conservation of undisturbed natural forests and economic impacts on agriculture

11
Land Use Policy 30 (2013) 344–354 Contents lists available at SciVerse ScienceDirect Land Use Policy jou rn al h om epa ge: www.elsevier.com/locate/landusepol Conservation of undisturbed natural forests and economic impacts on agriculture Michael Krause , Hermann Lotze-Campen, Alexander Popp, Jan Philipp Dietrich, Markus Bonsch Potsdam Institute for Climate Impact Research, PO Box 601203, 14412 Potsdam, Germany a r t i c l e i n f o Article history: Received 18 April 2011 Received in revised form 22 March 2012 Accepted 24 March 2012 Keywords: Undisturbed natural forest Carbon storage Opportunity costs Agricultural land expansion Avoided deforestation Avoided carbon emissions a b s t r a c t Conservation of undisturbed natural forests, which are important for biodiversity, carbon storage, and other ecosystem services, affects agricultural production and cropland expansion. We analyze the eco- nomic impacts of undisturbed natural forest conservation programs on agriculture and the magnitude of avoided deforestation and avoided carbon emissions in the tropics. We apply a global agricultural land use model to estimate changes in agricultural production costs for the period 2015–2055. Our forest conservation scenarios reflect two different policy goals: either maximize forest carbon storage or min- imize impacts on agricultural production. In all the scenarios, the economic impacts on agriculture are relatively low. Production costs would increase due to forest conservation by a maximum of 4%, predom- inantly driven by increased investments in agricultural productivity increase. We also show regional differences in Latin America, Sub-Saharan Africa, and Southeast Asia, due to different growth rates in food demand, land availability and crop productivity. The area of avoided deforestation does not exceed 1.5 million ha yr 1 in the period 2015–2055, while avoided carbon emissions reach a maximum of 1.9 Gt CO2 per year. According to our results on the potential changes in agricultural production costs, undis- turbed natural forest conservation appears to be a low-cost option for greenhouse gas emission reduction. © 2012 Elsevier Ltd. All rights reserved. Introduction There is rising awareness in science and policy of the poten- tial scarcity of land for an increasing number of future uses. Land is mainly used for food, feed, fiber, bioenergy, and wood produc- tion as well as infrastructure. In addition, land is reserved for carbon storage, biodiversity conservation, and other ecosystem services (Eliash, 2008; RFA, 2008; Roberts et al., 2008; Fischer et al., 2002; FAO, 2002; v. Velthuizen et al., 2007; Popp et al., 2011a; Lotze-Campen et al., 2010). Natural forest ecosystems, espe- cially tropical natural forests (Laurance, 2007) and tropical primary forests (Barlow et al., 2007), provide carbon storage (Jackson et al., 2008; Bonan, 2008; Gumpenberger et al., 2010) and maintain biodi- versity (Brooks et al., 2006). These valuable services are threatened by lasting deforestation (FAO, 2006) as well as human-induced degradation and fragmentation (Turner, 1996; Gullison et al., 2007). Several studies in the literature highlight the importance of forests with respect to ecosystem services. Schmitt et al. (2009) emphasize the insufficient protection of non-fragmented natural forests in the tropics and subtropics within global priority areas for ecosystem conservation. Brooks et al. (2006) prioritize undisturbed natural forest ecosystems (Bryant et al., 1997: 12; Greenpeace International, 2005) to be conserved for their high biodiversity. Corresponding author. Tel.: +49 0331 288 2457. E-mail address: [email protected] (M. Krause). Forest conservation programs will only be successful if they take the economic impacts on alternative land uses explicitly into account. Grig-Gran (2006) quantifies the costs of avoided deforestation in terms of foregone agricultural income, based on costs of alterna- tive production systems at the country scale. At the global scale, Kindermann et al. (2008) estimate the costs of avoided deforesta- tion at different price levels for forest carbon, based on the change in forestry land values relative to agriculture. Neither of the two studies focusses on particular forest types. Mittermeier et al. (2003) provide rough estimates of the costs of conserving partly forested wilderness for biodiversity purposes. Apart from the forest type to be conserved, the spatial design and prioritization of conservation programs have to be defined. Grig-Gran (2006) circumvents the issue by reducing historical deforestation rates at aggregate country scale, while Kindermann et al. (2008) build on spatially explicit changes in land values to gen- erate spatial patterns of avoided deforestation. There are basically two alternative goals to be pursued. First, forest conservation pro- grams could be designed to maximize the provision of ecosystem services (e.g. maximize stored carbon). Alternatively, the impact of forest conservation programs on alternative land uses could be minimized (e.g. minimize foregone income in agriculture). To our knowledge, no study has yet compared these two options based on a comprehensive bio-economic modeling approach. In this paper, we apply a spatially explicit global land use model to address the following research questions: What are the economic impacts of undisturbed natural forest conservation strategies on 0264-8377/$ see front matter © 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.landusepol.2012.03.020

Upload: michael-krause

Post on 29-Nov-2016

235 views

Category:

Documents


7 download

TRANSCRIPT

Page 1: Conservation of undisturbed natural forests and economic impacts on agriculture

C

MP

a

ARRA

KUCOAAA

I

titcse2cf2vbd

fefenI

0h

Land Use Policy 30 (2013) 344– 354

Contents lists available at SciVerse ScienceDirect

Land Use Policy

jou rn al h om epa ge: www.elsev ier .com/ locate / landusepol

onservation of undisturbed natural forests and economic impacts on agriculture

ichael Krause ∗, Hermann Lotze-Campen, Alexander Popp, Jan Philipp Dietrich, Markus Bonschotsdam Institute for Climate Impact Research, PO Box 601203, 14412 Potsdam, Germany

r t i c l e i n f o

rticle history:eceived 18 April 2011eceived in revised form 22 March 2012ccepted 24 March 2012

eywords:ndisturbed natural forestarbon storagepportunity costsgricultural land expansion

a b s t r a c t

Conservation of undisturbed natural forests, which are important for biodiversity, carbon storage, andother ecosystem services, affects agricultural production and cropland expansion. We analyze the eco-nomic impacts of undisturbed natural forest conservation programs on agriculture and the magnitude ofavoided deforestation and avoided carbon emissions in the tropics. We apply a global agricultural landuse model to estimate changes in agricultural production costs for the period 2015–2055. Our forestconservation scenarios reflect two different policy goals: either maximize forest carbon storage or min-imize impacts on agricultural production. In all the scenarios, the economic impacts on agriculture arerelatively low. Production costs would increase due to forest conservation by a maximum of 4%, predom-inantly driven by increased investments in agricultural productivity increase. We also show regional

voided deforestationvoided carbon emissions

differences in Latin America, Sub-Saharan Africa, and Southeast Asia, due to different growth rates infood demand, land availability and crop productivity. The area of avoided deforestation does not exceed1.5 million ha yr−1 in the period 2015–2055, while avoided carbon emissions reach a maximum of 1.9 GtCO2 per year. According to our results on the potential changes in agricultural production costs, undis-turbed natural forest conservation appears to be a low-cost option for greenhouse gas emission reduction.

ntroduction

There is rising awareness in science and policy of the poten-ial scarcity of land for an increasing number of future uses. Lands mainly used for food, feed, fiber, bioenergy, and wood produc-ion as well as infrastructure. In addition, land is reserved forarbon storage, biodiversity conservation, and other ecosystemervices (Eliash, 2008; RFA, 2008; Roberts et al., 2008; Fischert al., 2002; FAO, 2002; v. Velthuizen et al., 2007; Popp et al.,011a; Lotze-Campen et al., 2010). Natural forest ecosystems, espe-ially tropical natural forests (Laurance, 2007) and tropical primaryorests (Barlow et al., 2007), provide carbon storage (Jackson et al.,008; Bonan, 2008; Gumpenberger et al., 2010) and maintain biodi-ersity (Brooks et al., 2006). These valuable services are threatenedy lasting deforestation (FAO, 2006) as well as human-inducedegradation and fragmentation (Turner, 1996; Gullison et al., 2007).

Several studies in the literature highlight the importance oforests with respect to ecosystem services. Schmitt et al. (2009)mphasize the insufficient protection of non-fragmented naturalorests in the tropics and subtropics within global priority areas for

cosystem conservation. Brooks et al. (2006) prioritize undisturbedatural forest ecosystems (Bryant et al., 1997: 12; Greenpeace

nternational, 2005) to be conserved for their high biodiversity.

∗ Corresponding author. Tel.: +49 0331 288 2457.E-mail address: [email protected] (M. Krause).

264-8377/$ – see front matter © 2012 Elsevier Ltd. All rights reserved.ttp://dx.doi.org/10.1016/j.landusepol.2012.03.020

© 2012 Elsevier Ltd. All rights reserved.

