- Email: [email protected]
Assessing the effectiveness of payments for ecosystem services for diversifying rubber in Yunnan, China
Environmental Modelling & Software 69 (2015) 187e195
Contents lists available at ScienceDirect
Environmental Modelling & Software journal homepage: www.elsevier.com/locate/envsoft
Assessing the effectiveness of payments for ecosystem services for diversifying rubber in Yunnan, China Alex Smajgl a, e, *, Jianchu Xu b, Stephen Egan c, Zhuang-Fang Yi b, John Ward d, Yufang Su b a
Mekong Region Futures Institute, Bangkok, Thailand World Agroforestry Center, East and Central Asia Region, Kunming 650201, China c CSIRO Ecosystem Sciences and Climate Adaptation Flagship, Highett, VIC 3190, Australia d Mekong Region Futures Institute, Vientiane, Lao PDR, Thailand e CSIRO Ecosystem Sciences, Townsville, Australia b
a r t i c l e i n f o
a b s t r a c t
Article history: Received 25 February 2015 Received in revised form 19 March 2015 Accepted 23 March 2015 Available online
Monoculture rubber plantations are rapidly expanding throughout Southeast Asia. The Prefecture of Xishuangbanna is characterised as both an epicenter of Chinese biodiversity and rubber production. The rapid conversion of native primary forest into rubber plantations has introduced tensions between the competing political goals to conserve biodiversity and heritage landscapes e both driving the rapidly emerging tourism industry e and economic aspirations at household and district levels. In China, decision makers discuss payments for ecosystem service schemes to resolve these tensions. As a component of the policy development process, this research project was invited to inform the political debate. Agent-based simulations revealed perverse outcomes of payments for ecosystem services, intended to encourage the conversion of monoculture rubber into agroforestry rubber. As an outcome of this modelling-based initiative, managing agencies have revised previously drafted payment schemes and reconsidered the importance of monitoring and regulatory approaches. Crown Copyright © 2015 Published by Elsevier Ltd. All rights reserved.
Keywords: Agent-based modelling Payment for ecosystem services Conservation Rubber Monoculture China
1. Introduction Monoculture rubber (Hevea brasiliensis) plantations are expanding rapidly throughout Xishuangbanna Prefecture, Southwest China, largely replacing evergreen broadleaf trees and swidden-fallow secondary vegetation, moving toward marginal land in higher altitudes and steeper slopes. More than 424,000 ha of forested areas with a high biodiversity value have already been converted to rubber planations. Continued expansion has the potential to encroach into existing primary forest and protected areas (Xu et al., 2013). There are costs and benefits associated with the land use and land cover changes, providing higher income for local farmers, increased integration into global economies, but introducing substantial and enduring threats to terrestrial biodiversity, impaired hydrological services, and reduced total carbon biomass (Li et al., 2008; Yi et al., 2014; Ziegler et al., 2009). Local and Provincial agencies and the central government in Beijing have
* Corresponding author. CSIRO Ecosystem Sciences, Townsville, Australia. Tel.: þ66 9956571200. E-mail address: [email protected] (A. Smajgl). http://dx.doi.org/10.1016/j.envsoft.2015.03.014 1364-8152/Crown Copyright © 2015 Published by Elsevier Ltd. All rights reserved.
increasingly expressed concerns regarding the environmental consequences of unregulated rubber expansion. Yunnan is an inland province at a low latitude and high elevation, characterized by the richest biological and cultural diversity in China and connects two global biodiversity hotspots, the Mountains of Southwest China and Indo-Burma. Yunnan is the source of the headwaters and major tributaries of the Yangtze, Salween, Irrawaddy, Mekong, Red, and Pearl Rivers, which combined directly or indirectly influence the lives of more than 600 million people. Substantial private investments in land development including rubber plantation are coinciding with centrally planned increases in the economic development of the region, focused primarily on of the construction of transportation infrastructure and hydropower dams (Grumbine and Xu, 2011). Concurrent with economic development is a rapid deterioration of the ecological health of the “roof” of Southwest China and Southeast Asia. Chinese institutions, governance and policy makers play a pivotal role in resolving the tensions that arise when the imperatives of conservation and economic development coincide. Intensifying demands for arable land and the multitude of benefits derived from biodiversity conservation are likely to amplify the importance of that role. Determining whether rubber plantations
188
A. Smajgl et al. / Environmental Modelling & Software 69 (2015) 187e195
and tropical rainforest can co-exist and how the negative impacts of infrastructure development on wildlife habitats can be mitigated are two priority questions confronting Chinese policy makers. Consistent with the arguments of Stone (2008), the likelihood of a conflict between the imperatives of economic growth and habitat preservation in Yunnan is escalating. Decision makers are sending mixed signals about their intentions. The Central Government initiated two nationwide conservation policies in the late 1990s e the Natural Forest Conservation Program (NFCP) and the Grain to Green Program. The Yunnan Provincial Government advocated increased funding to expand conservation areas and statutory conservation mandates in 2007 (Stone, 2008). However, rubber is a very profitable livelihood for local people. “Getting rich” and increasing wealth are also priority mandates for local government officials particularly at the Xishuangbanna Prefecture level. Balancing the two potentially competing views of conservation and economic development is a challenge for decision makers. Restricting development is not considered a plausible solution. Recent debates among decisionmakers and research academia have identified trade-offs between commercial exploitation of the region's natural resources and the need to promote sustainability of the ecosystems in the area. In 2009, a participatory research process was invited to facilitate discussions between these levels of government and provide insights of the potential consequences of proposed incentive changes; details of the participatory process are provided in Smajgl et al. (in press). Methodologically, the learning exercise employed agent-based social simulation to address the high levels of complexity and cognitive demands the decision context entails. This paper describes the characteristics and implementation of the Mersim (Mekong region simulation) model. The following Section outlines the competing goals of economic growth and ecological conservation against the backdrop of rubber expansion in Xishuangbanna, considered in a multi-level governance environment. The model design is then explained, followed by the simulation results for the most relevant policy scenarios. The paper concludes with the analysis of simulation results and subsequent policy deliberations. 2. Rubber, biodiversity and livelihood in Xishuangbanna Rubber (H. brasiliensis) has emerged as a key cash crop replacing traditional agriculture and secondary forests in the Mekong region (Ziegler et al., 2009), a direct result of Chinese aspirations for improved national rubber security and Chinese market demands, the world's largest rubber consumer (Mann, 2009). Forecasts indicate that global demand for natural rubber may outpace supply by 1.4 million metric tons by 2020 (FAO, 2013). Asia accounts for 97% of the world's natural rubber supply, traditionally from Thailand, Indonesia, and Malaysia. Entrepreneurs from China, Vietnam and Thailand are investing in rubber plantations in the less developed countries of the Mekong RegiondLaos, Cambodia and Myanmar. Currently, China contributes approximately 6.8% of the total global production in natural dry rubber (Table 1), but consumed 37% of global production in 2011 (FAO, 2013).
