on private land intropical Australia Ecological Economics 66309-318 httpdxdoiorg101016jecolecon200709012

Irwin E G and J Geoghegan 2001 Theory data methodsdeveloping spatially explicit economic models of land usechange Agriculture Ecosystems amp Environment 857-24 httpdxdoiorg101016S0167-8809(01)00200-6

Landsberg J J and R H Waring 1997 A generalised modelof forest productivity using simplified concepts of radiation-use efficiency carbon balance and partitioning ForestEcology and Management 95209-228 httpdxdoiorg101016S0378-1127(97)00026-1

Lawson K K Burns K Low E Heyhoe and H Ahammad2008 Analysing the economic potential of forestry for carbonsequestration under alternative carbon price paths AustralianBureau of Agricultural and Resource Economics Canberra

Australia [online] URL httparchivetreasurygovaulowpollutionfutureconsultants_reportdownloadsEconomic_Potential_of_Forestrypdf

Lubowski R N A J Plantinga and R N Stavins 2006Land-use change and carbon sinks econometric estimation ofthe carbon sequestration supply function Journal ofEnvironmental Economics and Management 51135-152 httpdxdoiorg101016jjeem200508001

Lubowski R N A J Plantinga and R N Stavins 2008What drives land-use change in the United States A nationalanalysis of landowner decisions Land Economics 84529-550

Monjardino M D Revell and D J Pannell 2010 Thepotential contribution of forage shrubs to economic returnsand environmental management in Australian drylandagricultural systems Agricultural Systems 103187-197 httpdxdoiorg101016jagsy200912007

Nelson E G Mendoza J Regetz S Polasky H Tallis DR Cameron K M A Chan G C Daily J Goldstein P MKareiva E Lonsdorf R Naidoo T H Ricketts and M RShaw 2009 Modeling multiple ecosystem servicesbiodiversity conservation commodity production andtradeoffs at landscape scales Frontiers in Ecology and theEnvironment 74-11 httpdxdoiorg101890080023

Nelson E S Polasky D J Lewis A J Plantinga ELonsdorf D White D Bael and J JLawler 2008 Efficiencyof incentives to jointly increase carbon sequestration andspecies conservation on a landscape Proceedings of theNational Academy of Sciences of the United States of America 1059471-9476 httpdxdoiorg101073pnas0706178105

Newburn D S Reed P Berck and A Merenlender 2005Economics and land-use change in prioritizing private landconservation Conservation Biology 191411-1420 httpdxdoiorg101111j1523-1739200500199x

Obersteiner M H Boumlttcher and Y Yamagata 2010Terrestrial ecosystem management for climate changemitigation Current Opinion in Environmental Sustainability 2271-276 httpdxdoiorg101016jcosust201005006

Palm C A S M Smukler C C Sullivan P K Mutuo GI Nyadzi and M G Walsh 2010 Identifying potentialsynergies and trade-offs for meeting food security and climatechange objectives in sub-Saharan Africa Proceedings of theNational Academy of Sciences of the United States of America 10719661-19666 httpdxdoiorg101073pnas0912248107

Pannell D G R Marshall N Barr F Vanclay and RWilkinson 2011 Adoption of new practises by rurallandholders Pages 11-39 in D Pannell and F Vanclay editors

Ecology and Society 17(3) 21httpwwwecologyandsocietyorgvol17iss3art21

Changing land management CSIRO Publishing CollingwoodVictoria Australia

Parks P J and I W Hardie 1995 Least-cost forest carbonreserves cost-effective subsidies to convert marginalagricultural land to forests Land Economics 71122-136 httpdxdoiorg1023073146763

Plantinga A J R Alig and H T Cheng 2001 The supplyof land for conservation uses evidence from the conservationreserve program Resources Conservation and Recycling 31199-215 httpdxdoiorg101016S0921-3449(00)00085-9

Plantinga A J R N Lubowski and R N Stavins 2002 Theeffects of potential land development on agricultural landprices Journal of Urban Economics 52561-581 httpdxdoiorg101016S0094-1190(02)00503-X

Plantinga A J T Mauldin and D J Miller 1999 Aneconometric analysis of the costs of sequestering carbon inforests American Journal of Agricultural Economics 81812-824 httpdxdoiorg1023071244326