Forest conservation programs will only be successful if they take theeconomic impacts on alternative land uses explicitly into account.Grig-Gran (2006) quantifies the costs of avoided deforestation interms of foregone agricultural income, based on costs of alterna-tive production systems at the country scale. At the global scale,Kindermann et al. (2008) estimate the costs of avoided deforesta-tion at different price levels for forest carbon, based on the changein forestry land values relative to agriculture. Neither of the twostudies focusses on particular forest types. Mittermeier et al. (2003)provide rough estimates of the costs of conserving partly forestedwilderness for biodiversity purposes.

Apart from the forest type to be conserved, the spatial designand prioritization of conservation programs have to be defined.Grig-Gran (2006) circumvents the issue by reducing historicaldeforestation rates at aggregate country scale, while Kindermannet al. (2008) build on spatially explicit changes in land values to gen-erate spatial patterns of avoided deforestation. There are basicallytwo alternative goals to be pursued. First, forest conservation pro-grams could be designed to maximize the provision of ecosystemservices (e.g. maximize stored carbon). Alternatively, the impactof forest conservation programs on alternative land uses could beminimized (e.g. minimize foregone income in agriculture). To ourknowledge, no study has yet compared these two options based on

a comprehensive bio-economic modeling approach.

In this paper, we apply a spatially explicit global land use modelto address the following research questions: What are the economicimpacts of undisturbed natural forest conservation strategies on

Page 2: Conservation of undisturbed natural forests and economic impacts on agriculture

se Pol

aPse

hc

elSmravs

aatsot

M

ME

r2gcctdeavt2dpvctua2pcoptttarlcc

at

M. Krause et al. / Land U

gricultural production in Sub-Saharan Africa, Latin America andacific Asia? What are the benefits of different forest conservationtrategies in terms of avoided deforestation and avoided net carbonmissions from land use change?

A consistent, spatially explicit land use budgeting approachelps to better initialize different land pools and track land usehanges due to agricultural expansion and forest conservation.

Opponents of large-scale forest reservation for the provision ofcosystem services argue that enforcement without involvement ofocal stakeholders would be questionable (Hayes and Ostrom, 2005;chwartzman et al., 2000). In this paper we assume that decisionaking of local stakeholders is solely determined by the economic

ationale of optimizing their net benefits from using the land. Thispproach allows us to quantify the implicit costs of forest conser-ation through foregone benefits from other landuse activities, theo called “opportunity costs”.

The next section introduces the model, the land allocation mech-nism and underlying assumptions. The conceptual embeddingnd calculation of implicit costs (i.e. “opportunity costs”) of undis-urbed natural forest conservation as well as the forest conservationcenarios are briefly described. Section “Results” provides modelutput under different forest conservation scenarios followed byhe sections “Discussion” and “Conclusions”.

aterial and methods

odel of Agricultural Production and its Impact on thenvironment

The Model of Agricultural Production and its Impact on the Envi-onment (MAgPIE) (Lotze-Campen et al., 2008; Popp et al., 2010,011b; Schmitz et al., 2012) is a spatially explicit recursive-dynamiclobal land use optimization model which minimizes the totalosts of agricultural production in decadal time steps until 2055. Itovers the most important agricultural crop and livestock produc-ion types in 10 economic regions worldwide to meet commodityemand. Spatially explicit bio-physical constraints and regionalconomic conditions are taken into account. Obtainable crop yieldsnd carbon contents of forests are provided by the global dynamicegetation model LPJmL at 0.5 arc degree resolution, which is equalo 50 km × 50 km at the equator (Sitch et al., 2003; Bondeau et al.,007; Fader et al., 2010). Productive land enters commodity pro-uction as an input which is limited by the historically derivedhysical crop area (Fader et al., 2010) as well as additional con-ertible unused land. Varying crop yields based on bio-physicalonditions in different locations determine the production costs peron of output which leads to distinct patterns of agricultural landse. Input costs per hectare for labor, chemicals and other capitalre calculated from the GTAP database (Narayanan and Walmsley,008). Rotational constraints define maximum shares of crop typeser grid cell which are related to average crop rotations and agri-ultural management. Depending on the spatially explicit distancef cropland to major urban centers, intra-regional transport costser ton of agricultural output are added to agricultural produc-ion costs. Transport cost estimates are derived from GTAP totalransport costs (Narayanan and Walmsley, 2008), total transportime needed due to distance to major urban centers (Nelson, 2008),nd total production quantity. International trade is constrained byegional minimum self-sufficiency rates. This means that a certainevel of consumption has to be fulfilled within the region. The restan be produced in other world regions according to comparative

ost advantages (Lotze-Campen et al., 2010).

Cropland expansion into convertible unused land is regardeds one option to align agricultural production with projectedotal food consumption. It is associated with additional costs for

icy 30 (2013) 344– 354 345

infrastructure, land clearing, and site preparation. These costs arecalculated from the access costs of forest land in equilibrium pro-vided by the Global Timber Model (GTM) database (Sohngen et al.,2009). As a second option for increasing production, the model caninvest in agricultural research and development (R&D) for tech-nological change, in order to increase crop yields. The costs oftechnological change are a function of the regional technology leveland have been derived from data on public expenditure on agri-cultural R&D (Dietrich et al., 2010b). Input costs per hectare alsoincrease with the intensification of agricultural land use (Dietrichet al., 2010a, 2010b; Popp et al., 2011a). As a consequence, shiftsin land use patterns are determined by weighing marginal costs ofland conversion, transport and factor inputs against marginal costsof intensification and the associated increase in transport and factorinputs.

For the purpose of this analysis, the model has been furtherdeveloped. First, spatially explicit datasets for the initialization ofavailable land pools for cropland expansion have been compiled(Krause et al., 2009). The available land pool has been split into“undisturbed natural forest”, i.e. the union of large intact forestlandscapes (Greenpeace International, 2005) and frontier forests(Bryant et al., 1997) in forestry and unused land categories (Erbet al., 2007), and “other available land”, i.e. abandoned croplandplus other natural vegetation not delineated as forest or grazingland (see Appendix A). Datasets are harmonized at a resolution of0.5 arc degree.

Second, the land allocation mechanism has been expanded byincluding additional land pools, rules, and cost types for croplandexpansion and forest conservation activities. In our analysis wefocus on cropland. Urban land and grazing land are kept constantin area. If land in the cropland pool gets scarce, additional landcan be made available for cropland expansion from the two addi-tional land pools, undisturbed natural forest and other availableland (Table A.1). Abandoned cropland enters the other availableland pool as succession leads to natural re-vegetation.

Undisturbed natural forest conservation activities do notdirectly compete for land. They are normatively set by internationalconservation policies in different scenarios. Appendix B provides amathematical description of model changes.

Economic impact on agriculture due to forest protection

A number of calculations are made in a post-processing proce-dure after model outputs have been generated. Undisturbed naturalforest conservation activities restrict the location and magnitude ofavailable unused land for crop production. Therefore, undisturbednatural forest conservation activities change the costs of produc-tion. The analysis of economic impacts from undisturbed naturalforest conservation is based on the concept of opportunity costs.Generally, opportunity costs of an actual activity are defined as theforegone net benefits from not conducting the next best activity(v. Wieser, 1928). In our context, opportunity costs of a certainland-use activity are the foregone net returns per hectare of thenext-best alternative type of land use. If the actual land use activ-ity pertains to the conservation of undisturbed natural forest, theopportunity costs per hectare indicate the implicit economic valueof undisturbed natural forest and the related ecosystem services,e.g. stored carbon. Based on this concept, the costs of establishinga global undisturbed natural forest conservation program for theprovision of ecosystem services to global society can be calculatedas the foregone net returns from alternative land use activities. Theagricultural sector may thus be compensated for providing avoided

net carbon emissions or conserved biodiversity from foregone landuse change. Opportunity cost estimates may serve as an indica-tor for the magnitude of compensation payments to make globalsociety better off but agricultural producers not worse off.
Page 3: Conservation of undisturbed natural forests and economic impacts on agriculture