Situated in the upper Mekong, the Xishuangbanna Dai Autonomous Prefecture is a biologically diverse region in the tropical zone of southwest China. The Prefecture covers only 0.2% of the land area of China, it harbours some 16% of the vascular flora, 21.7% of mammals, and 36.2% of birds found in the country (Zhang and Cao, 1995). The northern, most tropical rainforest in the world is found here below 800 msl, which is the habitat that was first converted to rubber (Zhu et al., 2004). Xishuangbanna is also culturally diverse. More than two thirds of the population belongs to one of 12 ethnic minorities including the valley-dwelling Dai, who practice rice-paddy and home garden intensive agriculture, and upland people such as Hani (or Akha), Jinuo, Yao, Lahu and Bulang, largely engaged in swidden agriculture, with a forest mosaic landscape. It is one of the most famous tourism destinations regionally. Xishuangbanna National Nature Reserve was established in 1958, one of the earliest reserves in China, recognized for its high endemic species and for the conservation of Asian elephants, which are threatened by the expansion of rubber monoculture (Chen et al., 2013, submitted for publication; Zhang et al., 2007). After the 1949 Revolution, the new government of China identified rubber as an important strategic national resource. Rubber plantations and production were promoted through centralised governmental planning and regulatory initiatives. Commencing in the 1950's, the Central Government established rubber plantations or state rubber farms in Xishuangbanna Prefecture under the direction of the Ministry of Land Reclamation of Yunnan (Xu, 2006). As a consequence of the first Forestry Reform in 1983 (legislated as the Forestry Law of the People's Republic of China 1985), state forest tenure was reassigned to collectives and smallholder rubber plantations, initiated by lowland farmers assisted by technical guidance from the state coupled with government funded farm subsidies (Yi et al., 2013). From 1976 to 2007, rainforest in Xishuangbanna decreased from 13,193 km2 to 8336 km2, while rubber expanded from 249 km2 to 2256 km2 (Li et al., 2007a). The area of rubber plantations continued to increase to 4242 km2 in 2010, almost doubling the area estimated in 2007 (Xu et al., 2013). The rapid expansion of land devoted to rubber plantations has increased the income of some, while eroding the traditional livelihoods of many others, in particular those dependent on non-timber forest products. Moreover, the remaining forest area is increasingly fragmented and continues to degrade further through area and edge effects (Li et al., 2009; Xu et al., 2013). Assessments of biodiversity in rubber monocultures remain very low and dominated by introduced species (Wu et al., 2001). As a consequence of increasing monoculture rubber, biodiversity, as measured by the number of species found in the landscape (Sodhi et al., 2004; Xu and Melick, 2007) and carbon stocks both above (Bunker et al., 2005; Li et al., 2007b) and below ground (Guo and Gifford, 2002) are predicted to decrease. Consequentially, associated livelihood options are likely to decline (Xu et al., 2005). Ziegler et al. (2009) contend that expanded rubber monocultures are likely to affect the regional hydrology, as surface and groundwater recharge rates are likely to deteriorate because rubber
Table 1 Chinese rubber consumption and global production for 2009. Chinese rubber consumption and production for 2009
Tonnes
% Of world production
Domestic supply quantity Import Quantity Local Production Global production 2009
2,418,556 1,813,667 618,866 9,737,674
25% 19% 6%
Source FAO 2013.
A. Smajgl et al. / Environmental Modelling & Software 69 (2015) 187e195
plantations require more water compared to endemic rainforest (Guardiola-Claramonte et al., 2008; Tan et al., 2011). Increased application of fertilizers and residues of products used in rubber processing have led to eutrophication of water bodies, and the combination of herbicide use and reduced natural filtering has contaminated drinking supplies (Zhou and Hu, 2008; Zhou et al., 2006). Reduced and contaminated water supplies have, in some cases, resulted in conflicts in water access between villages. In summary, the degradation of ecosystem services reduces the environment's capacity to support a diversity of livelihoods, the broader economy and threatens the long term viability of the rubber industry. Maintaining productive rubber production and secured livelihoods coupled with policy support to sustain ecosystem services and biodiversity conservation remain as complementary policy objectives in Xishuangbanna. These joint imperatives manifest as initiatives to design policy instruments and economic incentives to promote environmentally friendly rubber across Xishuangbanna. Crucially, the political consensus and pathway for implementing such policy changes already exist. Under pressure from both national and provincial governments to address the environmental problems correlated with expanded rubber plantations, the Xishuangbanna Prefectural government and the rubber industry established the “Leadership Group for Environmentally Friendly Rubber” (LGEFR) in 2009. The paper describes the results of an invitation to provide research insights to evaluate the degree that conservation goals can be achieved without compromising the economic reality of sustained agricultural livelihoods and incomes. Payments for ecosystem services (PES) were identified as a key incentive to reduce the area under rubber. The PES design offered financial compensation to rubber farmers if they replaced 20% of their monoculture rubber by native vegetation in altitudes below 900 m and completely eliminated rubber in areas above 900 m. However, in the initial design the financial compensation replaced less than 10% of the subsequent loss in rubber income. As a corollary, farmers did not participate in the PES scheme. The resulting political impetus was to negotiate for full compensation of losses incurred by rubber farmers participating in the PES scheme. This research was co-designed by decision makers in Xishuangbanna to assess the potential impact of a revised PES scheme with full compensation of rubber farmers. 3. Model design 3.1. Methodological choice The overarching participatory process determined the research focus of the study as outlined above. Agent-based social simulation allows for the representation of such complex social-ecological interactions (Gilbert, 2008; Smajgl and Bohensky, 2013) and is suitable when the modelling aims for improved system understanding or social learning and where disaggregated effects of interactions between individuals and the environment need to be analysed (Kelly (Letcher) et al., 2013). Agent-based modelling is often employed to simulate land use change by incorporating prevailing environmental conditions, bio-physical processes, a diversity of human behaviours, decision-making, and policy alternatives (Bohensky et al., 2013; Brown et al., 2004; Marshall and Smajgl, 2013; Matthews et al., 2007; Robinson et al., 2007). Over recent years, modelling Land use and Land Cover Change (LUCC) has developed into a stronghold for ABM (Janssen et al., 2008; Robinson et al., 2007; Smajgl and Bohensky, 2013; Verburg et al., 2002). In China, ABM has been limited to a few studies, focused on LUCC and livelihood changes through PES, e.g. the Grain to Green
189