Polasky S E Nelson J Camm B Csuti P Fackler ELonsdorf C Montgomery D White J Arthur B Garber-Yonts R Haight J Kagan A Starfield and C Tobalske2008 Where to put things Spatial land management to sustainbiodiversity and economic returns Biological Conservation 1411505-1524 httpdxdoiorg101016jbiocon200803022

Polglase P K Paul C Hawkins A Siggins J Turner TBooth D Crawford T Jovanovic T Hobbs K Opie AAlmeida and J Carter 2008 Regional opportunities foragroforestry systems in Australia Rural Industries Researchand Development Corporation Canberra Australia [online]URL httpsrirdcinfoservicescomaudownloads08-176

Polglase P A Reeson C Hawkins K Paul A Siggins JTurner DCrawford T Jovanovic T Hobbs K Opie JCarwardine and A Almeida 2011 Opportunities for carbonforestry in Australia economic assessment and constraints toimplementation CSIRO Technical Report CSIROPublishing Collingwood Victoria Australia [online] URL httpwwwcsiroaufilesfilesp10tzpdf

Raymond C M B A Bryan D H MacDonald A Cast SStrathearn A Grandgirard and T Kalivas 2009 Mappingcommunity values for natural capital and ecosystem servicesEcological Economics 681301-1315 httpdxdoiorg101016jecolecon200812006

Reserve Bank of Australia (RBA) 2010 Cash Rate Target2001-2010 Reserve Bank of Australia Sydney Australia

Richards F J 1959 A flexible growth function for empiricaluse Journal of Experimental Botany 10290-301 httpdxdoiorg101093jxb102290

Schatzki T 2003 Options uncertainty and sunk costs anempirical analysis of land use change Journal ofEnvironmental Economics and Management 4686-105 httpdxdoiorg101016S0095-0696(02)00030-X

Smith P P J Gregory D van Vuuren M Obersteiner PHavlik M Rounsevell J Woods E Stehfest and J Bellarby2010 Competition for land Philosophical Transactions of theRoyal Society B-Biological Sciences 3652941-2957 httpdxdoiorg101098rstb20100127

Stavins R N 1999 The costs of carbon sequestration arevealed-preference approach American Economic Review 89994-1009 httpdxdoiorg101257aer894994

Stoms D M J Kreitler and F W Davis 2011 The powerof information for targeting cost-effective conservationinvestments in multifunctional farmlands EnvironmentalModelling amp Software 268-17 httpdxdoiorg101016jenvsoft201003008

Thomson A M K V Calvin L P Chini G Hurtt J AEdmonds B Bond-Lamberty S Frolking M A Wise andA C Janetos 2010 Climate mitigation and the future oftropical landscapes Proceedings of the National Academy ofSciences 10719633-19638 httpdxdoiorg101073pnas0910467107

Tilman D R Socolow J A Foley J Hill E Larson LLynd S Pacala J Reilly T Searchinger C Somerville andR Williams 2009 Beneficial biofuels - the food energy andenvironment trilemma Science 325270-271 httpdxdoiorg101126science1177970

Van der Werf E and S Peterson 2009 Modeling linkagesbetween climate policy and land use an overviewAgricultural Economics 40507-517 httpdxdoiorg101111j1574-0862200900394x

Waumltzold F and M Drechsler 2005 Spatially uniform versusspatially heterogeneous compensation payments forbiodiversity-enhancing land-use measures Environmentaland Resource Economics 3173-93 httpdxdoiorg101007s10640-004-6979-6

West P C H K Gibbs C Monfreda J Wagner C CBarford S R Carpenter and J A Foley 2010 Trading carbonfor food global comparison of carbon stocks vs crop yieldson agricultural land Proceedings of the National Academy ofSciences of the United States of America 10719645-19648 httpdxdoiorg101073pnas1011078107

Wise M K Calvin A Thomson L Clarke B Bond-Lamberty R Sands S J Smith A Janetos and J Edmonds2009 Implications of limiting CO2 concentrations for land useand energy Science 3241183-1186 httpdxdoiorg101126science1168475

Ecology and Society 17(3) 21httpwwwecologyandsocietyorgvol17iss3art21

Wu J J and W G Boggess 1999 The optimal allocationof conservation funds Journal of Environmental Economicsand Management 38302-321 httpdxdoiorg101006jeem19991091

Zhao-gang L and L Feng-ri 2003 The generalizedChapman-Richards function and applications to tree and standgrowth Journal of Forestry Research 1419-26 httpdxdoiorg101007BF02856757