3 se Policy 30 (2013) 344– 354

ufonaWnTtrfioontimoatsTw

S

fbafiuemns

encl

ei(sto

uusTtMeabotwossF

46 M. Krause et al. / Land U

In MAgPIE, opportunity costs can be calculated for differentndisturbed natural forest conservation programs because dif-erent spatial and temporal scales may show different impactsn agriculture. Total agricultural production costs in the Busi-ess As Usual (BAU) scenario are subtracted from those whichccrue in conservation scenarios for each time step (see below).hile production costs are comparable in the same time step, it is

ot methodologically sound to compare them directly over time.herefore, the present value (PV) of total opportunity costs overime is calculated, which constitutes the sum of foregone future costeductions multiplied by a discount factor. A global discount rate ofve percent is assumed to cover real capital costs. For comparingpportunity cost, if conservation programs differ in the durationf implementation over time, the average annual flow of opportu-ity costs over time is important, which can be derived from theotal PV. Furthermore, we calculate the share of opportunity costsn total agricultural production costs, in order to show the relative

agnitude of foregone benefits. Additionally, estimates of the totalpportunity costs are linked to the avoided deforestation area andssociated avoided net carbon emissions that constitute benefitso global society. Such benefits depend on agricultural productiontrategies and are not solely dependent on conservation efforts.herefore, they indicate the average compensation payments thatould accrue to the global society.

cenario analysis

In our scenario analysis we distinguish two undisturbed naturalorest conservation programs: one that aims at maximizing car-on storage as an ecosystem service, and a second one that aimst minimizing the economic impact on alternative land uses. In therst case, we put priority on the contribution of undisturbed nat-ral forests to climate change mitigation. In the second case, wemphasize the agricultural sector perspective that aims at mini-izing additional costs. The magnitude of conserved undisturbed

atural forest area is comparable in the two scenarios, but carbontorage has a lower priority in the latter one.

As a point of reference, the BAU scenario allows for croplandxpansion into undisturbed natural forest as well as unused otheratural vegetation that is at least marginally suitable for rainfedrop production. Crop suitability is based on the Global AgroEco-ogical Zones project (Fischer et al., 2002) (Table A.2).

In the full protection scenario FC100, all undisturbed natural for-st and unused naturally vegetated land (globally 734 million ha)s excluded from the land pool available for cropland expansionTable A.1). The scenario FC100 serves for contrasting purposes onlyince it is unlikely that 100% undisturbed natural forest conserva-ion will be successful. Therefore it aims at grasping an upper limitf potential opportunity costs.

In FC50-Y and FC50-C scenarios, we reduce the area of protectedndisturbed natural forest to 50%, i.e. 367 million ha globally. Thendisturbed natural forest conservation program is implementedtepwise over time and proportionally distributed across regions.hus, the time needed for effectively implementing the undis-urbed natural forest conservation programs is taken into account.

oreover, the heterogeneity of regional undisturbed natural forestndowments and associated carbon density is covered. The landllocation to protected undisturbed natural forest is determinedy the conservation strategy, i.e. either by minimizing agriculturalpportunity costs or by maximizing carbon storage. Technically,he first strategy gives priority to the conservation of forest areaith the lowest expected crop yields and thus minimizes expected

pportunity costs in agriculture (scenario FC50-Y). The secondtrategy focuses on carbon-rich area first to maximize carbontorage of undisturbed natural forest (scenario FC50-C). Both theC50-Y and FC50-C scenarios will differ from FC100 due to allowed

Fig. 1. Total opportunity costs, BAU versus FC100, 2015–2055.

cropland expansion into remaining undisturbed natural forest,which is referred to as leakage in undisturbed natural forest con-servation.

The analysis focuses on three tropical regions, Sub-SaharanAfrica (AFR), Latin America (LAM) and Pacific Asia (PAS). Theyaccount for 92% of suitable undisturbed natural forests and a sub-stantial part of total undisturbed natural forest area (43%) (Bryantet al., 1997; Greenpeace International, 2005) and have shown thehighest rates of deforestation (FAO, 2006). The complementary Restof World (ROW) is the aggregate of seven world regions in the MAg-PIE model: Centrally-Planned Asia, Europe, Former Soviet Union,Middle East and North Africa, North America, Pacific OECD, andSouth Asia. The country affiliations to these regions are listed inDietrich et al. (2010a: Appendix A).

Results

Total opportunity costs of undisturbed natural forest conservation

The first set of results deals with the total opportunity costsof undisturbed natural forest conservation. Fig. 1 shows both theabsolute value of opportunity costs and their share of agriculturalproduction costs for the full protection scenario FC100 comparedto the BAU scenario without any protection policy. As explainedabove, these opportunity costs are the foregone net income fromagricultural production as a consequence of undisturbed naturalforest conservation.

In total, three tropical regions account for more than 82%of global opportunity costs, i.e. 10 billion US$ from 2015 to2055. Regional opportunity costs per year are between 0.3 and0.5 billion US$, and do not exceed 4% of agricultural productioncosts (PAS) in relative terms.

Disaggregated results on the composition of total opportu-nity costs help to understand the drivers of change (Table C.1).Obviously, agricultural input costs for labor, capital, and chemi-cals are reduced, as less land is taken into production. Likewise,expenditures for preparing additional unused land for agriculturalproduction, e.g. through land clearing and infrastructure, are saved.PAS makes an adequate example of the trend in all regions, a par-tial cost saving in scenario FC100 compared to BAU. Input costs, landconversion costs and, to a minor extent, transport costs are reducedin PAS by −10%, −46%, and −1% respectively. On the other hand,additional investments in R&D over-compensate these cost reduc-tions, leading, in total, to positive opportunity costs in all regions(Table C.1).

In addition to results in FC100, total opportunity costs of50% undisturbed natural forest conservation (FC50-Y, FC50-C) areshown in Fig. 2.

Total opportunity costs in the minimized agricultural impacts

scenario FC50-Y are plausibly smaller or equal to the maximizedcarbon storage scenario FC50-C, except for LAM. LAM showszero opportunity costs in FC50-C which coincides with totaltransport costs that are comparable to FC50-Y, but significantly
Page 4: Conservation of undisturbed natural forests and economic impacts on agriculture

M. Krause et al. / Land Use Policy 30 (2013) 344– 354 347

htmFFauF

teooatt

tteeccsc

t(hY

Fig. 2. Total opportunity costs, FC50 versus BAU, 2015–2055.

igher cropland expansion. This means that transport costs tohe market per ton of output are lower in FC50-C in 2015, and

ore land is taken into production in the vicinity of markets.or AFR, the opportunity cost differences between FC50-Y andC50-C are related to significantly lower average agricultural yieldsnd carbon saved per hectare at similar magnitude of conservedndisturbed natural forest area in 2015 (e.g. 473 tons CO2 ha−1 forC50-Y versus 605 tons CO2 ha−1 for FC50-C).

There are at least three striking results in FC50 scenarioshat deviate from FC100. First, despite undisturbed natural for-st conservation areas being cut by half compared to FC100, totalpportunity cost estimates decline by more than 50%. Second, totalpportunity costs in AFR exceed the values in other regions, in bothbsolute and relative terms. However, they do not exceed 0.8% ofotal agricultural production costs. Third, in LAM the total oppor-unity costs drop to zero in FC50-C.

The economic impacts of undisturbed natural forest conserva-ion on the agricultural sector are linked to quantifiable benefitso global society. Avoided deforestation and avoided net carbonmissions are ascribed to agricultural activities and these benefitsxist in all regions. In Fig. 3a we present the average opportunityosts per hectare of avoided deforestation for the minimized agri-ultural impacts scenario FC50-Y and for the maximized carbontorage scenario FC50-C. The opportunity costs per ton of avoidedarbon emissions for the two scenarios are displayed in Fig. 3b.

The average annual opportunity costs of the minimized agricul-

ural impacts scenario (FC50-Y) and the maximized carbon storageFC50-C) scenario are similar, except for PAS where costs perectare are almost twice as high in FC50-C compared to FC50-. Furthermore, results for FC50-C in LAM and PAS differ, due to

Fig. 3. Average annual opportunity costs, FC50 versus BAU, 2015–2055.

Fig. 4. Avoided deforestation and avoided net carbon emissions, FC50 versus BAU,2015–2055.

similar avoided deforestation areas and avoided net carbon emis-sions, but relatively higher total opportunity costs in PAS.

For the full protection scenario FC100 (Fig. C.2a and b) averageannual opportunity costs of avoided deforestation are at same lev-els in AFR and LAM (370 US$ ha−1) but they are significantly higherin PAS (520 US$ ha−1). Regarding avoided net carbon emissions,average annual opportunity costs per ton CO2 remain only slightlylower in AFR than in LAM. This is related to a similar relationshipbetween avoided net carbon emissions and total opportunity costin the two scenarios.

Benefits in terms of avoided deforestation and avoided emissions

In our second set of results, we quantify the benefits in termsof avoided deforestation and avoided carbon emissions under thethree different conservation scenarios.