Program (Chen et al., 2012; Sun and Müller, 2013). Chen et al. (2014) is especially relevant, reporting the modelled impacts of PES schemes, in the form of cash and electricity payments, on forest cover. In this modelling context, albeit not related to rubber farming, all PES schemes resulted in increased forest cover. Other agent-based studies focus on urbanization processes (Gu and Cheng, 2007; Zhang et al., 2010, 2004). An et al. (2005) developed an ABM focussed on deforestation and forest degradation in China, to simulate the effects on forests and panda habitats associated with growing rural populations. At the time of writing, there are no published studies on the expansion of Chinese rubber monocultures and the economic effects on local people. Outside of China, rubber expansion has been modelled in the LB-LUDAS model to explore the impacts of PES on the trade-offs between goods and services of jungle rubber in Jambi Province, Indonesia (Villamor et al., 2014, 2012). Given the many effective LUCC-type applications of agent-based methodology stakeholders and the research team agreed to develop the Mekong region simulation (Mersim) model. The description of the agent-based model Mersim follows the ODD (Overview e Design concepts e Details) protocol (Grimm et al., 2006, 2010; Müller et al., 2014) and model details including Java code can be found in Smajgl et al. (2013). 3.2. Purpose of the model In initial rounds of facilitated consultations in Xishuangbanna, decision makers identified payments for replacing monoculture rubber plantations by agroforestry rubber as the policy intervention that needed to be assessed (details for the participatory process can be found in Smajgl et al., in press). This type of payment for ecosystem service was to be combined with full compensation of losses incurred by participating rubber farmers. Decision makers identified the aggregate area planted to rubber as the key indicator of PES performance, which required the spatially explicit modelling of behavioural changes of households and environmental dynamics (Smajgl et al., in press); see Lieske (2015) for the relevance of spatially explicit modelling. 3.3. State variables The participatory process placed the stakeholder priorities at the core of the model design and determined the state variables as: rubber production, household income and poverty, land use and household livelihoods, household location and migration. 3.4. Emergence Following from the stakeholder-defined modelling goal, the spatial extent of rubber and the broader land use change were critical emergent phenomena, which are inherently linked to the livelihood and migration decisions implemented by households. Livelihood changes emerge in response to changes in environmental, economic and institutional factors. Land use patterns evolve as a consequence of socialeenvironmental interaction, including the area used for cultivating rubber. The perception of changes in the agent's environmental demands typically requires a cognitive structure that allows for the evaluation of options and their likely consequences. Several scholars argue for cognitively rich agents in agent-based social simulation (i.e. Edmonds and Moss, 2001; Sun and Naveh, 2004) and alternate learning algorithms have been developed and tested (Brenner, 2006; Fudenberg and Levine, 1998). In an empirical model, however, the characterization of the cognitive architecture of agents is a challenge as individuals display a diversity of responses to manifold
190
A. Smajgl et al. / Environmental Modelling & Software 69 (2015) 187e195
drivers, subject to variable motivations and expressed as different behavioural thresholds. This introduces high levels of uncertainty for assumptions regarding cognitive structure and parameterisation. Therefore, the cognitive reasoning regarding potential behavioural adaptation was elicited through a randomised survey of Xishuangbanna households, providing intentional data for parameterising largely reactive agents. Thereby, the cognitive reasoning is implicit to the intentional data elicited for all relevant scenarios and the parameterisation process. This is best explained through the characterisation and parameterisation process itself. 3.5. Household data for parameterisation The parameterisation process is described based on the framework provided in Smajgl et al. (2011). Fig. 1 shows the principle parameterisation steps required in an empirical model (boxes) and which particular options were implemented for this study (arrows). The Mersim model formulation is based on theory articulated by Castellani and Hafferty (2009) that conceptualises social-ecological complexity, in particular the focus on a disaggregated systems approach that allows non-linear system components to interact and, thereby allowing for emerging phenomena (i.e. Funtowicz and Ravetz, 1994). Experts helped to identify principle agent classes, such as household agents, government agents and spatial agents. This expert-based process also identified principle agent behaviour such as the harvesting of tea and the tapping of rubber. These livelihood related activities were put into a typical calendar and linked to associated regions and altitudes where necessary. The next step involved the specification of household attributes and household behaviours. A random sample of 1000 households (20 randomly selected households from 50 randomly selected villages) across Xishuangbanna were surveyed to elicit their key characteristics (i.e. location, household size, livelihoods, production, and income), their self selected attributes of subjective wellbeing, the principle human values that guide their lives, and their adaptation intentions. Intentions represent responses to questions that frame a specified hypothetical change, see Annex A in Smajgl et al. (2013) for the questionnaire. In this case the change households were asked to imagine: - Government payments are made once rubber trees are replaced by native trees to convert rubber monoculture plantations to an agroforestry farms. - The reduction of household income by 50% for at least five consecutive years. - The availability of tourism sector employment. Households had four principle response options: either
- To maintain their livelihood activity in their current household location; - To change their livelihood in their current household location; - To migrate out but maintain their current livelihood; or - To migrate out and change their livelihood activity. In each of these categories, responses to follow-on question informed estimates of the magnitude or type of livelihood response, the impediments to adaptation and/or the location for migration. The intentional data and behavioural changes elicited from household survey responses delimited the cognitive complexity of household agents to a more parsimonious depiction of largely reactive agents in the model. The sample data for attributes and behavioural rules was mapped into the total agent population by disproportional upscaling. Proportional up-scaling refers to a technique in which the proportions of responses in the sample is maintained to parameterise the whole population by simply replicating (or cloning) each response by sample size divided by population size (in this case multiplied by 1000). Disproportional up-scaling on the other hand changes proportions as some responses are used more often than others due to some scaling factors. In this case the proportions were amended to match the number of rubber plantations by using remote sensing data to map areas under rubber into village areas and then applying an empirically based range of farm sizes. This approach provided the number of rubber farms that were initialised for each village with a realistic farm size that collectively matched the actual area under rubber in the Prefecture. Responses from rubber farmers were used to parameterise modelled rubber farms and other responses were used for other respective land uses. This GIS-based adjustment was intended to represent a more realistic spatial distribution of simulated household behaviour. 3.6. Adaptation & objective Given the way we reduced agent cognition, agents step through a simple adaptation process, which allowed a reduction in the run time of the model so that live runs were able to be performed during the participatory modelling process (Smajgl et al., in press). Fig. 2 depicts the steps for household agents. Household agents respond to income levels derived from paid labour or agricultural activities. Households' objectives are implicit to their behavioural response functions (or rules). Modelled agents respond to livelihood related changes based on intentional data derived from the household survey responses. No additional optimisation or satisficing assumption is implemented. As a corollary, household expectations and learning are not explicitly represented but implicitly captured by the empirically derived response strategies.
Fig. 1. Parameterisation Sequence for the Mersim model, adapted from Smajgl et al. (2011).
A. Smajgl et al. / Environmental Modelling & Software 69 (2015) 187e195
Fig. 2. Flowchart for household agents.
3.7. Stochasticity Most parameters are assumed to be stochastic to resemble more realistic model assumptions, including crop prices, productivity, wages, and rainfall. For instance, the range for weekly income from rubber is set at ¥140e¥350 per mu for plantations below 900 m altitude and ¥45e¥200 per mu for plantations above 900 m altitude. The ranges were developed by experts in conjunction with time series data. 3.8. Initialisation and submodels The Mersim model utilises five sets of GIS data: (1) administrative boundaries down to administrative villages, (2) soil data, (3) land cover data, (4) rainfall projections, and (5) a digital elevation model. These datasets were used to specify the artificial landscape while the household survey provided the necessary data on household attributes and behavioural responses. Household income was calculated in weekly steps as the sum of all livelihoods that all household members are engaged in,
Fig. 3. Spatial distribution of rubber-based income in Xishuangbanna.
191
192
A. Smajgl et al. / Environmental Modelling & Software 69 (2015) 187e195
including the monetisation of subsistence production to avoid a misleading, under-estimated quantification of poverty. Fig. 3 shows how income generated from the cultivation of rubber was distributed across Xishuangbanna after the model initialisation. Dark markers show the location of villages with very high levels of rubber-based income, largely located in the lowlands. The income from rubber drops with decreasing green pigment. Initial model validation was largely based on the spatial distribution of rubber production, the total production of rubber and the distribution of household income. 3.9. Rubber related household rules The above described characterisation and parameterisation translates into the following behavioural rules for households engaged in farming rubber. In weekly time steps the income of households below 900 m is calculated by multiplying the area under rubber (in Mu ¼ 1/15 ha) with the profit per Mu, estimated by experts to range between ¥140 to¥ 350 per Mu and per week. In areas above 900 m productivity is lower and estimated to range between ¥45 and ¥200 per Mu. The logic for estimating the income of rubber farming households is explained thus:
The second critical logic for rubber farming is the response to the PES scheme. Once initialised, each agent refers to a survey response that corresponds to livelihood activity and the land use. Depending on the survey response the agent will perceive the PESbased incentive change and execute the change reported during the survey. In the case of a change, the previous livelihood activity is replaced by the response the representative household gave during the survey when asked “what would you do differently if the debated PES scheme was introduced” (see also Fig. 2). 4. Simulation results Decision makers engaged during the participatory research process identified three principle scenarios requiring simulation and comparison: The impact of government payments on the spatial extent of rubber; the PES scenario The impact of new regulation (combined with effective monitoring) that prohibits rubber above 900 m and reduced rubber below 900 by 20% to create ecological buffer zones along water bodies; the regulation scenario and The impact of additional off-farm employment options, in particular in the tourism industry; the tourism employment scenario. The PES and Regulation scenarios allow for the actual assessment of market-based and the regulatory approaches and their comparative evaluation. Based on options debated among decision makers, the market-based scenario assumes that payments are
made if below 900 m altitude 20% of rubber trees are replaced by native trees and above 900 m altitude 100% trees are replaced by native trees. The modelled economic incentive was calculated to fully compensate income loss due to rubber replacement. For the regulatory mechanism we assume that the reduction of rubber was not a choice but mandatory and is not accompanied by compensation payments. The PES scenario, if implemented outright, would assume effective monitoring and enforcement. To test the current state of rubber expansion a remote sensing based assessment was undertaken which indicated that the 1) the area under rubber is approximately 39% greater than the area listed in the official government census and 2) substantial areas of rubber extended into protected areas (Xu et al., 2013). In the respective scenario a lack of monitoring can be interpreted that a PES is paid even if the previous cover was not monoculture rubber. Given the lack of actual effective monitoring, we implemented two steps into agents' behavioural responses. First agents activate the survey response to the PES related survey question (i.e. “Yes, I would reduce the number of rubber trees by 20%”). Second, those agents representing households that indicated in the survey their capacity and interest to expand rubber, test this strategy as their income related to rubber (independent from the PES) is actually declining. This approach allows approximating a fictitious play-type learning process without the computationally expensive cognitive architecture. We introduced an additional scenario to address the lack of monitoring. The PES scenario with effective monitoring assumes that the current plantation area is constrained and no additional areas can be planted to rubber. All losses in rubber-based income are fully compensated. The regulation approach is similar to the monitored PES scenario but without compensation. The Tourism employment scenario aimed to test how effectively off-farm income could replace rubber as a livelihood. 4.1. Area under rubber Table 2 depicts the average impacts of the four policy scenarios on the area under rubber. Results are shown as per cent changes compared to the baseline and are calculated as averages over 200 simulation runs. The market-based incentive was adopted by over 90% of the rubber farming agents, which is higher than reported in Villamor et al. (2012) for rubber farmers in Jambi in Indonesia. However, the government payments without reinforced monitoring lead to a further increase of areas under rubber by 19.4% as the additional payments create an incentive sufficient for landowners to expand rubber as a mixed crop. Thereby, the PES scheme (without effective monitoring and enforcement) translates as a subsidy scheme for rubber. The expanded area of rubber production combined with the decrease in rubber density per ha results in a mean reduction of prefecture rubber production of 0.44%. If, however, institutional changes were implemented to enforce current and new land use plans, including this PES scheme, the results would be as expected with the area under rubber being eliminated above 900 m and rubber farmers below 900 m altitude are reducing the area under rubber by 20% (see Fig. 4). Table 2 Simulation results for area in Xishuangbanna under rubber and for average household income for 2029 as an average of 200 model runs. In % compared to baselines