In the FC100 (full protection) scenario the area of avoided defor-estation in AFR (1.5 million ha yr−1 from 2015 to 2055) is 36% higherthan in LAM (1.1 million ha yr−1, Fig. C.1a), although AFR has merely22% of the undisturbed natural forest area of LAM (Table A.1). Whilethere is displacement of deforestation activity (leakage) into undis-turbed natural forest by definition in FC100, leakage into otheravailable land does not exceed 0.1 million ha yr−1 in LAM and PAScompared to BAU (Table C.2). The ratio of avoided net carbon emis-sions (Fig. C.1b) divided by avoided deforestation area indicateshigher average carbon density per hectare in AFR than in LAM(150 tons ha−1 versus 130 tons ha−1).

Fig. 4a shows the amount of forest area actually conserved annu-ally for the FC50-Y (minimized agricultural impacts) and the FC50-C(maximized carbon storage) scenarios. The total amount of avoidedcarbon emissions per year for the two scenarios is shown in Fig. 4b.

In LAM and PAS, the amount of avoided deforestation is sig-nificantly higher for the minimized agricultural impacts scenarioFC50-Y compared to FC50-C while avoided emissions are similarfor both scenarios. In AFR, the maximized carbon storage scenarioFC50-C does better in avoiding both deforestation and emissions.

Moreover, FC50-Y and FC50-C scenarios show a significant dis-crepancy between area change rates in available and conserved

undisturbed natural forest (Table C.2). The average annual areaof undisturbed natural forest converted to cropland peaks at1.1 million ha yr−1 between 2015 and 2055 in AFR. Similar to totalopportunity costs, avoided deforestation area drops more than
Page 5: Conservation of undisturbed natural forests and economic impacts on agriculture

3 se Pol

pa

D

opsahortcttlscimrtvr(

as

iamtetcIm(l(iSi

oatbtgcst

rfAn2ten

48 M. Krause et al. / Land U

roportionally compared to FC100 (Table C.2), as do changes ingricultural yields (Table C.3).

iscussion

In a first set of results, we have analyzed the economic impactsf undisturbed natural forest conservation on agricultural croproduction in tropical regions. The economic impacts are mea-ured by the opportunity costs, i.e. the foregone net returns ingriculture due to undisturbed natural forest conservation and,ence, restricted cropland expansion. Total opportunity costs turnut to be relatively low for all scenarios and all regions. Theatio of opportunity costs relative to total agricultural produc-ion costs is low, because input costs for labor, chemicals, orapital account for a high share of total agricultural costs, buthey remain insensitive to undisturbed natural forest conserva-ion strategies. This is caused by two contrary processes. First,ess cropland area is taken into production than in the baselinecenario which reduces input costs. Second, investments in agri-ultural R&D boost crop yields and lead to higher crop-specificnput costs (Dietrich et al., 2010b; Popp et al., 2011a) which

ainly explain differences in total opportunity costs betweenegions and scenarios. Apart from agricultural intensificationhere is also a feedback of undisturbed natural forest conser-ation on cropland expansion into other available land. Thisesult is consistent with Boserup (1965) and Miles and Kapos2008).

In Latin America, the yield based conservation allocation mech-nism (FC50-Y) cannot serve as the minimizing agricultural impactscenario because transport has a higher impact on costs than yields.

In Sub-Saharan Africa, total opportunity costs are higher thann Latin America. R&D costs per unit of output grow according to

power law in our model (Dietrich et al., 2010a, 2010b), whichainly drives opportunity costs in Sub-Saharan Africa in connec-

ion with strong yield increases. Less suitable land for croplandxpansion and the higher demand for food and feed products dueo population and income growth in Sub-Saharan Africa indirectlyontribute to higher total opportunity costs than in Latin America.n the past, production increases in African agriculture have been

ainly achieved by cropland expansion rather than intensificationGeist, 2006: 74). Moreover, Sub-Saharan Africa has experiencedower rates of urbanization than Latin America and Pacific AsiaButler and Laurance, 2008; UN, 2009). Nevertheless, investmentsn agricultural productivity increase have been low, which putsub-Saharan Africa at a disadvantage with respect to the economicmpacts of undisturbed natural forest conservation in the future.

We have also shown results on average annual opportunity costsf undisturbed natural forest conservation. These can be used tonalyze whether forest conservation will be economically attrac-ive as a climate change mitigation option in the future. This woulde the case if the average annual opportunity costs from undis-urbed natural forest conservation are smaller than or equal tolobal carbon prices in the future. However, as there are politi-al and technical constraints to the region-wide implementation ofuch conservation programs (Ebeling and Yasue, 2008), the poten-ial for climate change mitigation remains hypothetical.

The average costs of avoided deforestation in our analysis areelatively low compared to results from Kindermann et al. (2008)or the period 2005–2030 (Africa: 511 US$ ha−1 yr−1; South-Eastsia: 9064 US$ ha−1 yr−1). On the other hand, our results are sig-ificantly higher than those of Grig-Gran (2006) for the period

005–2035 (46–149 US$ ha−1 yr−1 for eight tropical countries). Inhese other studies, future technological change in agriculture isither neglected (Grig-Gran, 2006) or implemented as an exoge-ous trend (Kindermann et al., 2008). Kindermann et al. (2008)

icy 30 (2013) 344– 354

calculate costs of avoided deforestation based on carbon prices,but they do not take rising agricultural production costs dueto intensification or leakage into other natural forest areas intoaccount. Grig-Gran (2006) assumes zero leakage and takes nei-ther increasing agricultural demand nor required investment costsfor agricultural R&D into account. Both add to opportunity costs ifagricultural land is kept constant.

In a second set of results we have addressed the benefits ofundisturbed natural forest conservation, i.e. avoided deforesta-tion and avoided net carbon emissions. Benefits are obtained inall regions. If 100% of undisturbed natural forests are conservedin Sub-Saharan Africa, Latin America and Pacific Asia, the area ofavoided deforestation from 2015 to 2055 in our scenario is between30 and 43% of projected areas of avoided deforestation in the lit-erature (Kindermann et al., 2008). This is mainly due to the factthat we focus on the conservation of undisturbed natural forests,rather than on total forest areas. Accessibility of undisturbed natu-ral forest is limited, compared to other available land with naturalvegetation, which has also been shown by Andam et al. (2008).Loosening conservation efforts to 50% of undisturbed natural for-est area leads to substantial cropland expansion into remainingundisturbed natural forest in Sub-Saharan Africa (i.e. leakage ofcarbon emissions). From an economic perspective, expansion intounprotected undisturbed natural forest delivers cost advantagescompared to intensification. The spatial patterns of leakage aredetermined, by heterogeneous yield distribution across crop typesand grid cells, distance-depending transport costs, and scarcity ofavailable land in non-forest areas.

Large parts of previously intact primary forest in the CongoBasin have already been deforested or are threatened by deforesta-tion (Bryant et al., 1997; Greenpeace International, 2005; Wilkieet al., 2001). In the current situation, expenditures for forestconservation, e.g. in the Congo basin do not match the currentopportunity costs in foregone land uses (Wilkie et al., 2001). Thus,the adequate funding for forest conservation strongly dependson the global society’s willingness to pay for local conservationefforts. Without compensation payments and local stakeholderinvolvement the uncertainty in successful forest conservationefforts even rises (Hayes and Ostrom, 2005; Schwartzman et al.,2000). Moreover, Ebeling and Yasue (2008) stress the governancechallenge for the successful implementation of avoided defor-estation programs in tropical countries. Even with hypotheticaladequate conservation funding, the conservation of undisturbednatural forest will put additional pressure on other ecosys-tem types, such as savannahs or wetlands (Miles and Kapos,2008) which may result in rising carbon emissions. Neverthe-less, the comparative analysis and prioritization of non-forestecosystem types for conservation in an optimization frameworkwould add substantial complexity and is beyond the scope of thispaper.

Agricultural expansion due to improved accessibility of forestsmay trigger further economic development which raises the costof conserving the remaining forest. On top, policies to promotebioenergy crops such as sugar cane in Brazil (Koplow, 2006) sub-stantially add to average opportunity costs of forest conservation.So, if derived demand for land in other sectors is taken into account,opportunity costs will rise significantly. Historical drivers of defor-estation include not only agricultural land expansion, but alsocommercial logging, mining or bush meet hunting (Bryant et al.,1997; Wilkie et al., 2001). These arguments are not covered inthis article and leave room for improvement in the presentedmodeling approach. Important aspects, like enforcement of conser-

vation status, administration costs, timber revenues from clearedforest, additional demand for forest land from other sectors aswell as a refinement of land expansion costs are left for futureresearch.
Page 6: Conservation of undisturbed natural forests and economic impacts on agriculture

se Pol

C

ufsTcap

aguifioa

StstttbA

mwtalals

A

F0CmitBI

Ae

uiito2Whiaad

M. Krause et al. / Land U

onclusions

We have presented an expanded version of the global landse model MAgPIE, in order to analyze the economic impacts oforest conservation strategies on agriculture. The approach pre-ented here has several advantages compared to other studies.he focus of our analysis is on undisturbed natural forests, itovers relocation of deforestation into other natural forest areas,nd it allows for endogenous technological change in agriculturalroduction.