Regulation
PES
Tourism employment
Area under rubber Average household income
61% 32%
þ19.4% þ0.1%
0% þ1.8%
A. Smajgl et al. / Environmental Modelling & Software 69 (2015) 187e195
193
Table 3 Farm income and livelihood activity proportions of 1000 randomly sampled households in Xishuangbanna. Income source
Off farm income Total farm income Subsistence income Total median income
Altitude 550e700 m
701e900 m
901e1200 m
Median (RMB)
Median (RMB)
Median (RMB)
4769 52,397 1828 58,994
865 22,064 3386 26,315
1403 22,377 3313 27,093
Table 4 Livelihood activity proportions of 1000 randomly sampled households in Xishuangbanna. Livelihood activity Fig. 4. Changes in area under rubber for 2029 under the “PES & Monitoring” scenario, with squares indicating location of villages and their colour indicating the change in area under rubber. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
Fig. 4 shows the spatial distribution of changes in rubber plantations for the PES scenario with effective monitoring and enforcement. The map shows Xishuangbanna with markers for each village location. Villages without rubber are white, largely located in the above 900 m altitudes. Red markers represent villages where rubber production ceases following the 100% reduction rule of the PES scheme. The yellow markers show the location of villages where production drops by 20%. The modelled legislative solution reduces the area under rubber by 61%, revealing that nearly a quarter of all rubber is currently located at altitudes above 900 m (see Fig. 3), land considered as marginal. The modelling assumes a 20% reduction in lowland rubber and 100% reduction in upland districts to comply with the legislation. Therefore 76.25% of the reduction occurs in lowland areas (ie 80% of 61%), revealing that 23.75% is currently located in upland districts. The provision of additional tourism employment has no impact on the total area planted to rubber. 5. Discussion The results indicate that the proposed PES scheme is likely to create perverse incentives if not accompanied with effective monitoring and enforcement. This is likely to trigger a further expansion of land planted to rubber in Xishuangbanna. If one assumes perfect monitoring and perfect compliance then the area under rubber would drop according to the 20%/100% rule. However, the macro-economic price-tag is substantial with annual compensation payments exceeding ¥34b. In contrast, regulatory instruments effectively achieve the goal of limiting rubber plantations. However, the environmental success comes at a cost as the ¥34b would be incurred by rubber farming households in the form of decreasing income levels. The lack of attractiveness of alternative land uses to households should not be a surprise as the difference between the high incomes derived from rubber and most other livelihoods is substantial (Table 3 and Table 4). Off-farm income relates to non-farm related income sources, subsistence income refers to the monetised values of household agricultural production consumed by the household, total farm income refers to the aggregate of subsistence income and production sold in markets and total income is the aggregate of all reported income sources. The column % main activities refers to percentage of respondents engaged in the main livelihood activities across the three altitude zones. The expansion
Growing rice Plantation (rubber) Plantation (tea) Growing maize Livestock Tourism
9.0% 70.6% 0.8% 0.4% 4.1% 0.4%
10.0% 57.5% 3.2% 11.8% 5.0% 0.5%
12.7% 8.0% 62.0% 0.8% 1.7% 0.4%
of alternative high-income livelihoods such as employment in tourism seems ineffective for reducing the cultivation of rubber, although this instrument results in relatively small increases in average income. The research results do not indicate that PES is an ineffective conservation mechanism in any institutional setting. In contrast, the results emphasise the need to design monetary incentives very carefully to avoid perverse incentives and ensure cost effective outcomes (Connor et al., 2008; Vatn, 2010). Conservation goals could be achieved without compromising income levels if existing initiatives are combined with an additional regulation that prohibits the establishment of new rubber cultivations and is aligned with effective field-based monitoring of land use change. The success of a regulatory alternative to PES is based on the assumption that statutory requirements are actually implemented. Considering that rubber is also currently cultivated in protected areas (Xu et al., 2013) statutory compliance is most likely an unrealistic assumption. This makes the comparisons between the payment schemes and regulation difficult. 6. Conclusions From a policy perspective four recommendations emerged to manage rubber expansion in Xishuangbanna. First, the currently proposed PES design has the potential to expand the area planted to rubber instead of meeting the reduction objective. Second, regulation could be successful but nearly 100,000 households are likely to experience a substantial reduction in income, demanding a carefully sequenced policy implementation that aligns income loss with some form of temporary compensation measures. Third, both the regulatory and PES options will require an effective improvement of field monitoring coupled with enforceable sanctions and penalties. Fourth, the development of alternative sectors is unlikely to achieve conservation goals. In synthesis, conservation goals require an effective combination of regulation, household payments and monitoring to achieve the land use change desired by national and local prefecture governments without either worsening the environmental degradation or creating social distress. Methodologically, the combination of fieldwork and agentbased social simulation within a participatory process has proven