The synthesis of strong baseline deforestation, projected leak-ge, and the historical deforestation trend leads us to the followingeneral conclusion. Deforestation continues, (1) if undisturbed nat-ral forests are not considered for conservation programs at all, (2)

f payments to stakeholders based on opportunity costs are insuf-cient to serve as an incentive for avoided deforestation, or (3) ifther factors such as the monitoring of leakage are not taken intoccount.

In particular, undisturbed natural forest conservation inub-Saharan Africa requires substantial investments in agricul-ural productivity increase to meet rising food demand, whileubstantial relocation of deforestation still occurs. Based onhe historical drivers of deforestation and factors of uncer-ainty not accounted for, we conclude that full implementa-ion of a comprehensive forest conservation program solelyased on opportunity costs appears unlikely in Sub-Saharanfrica.

In Latin America, a forest conservation strategy with a priority onaximum carbon storage results in zero opportunity costs. Here, ain-win situation could be created, where carbon emission reduc-

ions may be obtained with very low compensation payments togriculture. Relatively small annual opportunity costs in all regionsead to the conclusion that, even if agricultural R&D expendituresre taken into account, undisturbed natural forest conservation is aow-cost option to reduce emissions and maintain other ecosystemervices.

cknowledgements

We gratefully acknowledge financial support by the Germanederal Ministry of Education and Research (GLUES project, grant1LL0901A) and the European Union funded VOLANTE project (FP7ollaborative Project, grant 265104). We wish to thank the anony-ous reviewers for their constructive remarks which substantially

mproved the paper. We are grateful for valuable comments onhis paper by Christoph Mueller, Christoph Schmitz, Benjaminodirsky, and Isabelle Weindl at Potsdam Institute for Climate

mpact Research.

ppendix A. Elaboration of consistent land pool database,mployed datasets and assumptions

One of the major challenges in this study has been thenambiguous definition of spatially explicit initial land pools

n a consistent way. A hierarchical nested structure is assumedn data integration with land use classes (Erb et al., 2007) athe first level, suitable land (Fischer et al., 2002) at the sec-nd level, intact and frontier forest (Greenpeace International,005; Bryant et al., 1997) at the third and IUCN areas (UNEP-CMC, 2007) at the fourth level. Third and fourth level datasets

ave been selected to include land worth being conserved

n addition to already protected land for nature conservationnd assumed high opportunity costs of cropland expansionnd non-convertibility by political consensus. The integratedatasets constitute land modules that are deliberately combined

icy 30 (2013) 344– 354 349

in three overarching scenario groups to come up with poten-tially available land pools for agricultural land expansion(Fig. A.1).

Data has been integrated by taking raster-based land usedatasets from Erb et al. (2007) at 5 arc min resolution as startingpoint. Land suitability, intact forest, frontier forest and protectedarea datasets have been converted to raster data in the same pro-jection and enter as Boolean data at 5 arc min resolution. Thus,fractions of land use with particular land suitability have beenobtained in each grid cell. The problem of missing values has beenovercome by assigning values from neighboring cells based on theEuclidean distance approach. The union of intact and frontier for-est has been integrated in existing land use categories that areplausible to comprise forest cover. The categories of unused andforestry land use potentially incorporate intact and frontier forestcover whereas allocation priority has been given to unused landby definition of large intact forest landscapes and frontier forest(Greenpeace International, 2005; Bryant et al., 1997). Intact andfrontier forest on unused land (as defined by Erb et al., 2007) isdenoted as “undisturbed natural forest”. Residual intact and fron-tier forest cover has been allocated to forestry land use and is alsodefined as undisturbed natural forest. But, it is more prone to defor-estation than intact and frontier forest in wilderness because ofits accessibility in the vicinity of infrastructure, rivers and shores(Erb et al., 2007). The complementary subset in forestry land use isdefined as the sum of “managed forest” and “potentially managednatural forest” which covers age-class forests like forest plantationsand semi-natural forests. Finally, non-matching intact and frontierforest shares in forestry land use have been capped. Addition-ally, protected areas by IUCN have been proportionally allocatedto managed forest, potentially managed natural forest and undis-turbed natural forest in forestry land use, undisturbed natural foreston unused land and grazing land categories. Owing to the non-presence of nature reserves, wilderness area and national parks incropland and urban land categories, they have not been consideredin IUCN area allocation. In order to avoid redundant and spuriousways of integration, the strictest terrestrial conservation categoriesI and II are assumed to be covered only. The grazing suitabilitycategories (Erb et al., 2007) and crop suitability indicator values(Fischer et al., 2002; v. Velthuizen et al., 2007) have been separatelyaggregated by sum from 5 arc min resolution to 0.5 arc degree res-olution. Then, the crop suitability share of the 0.5 arc degree gridcell has been proportionally allocated to all land use types includ-ing each grazing suitability class. The non-redundant delineationof grazing land, i.e. permanent pasture and rangelands, includ-ing grazing suitability categories from other land use types likeforestry and unused land has not been violated by data integrationand aggregation and is thoroughly explained in Erb et al. (2007).Grazing land and unused land, with the latter comprising othernatural vegetation, are delineated by the distance to infrastruc-ture, shores and rivers and the net primary productivity of naturalvegetation (Erb et al., 2007). Erb et al. (2007) treat grazing land(permanent pasture and rangelands) as a residual land use cat-egory after land has been allocated to other land uses already.At 0.5 arc degree resolution, the consistent cropland dataset hasbeen substituted by a cropland dataset produced by Fader et al.(2010) for the sake of smoothness with historical time-series data.It comprises rainfed and irrigated areas for 13 crop functional types(cfts) and constitutes a synthesis of previous mapping approaches(Portmann et al., 2008; Ramankutty et al., 2008). The proportionalallocation of residual land area to remaining land use categorieshas finalized data harmonization. The output consists of global

datasets showing land use fractions at 0.5 arc degree resolution,which is about 50 km × 50 km grid cell size at the equator. Fig. A.2shows the global distribution of intact and frontier forest and totalforest.
Page 7: Conservation of undisturbed natural forests and economic impacts on agriculture

350 M. Krause et al. / Land Use Policy 30 (2013) 344– 354

io gro

fwtb7m

Fig. A.1. Land modules and scenar

The union of intact and frontier forest, i.e. undisturbed naturalorest, comprises about 1.64 billion ha globally and defines forestorth being conserved in a rather conservative way compared

o 1.34 billion ha primary forest calculated by FAO (2006), whichy definition includes undisturbed natural forests. However, only34 million ha are suitable for crop production. Table A.1 shows theagnitude of all initialized land pools.

Fig. A.2. Global distribution of intact an

ups to define available land pools.

Table A.2 grants an overview on employed datasets.

Appendix B. Mathematical description

The list of indices, parameters and variables is provided inTable B.1.

d frontier forest and total forest.

Page 8: Conservation of undisturbed natural forests and economic impacts on agriculture

M. Krause et al. / Land Use Policy 30 (2013) 344– 354 351

Table A.1Initialized land pools in 1995 (million ha).

Economic region Cropland % of total Pool of available land

IFF % of total Other land % of total

AFR 192 7.9 108 4.5 21 0.9LAM 153 7.6 487 24.2 25 1.2PAS 80 22.5 7ROW 1021 12.7 6World 1446 11.2 73

F2

ap

k

+10∑i

2324∑j

2∑m

Yi,j,m,“crop” × lcc +10∑i=1

YLDTCi × TCCi

+10∑2324∑

Zi,j ×{

ayldi,j × ̨ if ̌ = TRUE−carb × ̨ if ̌ = FALSE

(B.3)

TE

ig. C.1. Avoided deforestation and avoided net carbon emissions, FC100 versus BAU,015–2055.

During optimization the land constraint of land types, i.e. cropnd non-cropland, in each grid cell is binding for the sum of crop

roduction and conversion activities (Eqs. (B.1) and (B.2)).

21∑cr=1

Xi,j,kcr ≤ ai,j,“crop” +2∑

m=1

Yi,j,m,“crop” (B.1)

able A.2mployed geographic datasets.