194
A. Smajgl et al. / Environmental Modelling & Software 69 (2015) 187e195
very effective in revealing the unexpected emergence of potentially counter-productive effects of PES. The dominant expectation of participating decision makers was that the proposed PES design would be a cost effective instrument to achieve the balanced conservation productivity objective. Exposure to the participatory and simulated modelling process facilitated an effective learning experience for decision makers and decision influencers. It is critical to emphasise that this work was implemented in a learningfocused process to challenge the existing beliefs held by stakeholders and decision makers. The research focus distinguishes this work from modelling enterprises focused on predictive outcomes and estimates and promoted as validated truths (Bankes, 1993). The modelling results were presented to stakeholders to seed doubt as to the efficacy of PES schemes that do not account for localised context. Decision makers engaged in the present round of discussions concerned with biodiversity conservation and rubber production in Xishuangbanna perceived PES schemes as a panacea for achieving conservation goals. To the surprise of all stakeholders participating in the engagement process, the modelling revealed the potential for perverse incentives and shifted the focus to important design details of PES and the relevance of effective monitoring and enforcement mechanisms. Acknowledgements This research was funded by the DFAT-CSIRO Research for Development Alliance. The authors wish to thank also Beth Fulton for her constructive comments. References An, L., Linderman, M., Qi, J.G., Shortridge, A., Liu, J.G., 2005. Exploring complexity in a human-environment system: an agent-based spatial model for multidisciplinary and multiscale integration. Ann. Assoc. Am. Geogr. 95 (1), 54e79. Bankes, S., 1993. Exploratory modeling for policy analysis. Oper. Res. 41, 435e449. Bohensky, E., Smajgl, A., Brewer, T., 2013. Patterns in household-level engagement with climate change in Indonesia. Nat. Clim. Change 3, 348e351. Brenner, T., 2006. Agent learning representation: advice on modelling economic learning. In: Tesfatsion, L., Judd, K.L. (Eds.), Handbook of Computational Economics. North Holland, Amsterdam, pp. 895e947. Brown, D.G., Page, S.E., Riolo, R., Rand, W., 2004. Agent-based and analytical modeling to evaluate the effectiveness of greenbelts. Environ. Model. Softw. 19 (12), 1097e1109. Bunker, D.E., DeClerck, F., Bradford, J.C., Colwell, R.K., Perfecto, I., Phillips, O.L., Sankaran, M., Naeem, S., 2005. Species loss and aboveground carbon storage in a Tropical Forest. Science 310 (5750), 1029e1031. Castellani, B., Hafferty, F., 2009. Sociology and Complexity Science: a New Field of Inquiry. Springer, Berlin, Heidelberg. Chen, S., Yi, Z.-F., Campos-Arceiz, A., Chen, M.-Y., Webb, E.L., 2013. Developing a spatially-explicit, sustainable and risk-based insurance scheme to mitigate humanewildlife conflict. Biol. Conserv. 168 (0), 31e39. Chen, S., Yi, Z.F., Campos-Arceiz, A., Chen, M.Y., Webb, E., 2015. Fair and Sustainable Insurance-based Compensation to Mitigate Human-wildlife Conflict. submitted for publication. Chen, X., Vina, A., Shortridge, A., An, L., Liu, J., 2014. Assessing the effectiveness of payments for ecosystem services: an agent-based modeling approach. Ecol. Soc. 19 (1). Chen, X.D., Lupi, P., An, L., Sheely, R., Vina, A., Liu, J.G., 2012. Agent-based modeling of the effects of social normas on enrollment in payments for ecosystem services. Ecol. Model. 229, 16e24. Connor, J., Ward, J., Bryan, B., 2008. Exploring the cost effectiveness of land conservation auctions and payment policies. Aust. J. Agric. Resour. Econ. 52 (3), 303e319. Edmonds, B., Moss, S., 2001. The importance of representing cognitive processes in multi-agent models. In: Dorffner, G., Bischof, H., Hornik, K. (Eds.), Artificial Neural NetworkseICANN'2001. Springer, Heidelberg, pp. 759e766. FAO, 2013. FAOSTAT. Food and Agriculture Organisation of the United Nations (FAO). Fudenberg, D., Levine, D.K., 1998. The Theory of Learning in Games. MIT Press, Cambridge. Funtowicz, S.O., Ravetz, J.R., 1994. Emergent complex systems. Futures 26 (6), 568e582. Gilbert, N., 2008. Agent-based Models. SAGE Publications, Los Angeles. Grimm, V., Berger, U., Bastiansen, F., Eliassen, S., Ginot, V., Giske, J., Goss-Custard, J., Grand, T., Heinz, S.K., Huse, G., Huth, A., Jepsen, J.U., Jørgensen, C., Mooij, W.M., Müller, B., Pe'er, G., Piou, C., Railsback, S.F., Robbins, A.M., Robbins, M.M.,
Rossmanith, E., Rüger, N., Strand, E., Souissi, S., Stillman, R.A., Vabø, R., Visser, U., DeAngelis, D.L., 2006. A standard protocol for describing individual-based and agent-based models. Ecol. Model. 198 (1e2), 115e126. Grimm, V., Berger, U., DeAngelis, D.L., Polhill, J.G., Giske, J., Railsback, S.F., 2010. The ODD protocol: a review and first update. Ecol. Model. 221 (23), 2760e2768. Grumbine, R.E., Xu, J., 2011. Mekong hydropower development. Science 332 (6026), 178e179. Gu, L., Cheng, C.Q., 2007. Research on simulation of Wuhan land-use change based on GIS-agent Models (in Chinese). Urban Stud. 14 (6), 47e51. Guardiola-Claramonte, M., Troch, P.A., Ziegler, A.D., Giambelluca, T.W., Vogler, J.B., Nullet, M.A., 2008. Local hydrologic effects of introducing non-native vegetation in a tropical catchment. Ecohydrology 1 (1), 13e22. Guo, L., Gifford, R., 2002. Soil carbon stocks and land use change: a meta analysis. Glob. Change Biol. 8 (4), 345e360. Janssen, M., Goldstone, R.L., Menczer, F., Ostrom, E., 2008. The effect of rule choice in dynamic interactive spatial commons. Int. J. Commons 2 (2), 288e312. Kelly (Letcher), R.A., Jakeman, A.J., Barreteau, O., Borsuk, M.E., ElSawah, S., Hamilton, S.H., Henriksen, H.J., Kuikka, S., Maier, H.R., Rizzoli, A.E., van Delden, H., Voinov, A.A., 2013. Selecting among five common modelling approaches for integrated environmental assessment and management. Environ. Model. Softw. 47, 159e181. Li, H., Ma, Y., Aide, T.M., Liu, W., 2008. Past, present and future land-use in Xishuangbanna, China and the implications for carbon dynamics. For. Ecol. Manag. 255 (1), 16e24. Li, H., Mitchell Aide, T., Ma, Y., Liu, W., Cao, M., 2007a. Demand for rubber is causing the loss of high diversity rainforest in Southwest China. Biodivers. Conserv. 16, 1731e1745. Li, H.M., Ma, Y.X., Liu, W.J., Liu, W.J., 2009. Clearance and fragmentation of tropical rain forest in Xishuangbanna, SW, China. Biodivers. Conserv. 18, 3421e3440. Li, W., Dickinson, R.E., Fu, R., Niu, G.-Y., Yang, Z.-L., Canadell, J.G., 2007b. Future precipitation changes and their implications for tropical peatlands. Geophys. Res. Lett. 34 (1), L01403. Lieske, D.J., 2015. Coping with climate change: the role of spatial decision support tools in facilitating community adaptation. Environ. Model. Softw. 68 (0), 98e109. Mann, C.C., 2009. Addicted to rubber. Science 325, 564e566. Marshall, N.A., Smajgl, A., 2013. Understanding types of cattle graziers in the Burdekin River region, Australia, based on their adaptive capacity to climate variability and level of resource dependency. Rangel Ecol. Manag. 66 (1), 88e94. Matthews, R., Gilbert, N., Roach, A., Polhill, J., Gotts, N., 2007. Agent-based land-use models: a review of applications. Landsc. Ecol. 22 (10), 1447e1459. Müller, B., Balbi, S., Buchmann, C.M., de Sousa, L., Dressler, G., Groeneveld, J., Klassert, C.J., Le, Q.B., Millington, J.D.A., Nolzen, H., Parker, D.C., Polhill, J.G., Schlüter, M., Schulze, J., Schwarz, N., Sun, Z., Taillandier, P., Weise, H., 2014. Standardised and transparent model descriptions for agent-based models: current status and prospects. Environ. Model. Softw. 55 (0), 156e163. Robinson, D.T., Brown, D.G., Schreinemachers, P., Janssen, M.A., Huigen, M., Wittmer, H., Gotts, N., Promburom, P., Irwin, E., Berger, T., Gatzweiler, F., Barnaud, C., 2007. Comparison of empirical methods for building agent-based models in land use science. J. Land Use Sci. 2 (1), 31e55. Smajgl, A., Bohensky, E., 2013. Behaviour and space in agent-based modelling: poverty patterns in East Kalimantan, Indonesia. Environ. Model. Softw. 45, 8e14. Smajgl, A., Brown, D.G., Valbuena, D., Huigen, M.G.A., 2011. Empirical characterisation of agent behaviours in socio-ecological systems. Environ. Model. Softw. 26 (7), 837e844. Smajgl, A., Egan, S., Kirby, M., Mainuddin, M., Ward, J., Kroon, F., 2013. The Mekong Region Simulation (Mersim) Model - Design Document. CSIRO Climate Adaptation Flagship (Townsville). Smajgl, A., Foran, T., Dore, J., Ward, J., Larson, S., 2015. Visions, beliefs and transformation: exploring cross-sector and trans-boundary dynamics in the wider Mekong region. Ecol. Soc. in press. Sodhi, N.S., Koh, L.P., Brook, B.W., Ng, P.K.L., 2004. Southeast Asian biodiversity: an impending disaster. Trends Ecol. Evol. 19 (654e660). Stone, R., 2008. Showdown looms over a biological treasure trove. Science 319, 1604. Sun, R., Naveh, I., 2004. Simulating organizational decision-making using a cognitively realistic agent model. J. Artif. Soc. Soc. Simul. 7 (3), 5. Sun, Z., Müller, D., 2013. A framework for modeling payments for ecosystem services with agent-based models, Bayesian belief networks and opinion dynamics models. Environ. Model. Softw. 45 (0), 15e28. Tan, Z.H., Zhang, Y.P., Song, Q.H., Liu, W.J., Deng, X.B., Tang, J.W., Deng, Y., Zhou, W.J., Yang, L.Y., Yu, G.R., Sun, X.M., Liang, N.S., 2011. Rubber plantations act as water pumps in tropical China. Geophys. Res. Lett. 38 (24), L24406. Vatn, A., 2010. An institutional analysis of payments for environmental services. Ecol. Econ. 69, 1245e1252. Verburg, P.H., Soepboer, W., Veldkamp, A., Limpiada, R., Espaldon, V., Mastura, S.S.A., 2002. Modeling the spatial dynamics of regional land use: the CLUE-S model. Environ. Manag. 30 (3), 391e405. Villamor, G.B., Le, Q.B., Djanibekov, U., van Noordwijk, M., Vlek, P.L.G., 2014. Biodiversity in rubber agroforests, carbon emissions, and rural livelihoods: an agentbased model of land-use dynamics in lowland Sumatra. Environ. Model. Softw. 61 (0), 151e165. Villamor, G.B., Noordwijk, M.V., Troitzsch, K.G., Vlek, P.L., 2012. Human Decision Making for Empirical Agent-based Models: Construction and Validation.
A. Smajgl et al. / Environmental Modelling & Software 69 (2015) 187e195 International Environmental Modelling and Software Society (iEMSs). In: http://www.iemss.org/society/index.php/iemss-2012-proceedings. Wu, Z.L., Liu, H.M., Liu, L.Y., 2001. Rubber cultivation and sustainable development in Xishuangbanna, China. J. Sustain. Dev. World Ecol. 8, 337e345. Xu, J., 2006. The political, social, and ecological transformation of a landscape: the case of rubber in Xishuangbanna, China. Mt. Res. Dev. 26, 254e262. Xu, J., Grumbine, R., Beckschaefer, P., 2013. Landscape transformation and use of ecological and socioeconomic indicators in Xishuangbanna, Southwest China, Mekong Region. Ecol. Indic. 36, 749e756. Xu, J., Ma, E.T., Tashi, D., Fu, Y., Lu, Z., Melick, D., 2005. Integrating sacred knowledge for conservation: cultures and landscapes in southwest China. Ecol. Soc. 10 (2), 7. Xu, J.C., Melick, D.R., 2007. Rethinking the effectiveness of public protected areas in southwestern China. Conserv. Biol. 21, 318e328. €fer, P., Swetnam, R.D., 2014. Can Yi, Z.-F., Wong, G., Cannon, C.H., Xu, J., Beckscha carbon-trading schemes help to protect China's most diverse forest ecosystems? A case study from Xishuangbanna, Yunnan. Land Use Policy 38 (0), 646e656. Yi, Z.F., Cannon, C.H., Chen, J., Ye, C.X., Swetnam, R.D., 2013. Developing indicators of economic value and biodiversity loss for rubber planations in Xishuangbanna, Southwest China: a case study from Menglun township. Ecol. Indic. 36, 788e797.
195
Zhang, H.H., Zeng, Y.N., Bian, L., Yu, X.J., 2010. Modelling urban expansion using a multi agent-based model in the city of Changsha. J. Geogr. Sci. 20 (4), 540e556. Zhang, J., Cao, M., 1995. Tropical forest vegetation of Xishuangbanna Southwest China and its secondary changes, with special reference to some problems in local nature conservation. Biol. Conserv. 73, 225e238. Zhang, J.M., Wu, B., Shen, T.Y., 2004. Research on dynamic simulation of Beijing land covering and changing by applying agent modeling (in Chinese). J. East China Inst. Technol. 27 (1), 80e83. Zhang, L., Ma, L.C., Feng, L.M., 2007. New challenges facing traditional nature reserve: Asian elephant (Elephas maximus) conservation in China. Integr. Zool. 1 (4), 179e187. Zhou, Z., Hu, S., 2008. Study on impacts of rubber industry on ecological environment of Xishuangbanan. Environ. Sci. Surv. 27, 73e75. Zhou, Z., Hu, S., Tan, Y., 2006. Ecological envrionment impact from large-scale rubber planting in Xishuangbanna. Yunnan Eviron. Sci. 25, 67e69. Zhu, H., Xu, Z.F., Wang, H., Li, B.G., 2004. Tropical rainforest fragmentation and its ecological and species diversity change in Southern Yunnan. Biol. Conserv. 13, 1355e1372. Ziegler, A.D., Fox, J.M., Xu, J., 2009. The RubberJuggernaut. Science 324, 1024e1025.