Type of dataset Name of dataset Year Spatial resolu-tion/Projection/Cover

Land use Land-use dataset for the year 2000

2000 5 arc min res., geograpprojection,90◦/−90◦ lat, −180◦/1lon

Land suitability Suitability of globalland area for rainfedcrops, using max. cropand tech. Mix

2002/2005 5 arc min res., geograpprojection,90◦/−90◦ lat, −180◦/1lon

Protected areas Protected AreasNational

2004 Polygons, geographicprojection90◦/−90◦ lat, −180◦/1lon

Intact forest World intact forestlandscapes map

2005 Polygons, geographicprojection, 69◦/−55◦

−172◦/178◦ lonFrontier forest The Last Frontier

Forests1997 Polygons,

pseudo-cylindricalequal-area projection7984568m/−6417752−10138882/1531610

Land use Rainfed and irrigatedcropland and managedgrassland

1700–2005 30 arc sec res., 90◦/−9lat, −180◦/180◦ lon

Land cover The Global Land CoverMap for the Year 2000

2000 32.1 arc sec. res.,geographic projection89.991071◦/56.00892−180◦/179.991070◦ l

7 21.7 2 0.62 0.8 74 0.94 5.7 122 0.9

Yi,j,“oavl”,“crop” ≤ ai,j,“oavl” (B.2)

In addition, we implement land allocation between agriculturalactivities and an undisturbed natural forest conservation activity.First, undisturbed natural forest conservation takes place until aprescribed policy-induced demand for conserved undisturbed nat-ural forest is met under the assumption of homogeneous provisionof ecosystem services per grid cell. Second, the optimization of agri-cultural production costs is conducted. We modified the objectivefunction from Lotze-Campen et al. (2008) accordingly (Eq. (B.3)).

Ct =10∑i=1

2324∑j=1

31∑k=1

Xi,j,k × FCi,j,kve + Xi,j,k × YLDi,j,kve × bi,j × tckve

i ji,j

ageCat. used Institution Reference

hic

80◦

All Institute of SocialEcology,KlagenfurtUniversity

Erb et al. (2007)

hic

80◦

SI0, SI5, SI40, SI85 FAO/IIASA Fischer et al. (2002)and v. Velthuizenet al. (2007)

80◦

Cat. I&II UNEP-WCMC UNEP-WCMC(2007)

lat,All Greenpeace

InternationalGreenpeaceInternational(2005)

,,

0m

All WRI Bryant et al. (1997)

0◦ Cropland 2000 Potsdam-Institutefor Climate ImpactResearch

Fader et al. (2010)

,8◦ lat,on

Cat. 37 (Snow) EuropeanCommission JointResearch Centre

EuropeanCommission (2003)

Page 9: Conservation of undisturbed natural forests and economic impacts on agriculture

352 M. Krause et al. / Land Use Policy 30 (2013) 344– 354

Table B.1List of indices, parameters and variables.

Statement Definition

IndicesI RegionsJ Grid cellsK Production activitiesKve Vegetal production activitiesKcr Crop activitiesM Available land pools {oavl, iff}T Time stepsParametersA Reference area from previous time stepB Transport distance to marketC Predefined global area of conserved IFFD Share of regional IFF at global IFFR Discount rateTc Unit transport costsLcc Unit land conversion costsAyld Average obtainable yield of crop typesCarb Carbon content in natural vegetationAlpha Scaling parameterBeta Scenario selection parameterSce Scenarios {bau, iff cons}VariablesX Optimized production areaY Converted land area to croplandYLD Obtainable yield of crop typesYLDTC Yield increasing technological changeZ IFF conservation areaAD Total avoided deforestation areaC Total costs of productionFC Unit factor costsTCC Unit costs of technological changeOC Total opportunity costs

(cisvc

va

Z

TD

PVOC Present value of total opportunity costsAOC Annuity of total opportunity costs

The global undisturbed natural forest conservation constraintEq. (B.4)) ensures that land allocated to undisturbed natural forestonservation equals the prescribed scenario-depending magnituden conservation. Furthermore, the prescribed regional share of con-erved undisturbed natural forest (Eq. (B.5)) aims at covering theariety of tropical undisturbed natural forest ecosystems and asso-iated ecosystem services in conservation.

10

i=1

2324∑j=1

Zi,j = c (B.4)

10

i=1

2324∑j=1

Zi,j ≥ c ∗ di (B.5)

A cellular constraint keeps the demand for land for land con-

ersion activities and undisturbed natural forest conservationctivities below the supply of potentially available land (Eq. (B.6)).

i,j + Yi,j,“iff ”,“crop” ≤ ai,j,“iff ” (B.6)

able C.1isaggregation of total opportunity costs by cost types, 2015–2055 (million US$ per year

Economic region Factor costs (1) R&D costs (2) Land c

FC 100 FC 50-Y FC 50-C FC 100 FC 50-Y FC 50-C FC 100

AFR 0.00 0.00 0.07 0.88 0.12 0.20 −0.28

LAM −0.07 −0.04 0.01 0.66 0.34 0.08 −0.27

PAS −0.04 −0.01 0.04 0.71 0.11 0.08 −0.21

ROW −0.03 −0.01 0.01 0.65 0.07 0.14 −0.15

World −0.14 −0.07 0.13 2.90 0.64 0.50 −0.91

Fig. C.2. Average annual opportunity costs, FC100 versus BAU, 2015–2055.

Avoided deforestation is result of shifts in patterns of land usealong the land productivity gradient taking expected transportcosts of produced commodities into account. It is calculated as thedifference in BAU deforestation to scenario-based deforestation (Eq.(B.7)).

AD =2055∑

t=2015

10∑i=1

2324∑j=1

Yt,i,j,“iff ”,“crop” − Yt,i,j,“iff ”,“crop”

for sce{

bau, iff cons}

(B.7)

The total opportunity costs at present value (Eq. (B.8)) and asannuity for the time horizon of undisturbed natural forest conser-vation programs (Eq. (B.9)) are calculated as follows.

PVOCi =5∑

t=1

(Ct,i,“iff cons” − Ct,i,“bau”)

× 1

(1 + r)(t×10)for sce

{bau, iff cons

}(B.8)

AOCi = PVOCi × r

1 − (1 + r)−50(B.9)

Appendix C. Supplementary results

See Tables C.1–C.3 and Figs. C.1 and C.2.

).

onversion costs (3) Transport costs (4) Total

4∑i=1

i

FC 50-Y FC 50-C FC 100 FC 50-Y FC 50-C FC 100 FC 50-Y FC 50-C

−0.05 −0.09 −0.04 0.03 −0.03 0.55 0.10 0.15−0.16 −0.04 0.00 −0.06 −0.04 0.33 0.07 0.00−0.04 −0.03 0.00 −0.01 −0.03 0.46 0.05 0.05−0.02 −0.03 −0.13 0.00 −0.02 0.33 0.04 0.09−0.26 −0.20 −0.18 −0.04 −0.13 1.67 0.26 0.30

Page 10: Conservation of undisturbed natural forests and economic impacts on agriculture

M. Krause et al. / Land Use Policy 30 (2013) 344– 354 353

Table C.2Projected annual change in land pools (million ha yr−1), 2015–2055.

Economic region Time period Cropland Available IFF Conserved IFF Other available land

BAU FC 100 FC 50-Y FC 50-C BAU FC 100 FC 50-Y FC 50-C BAU FC 100 FC 50-Y FC 50-C BAU FC 100 FC 50-Y FC 50-C

AFR 2015 3.5 2.5 2.5 2.5 −1.0 −7.5 −0.7 −0.7 Nil 7.5 0.7 0.7 −2.5 −2.5 −2.5 −2.52055 2.8 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Nil 0.0 0.0 0.0 −2.8 0.0 0.0 0.0

Mean 3.1 1.6 2.7 2.6 −1.5 −1.5 −1.5 −1.5 Nil 1.5 0.4 0.5 −1.6 −1.6 −1.6 −1.6

LAM 2015 0.9 0.1 0.3 1.7 −0.9 −46.9 −4.9 −6.3 Nil 46.9 4.7 4.7 −0.1 −0.1 −0.1 −0.12055 1.6 0.0 1.5 1.6 0.0 0.0 −6.0 −5.8 Nil 0.0 6.0 5.8 −1.6 0.0 −1.5 −1.6

Mean 1.6 0.6 1.2 1.5 −1.1 −9.4 −6.0 −6.2 Nil 9.4 5.3 5.2 −0.5 −0.6 −0.5 −0.5

PAS 2015 0.7 0.1 0.4 0.4 −0.6 −6.4 −0.9 −0.9 Nil 6.4 0.6 0.6 −0.1 −0.1 −0.1 −0.12055 1.2 0.0 0.0 1.0 −1.2 0.0 −0.3 −1.1 Nil 0.0 0.3 0.4 0.0 0.0 0.0 −0.2

Mean 0.9 0.1 0.7 0.8 −0.9 −1.3 −1.1 −1.2 Nil 1.3 0.5 0.5 0.0 −0.1 0.0 −0.1

ROW 2015 2.5 1.8 2.5 2.4 −0.7 −3.3 −1.0 −0.9 Nil 3.3 0.3 0.3 −1.8 −1.8 −1.8 −1.82055 1.0 0.3 0.2 0.7 −0.7 0.0 −0.1 −0.6 Nil 0.0 0.1 0.2 −0.3 −0.3 −0.2 −0.3

Mean 1.1 0.6 1.0 1.0 −0.5 −0.7 −0.6 −0.6 Nil 0.7 0.2 0.3 −0.6 −0.6 −0.6 −0.6

World 2015 7.6 4.6 5.7 7.0 −3.1 −64.1 −7.6 −8.9 Nil 64.1 6.4 6.4 −4.5 −4.6 −4.5 −4.52055 6.6 0.3 1.7 3.2 −1.8 0.0 −6.4 −7.5 Nil 0.0 6.4 6.4 −4.7 −0.3 −1.7 −2.1

Mean 6.7 2.9 5.5 5.9 −4.0 −12.8 −9.2 −9.5 Nil 12.8 6.4 6.4 −2.7 −2.9 −2.7 −2.8

Table C.3Projected annual change in crop yield level due to R&D input (% yr−1), 2015–2055.

Economic region Time period Scenario

BAU FC100 FC50-Y FC50-C

AFR 2015 1.0 1.5 1.5 1.52055 0.8 1.8 1.6 1.3

Mean 1.2 1.8 1.3 1.4

LAM 2015 1.4 1.8 1.8 1.02055 0.4 1.0 0.4 0.4

Mean 0.7 1.3 1.0 0.8

PAS 2015 1.1 1.7 1.4 1.42055 0.1 1.1 1.0 0.3

Mean 0.7 1.5 0.9 0.9

ROW 2015 1.4 1.4 1.4 1.42055 0.6 0.6 0.6 0.6

Mean 1.1 1.1 1.1 1.1

World 2015 1.3 1.5 1.4 1.3

R

A

B

B

B

2055 0.5

Mean 1.0

eferences

ndam, K.S., Ferraro, P.J., Pfaff, A., Sanchez-Azofeifa, G.A., Robalino, J.A., 2008. Mea-suring the effectiveness of protected area networks in reducing deforestation.Proceedings of the National Academy of Science of the United States of America105 (42), 16089–16094.

arlow, J., Gardner, T.A., Araujo, I.S., Avila-Pires, T.C., Bonaldo, A.B., Costa, J.E., Espos-ito, M.C., Ferreira, L.V., Hawes, J., Hernandez, M.I.M., Hoogmoed, M.S., Leite, R.N.,Lo-Man-Hung, N.F., Malcom, J.R., Martins, M.B., Mestre, L.A.M., Miranda-Santos,R., Nunes-Gutjahr, A.L., Overal, W.L., Parry, L., Peters, S.L., Ribeiro-Junior, M.A.,da Silva, M.N.F., da Silva Motta, C., Peres, C.A., 2007. Quantifying the biodiver-sity value of tropical primary, secondary and plantation forests. Proceedingsof the National Academy of Science of the United States of America 104 (47),18555–18560.

onan, G.B., 2008. Forests and climate change: forcings, feedbacks, and the climate

benefits of forests. Science 320, 1444–1449.

ondeau, A., Smith, P., Zaehle, S., Schaphoff, S., Lucht, W., Cramer, W., Gerten, D.,Lotze-Campen, H., Mueller, C., Reichstein, M., Smith, B., 2007. Modelling therole of agriculture for the 20th century global terrestrial carbon balance. GlobalChange Biology 13, 679–706.

0.9 0.7 0.6

1.3 1.1 1.1

Boserup, E., 1965. The Conditions of Agricultural Growth: The Economics of Agrar-ian Change Under Population Pressure. George Allen & Unwin Ltd., London,124 pp.

Brooks, T.M., Mittermeier, R.A., da Fonseca, G.A.B., Gerlach, J., Hoffmann, M.,Lamoreux, J.F., Mittermeier, C.G., Pilgrim, J.D., Rodrigues, A.S.L., 2006. Globalbiodiversity conservation priorities. Science 313, 58–61.

Bryant, D., Nielsen, D., Tangley, L., 1997. The Last Frontier Forests – Ecosystems &Economies on the Edge. World Resources Institute (WRI), Washington, 49 pp.

Butler, R.A., Laurance, W.F., 2008. New strategies for conserving tropical forests.Trends in Ecology & Evolution 23 (9), 469–472.

Dietrich, J.P., Schmitz, C., Mueller, C., Fader, M., Lotze-Campen, M., Popp, A., 2010a.Measuring agricultural land-use intensity. In: Paper Presented at HAWEPA2010 Workshop, Halle, Germany, June 28–29, 2010, http://www.iamo.de/fileadmin/veranstaltungen/hawepa10/Dietrich et.al. Hawepa 2010.pdf,12th January 2011.

Dietrich, J.P., Schmitz, C., Lotze-Campen, H., Mueller, C., Popp, A., 2010b. Imple-menting endogenous technological change in a global land use model. In: Paper

Presented at the 13th Annual Global Trade Analysis Project (GTAP) Conference,Penang, Malysia, June 9–11, 2010, GTAP Resource #3283, 24 pp., 12th January2011.
Page 11: Conservation of undisturbed natural forests and economic impacts on agriculture

3 se Pol

E

E

E

E

F

F

F

F

G

G

G

G

G

H

J

K

K

K

L

L

L

M

v. Wieser, F., 1928. Social Economics. Reprinted in 2003. Routledge Library Editions,

54 M. Krause et al. / Land U

beling, J., Yasue, M., 2008. Generating carbon finance through avoided deforesta-tion and its potential to create climatic, conservation and human developmentbenefits. Philosophical Transactions of the Royal Society B 363, 1917–1924.

C (European Commission), 2003. Global Land Cover 2000 Database.Joint Research Centre, Brussels, Belgium, http://bioval.jrc.ec.europa.eu/products/glc2000/glc2000.php, 20th April 2009.

liash, J., 2008. Climate Change: Financing Global Forests – The Eliash Review. TheStationary Office, London, 250 pp.

rb, K.-H., Gaube, V., Krausmann, F., Plutzar, C., Bondeau, A., Haberl, H., 2007. A com-prehensive global 5 min resolution land-use data set for the year 2000 consistentwith national census data. Journal of Land Use Science 2 (3), 191–224.

ader, M., Rost, S., Mueller, C., Bondeau, A., Gerten, D., 2010. Virtual water contentof temperate cereals and maize: present and potential future patterns. Journalof Hydrology 384 (3–4), 218–231.

AO (Food and Agriculture Organization of the United Nations), 2002. WorldAgriculture: Towards 2015/2030. Summary report. FAO, Rome, Italy.http://www.fao.org/docrep/004/Y3557E/\-Y3557E00.HTM, 20th November2008.

AO, 2006. Global Forest Resources Assessment 2005. FAO Forestry Paper 147.FAO, Rome, Italy. http://www.fao.org/docrep/008/a0400e/a0400e00.htm, 15thSeptember 2010.

ischer, G., Shah, M., v. Velthuizen, H., Nachtergaele, F.O., 2002. Global Agro-ecological Assessment for Agriculture in the 21st Century. InternationalInstitute for Applied Systems Analysis (IIASA), FAO. http://www.iiasa.ac.at/Research/LUC/Papers/gaea.pdf, 14th June 2008.

eist, H.J., 2006. Multiple impacts of land-use/cover change. In: Lambin, E.F., Geist, H.(Eds.), Land-Use and Land-Cover Change: Local Processes and Global Impacts.Global Change – International Geosphere-Biosphere Programme Book Series.Springer, Berlin, pp. 71–116.

reenpeace International, 2005. The World’s Last Intact Forest Landscapes.www.intactforest.org/download/download.htm, 15th September 2010.

rig-Gran, M., 2006. The cost of avoiding deforestation. Report Prepared for the SternReview of the Economics of Climate Change. International Institute for Environ-ment and Development (IIED), London, 20 pp. http://pubs.iied.org/G02489.html,15th September 2010.

ullison, R.E., Frumhoff, P.C., Canadell, J.G., Field, C.B., Nepstad, D.C., Hayhoe, K.,Avissar, R., Curran, L.M., Friedlingstein, P., Jones, C.D., Nobre, C., 2007. Tropicalforests and climate policy. Science 316, 985–986.

umpenberger, M., Vohland, K., Heyder, U., Poulter, B., Macey, K., Rammig, A., Popp,A., Cramer, W., 2010. Predicting pan-tropical climate change induced foreststock gains and losses—implications for REDD. Environmental Research Letters5, 014013, 15 pp.

ayes, T., Ostrom, E., 2005. Conserving the world’s forests: Are protected areas theonly way? Indiana Law Review 38, 595–617.

ackson, R.B., Randerson, J.T., Canadell, J.G., Anderson, R.G., Avissar, R., Baldocchi,D.D., Bonan, G.B., Caldeira, K., Diffenbaugh, N.S., Field, C.B., Hungate, B.A., Job-bagy, E.G., Kueppers, L.M., Nosetto, M.D., Pataki, D.E., 2008. Protecting climatewith forests. Environmental Research Letters 3, 044006, 5 pp.

indermann, G., Obersteiner, M., Sohngen, B., Sathaye, J., Andrasko, K., Rametsteiner,E., Schlamadinger, B., Wunder, S., Beach, R., 2008. Global cost estimates of reduc-ing carbon emissions through avoided deforestation. Proceedings of the NationalAcademy of Science of the United States of America 105 (30), 10302–10307.

oplow, D., 2006. Biofuels at What Cost? Government Support for Ethanoland Biodiesel in the United States. International Institute for SustainableDevelopment, Global Subsidies Initiative. Geneva, Switzerland. http://www.globalsubsidies.org/files/assets/Brochure - US Report.pdf, 04th March 2011.

rause, M., Lotze-Campen, H., Popp, A., 2009. Spatially-explicit scenarios on globalcropland expansion and available forest land in an integrated modelling frame-work. Selected Reviewed Paper Presented at the 27th International Associationof Agricultural Economists Conference in Beijing, China, August 16–22, 2009.International Association of Agricultural Economists (IAAE), Milwaukee, USA.AgEcon, 22 pp. http://purl.umn.edu/51751, 15th September 2010.

aurance, W.F., 2007. Have we overstated the tropical biodiversity crisis? Trends inEcology and Evolution 22 (2), 2265–2270.

otze-Campen, H., Mueller, C., Bondeau, A., Rost, S., Popp, A., Lucht, W., 2008.Global food demand, productivity growth and the scarcity of land and waterresources: a spatially explicit mathematical programming approach. Agricul-tural Economics 39 (3), 325–338.

otze-Campen, H., Popp, A., Beringer, T., Mueller, C., Bondeau, A., Rost, S., Lucht, W.,2010. Scenarios of global bioenergy production: the trade-offs between agricul-tural expansion, intensification and trade. Ecological Modelling 221, 2188–2196.

iles, L., Kapos, V., 2008. Reducing greenhouse gas emissions from deforestationand forest degradation: global land-use implications. Science 320, 1454–1455.

icy 30 (2013) 344– 354

Mittermeier, R.A., Mittermeier, C.G., Brooks, T.M., Pilgrim, J.D., Konstant, W.R., daFonseca, G.A.B., Kormos, C., 2003. Wilderness and biodiversity conservation. Pro-ceedings of the National Academy of Science of the United States of America 100(18), 10309–10313.

Narayanan, B., Walmsley, T.L., 2008. Global Trade, Assistance, and Production: TheGTAP 7 Data Base. Center for Global Trade Analysis, Purdue University, WestLafayette.

Nelson, A., 2008. Estimated travel time to the nearest city of 50,000 or morepeople in year 2000. Global Environment Monitoring Unit – Joint ResearchCentre of the European Commission, Ispra Italy. http://bioval.jrc.ec.europa.eu/products/gam/download.htm, 12th December 2009.

Popp, A., Lotze-Campen, H., Bodirsky, B., 2010. Food consumption, diet shifts andassociated non-CO2 greenhouse gas emissions from agricultural production.Global Environmental Change 20, 451–462.

Popp, A., Dietrich, J.P., Lotze-Campen, H., Klein, D., Bauer, N., Krause, M., Beringer,T., Gerten, D., Edenhofer, O., 2011a. The economic potential of bioenergy forclimate change mitigation with special attention given to implications for theland system. Environmental Research Letters 6, 034017.

Popp, A., Lotze-Campen, H., Leimbach, M., Knopf, B., Beringer, T., Bauer, N.,Bodirsky, B., 2011b. On sustainability of bio-energy production: integrat-ing co-emissions from agricultural intensification. Biomass & Bioenergy 35,4770–4780.

Portmann, F., Siebert, S., Bauer, C., Döll, P., 2008. Global data set ofmonthly growing areas of 26 irrigated crops. Frankfurt Hydrology Paper06, Institute of Physical Geography, University of Frankfurt, Frankfurtam Main, Germany, 400 pp, http://www.geo.uni-frankfurt.de/ipg/ag/dl/fpublikationen/2008/FHP 06 Portmann et al 2008.pdf, 15th September 2010.

Ramankutty, N., Evan, A.T., Monfreda, C., Foley, J.A., 2008. Farming the planet. 1.Geographic distribution of global agricultural lands in the year 2000. GlobalBiogeochemical Cycles 22, GB1003.

RFA (Renewable Fuels Agency), 2008. The Gallagher Review of the indirect effectsof biofuels production. St Leonards-on-Sea. http://www.renewablefuelsagency.gov.uk/reportsandpublications/reviewoftheindirecteffectsofbiofuels, 14th June2010.

Roberts, D., White, A., Nilsson, S., 2008. Convergence of food, fuel and fibre markets:driving change in the world’s forests. IUCN Arborvitae 37, 8–9.

Sitch, S., Smith, B., Prentice, C., Arneth, A., Bondeau, A., Cramer, W., Kaplans, J.O.,Levis, S., Lucht, W., Sykes, M.T., Thonicke, K., Venevsky, S., 2003. Evaluation ofecosystem dynamics, plant geography and terrestrial carbon cycling in the LPJdynamic global vegetation model. Global Change Biology 9, 161–185.

Schmitt, C.B., Burgess, N.D., Coad, L., Belokurov, A., Besancon, C., Boisrobert, L., Camp-bell, A., Fish, L., Gliddon, D., Humphries, K., Kapos, V., Loucks, C., Lysenko, I., Miles,L., Mills, C., Minnemeyer, S., Pistorius, T., Ravilious, C., Steininger, M., Winkel, G.,2009. Global analysis of the protection status of the world’s forests. BiologicalConservation 142, 2122–2130.

Schmitz, C., Biewald, A., Lotze-Campen, H., Popp, A., Dietrich, J.P., Bodirsky, B., Krause,M., Weindl, I., 2012. Trading more food – implications for land use, green-house gas emissions, and the food system. Global Environmental Change 22(1), 189–209.

Schwartzman, S., Moreira, A., Nepstad, D., 2000. Rethinking tropical forest conser-vation: perils in parks. Conservation Biology 14 (5), 1351–1357.

Sohngen, B., Tennity, C., Hnytka, M., 2009. Global forestry data for the economicmodeling of land use. In: Hertel, T.W., Rose, S., Tol, R.S.J. (Eds.), Economic Analysisof and Use in Global Climate Change Policy. Routledge, New York, pp. 49–71.

Turner, I.M., 1996. Species loss in fragments of tropical rain forest: a review of theevidence. Journal of Applied Ecology 33, 200–209.

UN (United Nations), 2009. Revision of World Urbanization Prospects. UN PopulationDivision. http://esa.un.org/unpd/wup/index.htm, 04th March 2011.

UNEP-WCMC (United Nations Environment Programme – World Conservation Mon-itoring Centre), 2007. World Database on Protected Areas (WDPA). CD-ROM,Cambridge, UK. http://sea.unep-wcmc.org/wdbpa/, 14th June 2008.

v. Velthuizen, H., Huddleston, B., Fischer, G., Salvatore, M., Ataman, E., Nachter-gaele, F.O., Zanetti, M., Bloise, M., Antonicelli, A., Bel, J., De Liddo, A., De Salvo, P.,Franceschini, G., 2007. Mapping biophysical factors that influence agriculturalproduction and rural vulnerability. Environment and Natural Resources Series11. FAO, Rome. http://www.fao.org/docrep/010/\-a1075e/a1075e00.htm, 14thJune 2008.

Routledge, London, Great Britain.Wilkie, D.S., Carpenter, J.F., Zhang, Q., 2001. The under-financing of protected areas

in the Congo Basin: so many parks and so little willingness-to-pay. Biodiversityand Conservation 10 (5), 691–709.