Contents lists available at ScienceDirect
Environmental Modelling & Software journal homepage: www.elsevier.com/locate/envsoft
Assessing the effectiveness of payments for ecosystem services for diversifying rubber in Yunnan, China Alex Smajgl a, e, *, Jianchu Xu b, Stephen Egan c, Zhuang-Fang Yi b, John Ward d, Yufang Su b a
Mekong Region Futures Institute, Bangkok, Thailand World Agroforestry Center, East and Central Asia Region, Kunming 650201, China c CSIRO Ecosystem Sciences and Climate Adaptation Flagship, Highett, VIC 3190, Australia d Mekong Region Futures Institute, Vientiane, Lao PDR, Thailand e CSIRO Ecosystem Sciences, Townsville, Australia b
a r t i c l e i n f o
a b s t r a c t
Article history: Received 25 February 2015 Received in revised form 19 March 2015 Accepted 23 March 2015 Available online
Monoculture rubber plantations are rapidly expanding throughout Southeast Asia. The Prefecture of Xishuangbanna is characterised as both an epicenter of Chinese biodiversity and rubber production. The rapid conversion of native primary forest into rubber plantations has introduced tensions between the competing political goals to conserve biodiversity and heritage landscapes e both driving the rapidly emerging tourism industry e and economic aspirations at household and district levels. In China, decision makers discuss payments for ecosystem service schemes to resolve these tensions. As a component of the policy development process, this research project was invited to inform the political debate. Agent-based simulations revealed perverse outcomes of payments for ecosystem services, intended to encourage the conversion of monoculture rubber into agroforestry rubber. As an outcome of this modelling-based initiative, managing agencies have revised previously drafted payment schemes and reconsidered the importance of monitoring and regulatory approaches. Crown Copyright © 2015 Published by Elsevier Ltd. All rights reserved.
Keywords: Agent-based modelling Payment for ecosystem services Conservation Rubber Monoculture China
1. Introduction Monoculture rubber (Hevea brasiliensis) plantations are expanding rapidly throughout Xishuangbanna Prefecture, Southwest China, largely replacing evergreen broadleaf trees and swidden-fallow secondary vegetation, moving toward marginal land in higher altitudes and steeper slopes. More than 424,000 ha of forested areas with a high biodiversity value have already been converted to rubber planations. Continued expansion has the potential to encroach into existing primary forest and protected areas (Xu et al., 2013). There are costs and benefits associated with the land use and land cover changes, providing higher income for local farmers, increased integration into global economies, but introducing substantial and enduring threats to terrestrial biodiversity, impaired hydrological services, and reduced total carbon biomass (Li et al., 2008; Yi et al., 2014; Ziegler et al., 2009). Local and Provincial agencies and the central government in Beijing have
* Corresponding author. CSIRO Ecosystem Sciences, Townsville, Australia. Tel.: þ66 9956571200. E-mail address: [email protected] (A. Smajgl). http://dx.doi.org/10.1016/j.envsoft.2015.03.014 1364-8152/Crown Copyright © 2015 Published by Elsevier Ltd. All rights reserved.
increasingly expressed concerns regarding the environmental consequences of unregulated rubber expansion. Yunnan is an inland province at a low latitude and high elevation, characterized by the richest biological and cultural diversity in China and connects two global biodiversity hotspots, the Mountains of Southwest China and Indo-Burma. Yunnan is the source of the headwaters and major tributaries of the Yangtze, Salween, Irrawaddy, Mekong, Red, and Pearl Rivers, which combined directly or indirectly influence the lives of more than 600 million people. Substantial private investments in land development including rubber plantation are coinciding with centrally planned increases in the economic development of the region, focused primarily on of the construction of transportation infrastructure and hydropower dams (Grumbine and Xu, 2011). Concurrent with economic development is a rapid deterioration of the ecological health of the “roof” of Southwest China and Southeast Asia. Chinese institutions, governance and policy makers play a pivotal role in resolving the tensions that arise when the imperatives of conservation and economic development coincide. Intensifying demands for arable land and the multitude of benefits derived from biodiversity conservation are likely to amplify the importance of that role. Determining whether rubber plantations
188
A. Smajgl et al. / Environmental Modelling & Software 69 (2015) 187e195
and tropical rainforest can co-exist and how the negative impacts of infrastructure development on wildlife habitats can be mitigated are two priority questions confronting Chinese policy makers. Consistent with the arguments of Stone (2008), the likelihood of a conflict between the imperatives of economic growth and habitat preservation in Yunnan is escalating. Decision makers are sending mixed signals about their intentions. The Central Government initiated two nationwide conservation policies in the late 1990s e the Natural Forest Conservation Program (NFCP) and the Grain to Green Program. The Yunnan Provincial Government advocated increased funding to expand conservation areas and statutory conservation mandates in 2007 (Stone, 2008). However, rubber is a very profitable livelihood for local people. “Getting rich” and increasing wealth are also priority mandates for local government officials particularly at the Xishuangbanna Prefecture level. Balancing the two potentially competing views of conservation and economic development is a challenge for decision makers. Restricting development is not considered a plausible solution. Recent debates among decisionmakers and research academia have identified trade-offs between commercial exploitation of the region's natural resources and the need to promote sustainability of the ecosystems in the area. In 2009, a participatory research process was invited to facilitate discussions between these levels of government and provide insights of the potential consequences of proposed incentive changes; details of the participatory process are provided in Smajgl et al. (in press). Methodologically, the learning exercise employed agent-based social simulation to address the high levels of complexity and cognitive demands the decision context entails. This paper describes the characteristics and implementation of the Mersim (Mekong region simulation) model. The following Section outlines the competing goals of economic growth and ecological conservation against the backdrop of rubber expansion in Xishuangbanna, considered in a multi-level governance environment. The model design is then explained, followed by the simulation results for the most relevant policy scenarios. The paper concludes with the analysis of simulation results and subsequent policy deliberations. 2. Rubber, biodiversity and livelihood in Xishuangbanna Rubber (H. brasiliensis) has emerged as a key cash crop replacing traditional agriculture and secondary forests in the Mekong region (Ziegler et al., 2009), a direct result of Chinese aspirations for improved national rubber security and Chinese market demands, the world's largest rubber consumer (Mann, 2009). Forecasts indicate that global demand for natural rubber may outpace supply by 1.4 million metric tons by 2020 (FAO, 2013). Asia accounts for 97% of the world's natural rubber supply, traditionally from Thailand, Indonesia, and Malaysia. Entrepreneurs from China, Vietnam and Thailand are investing in rubber plantations in the less developed countries of the Mekong RegiondLaos, Cambodia and Myanmar. Currently, China contributes approximately 6.8% of the total global production in natural dry rubber (Table 1), but consumed 37% of global production in 2011 (FAO, 2013).
Situated in the upper Mekong, the Xishuangbanna Dai Autonomous Prefecture is a biologically diverse region in the tropical zone of southwest China. The Prefecture covers only 0.2% of the land area of China, it harbours some 16% of the vascular flora, 21.7% of mammals, and 36.2% of birds found in the country (Zhang and Cao, 1995). The northern, most tropical rainforest in the world is found here below 800 msl, which is the habitat that was first converted to rubber (Zhu et al., 2004). Xishuangbanna is also culturally diverse. More than two thirds of the population belongs to one of 12 ethnic minorities including the valley-dwelling Dai, who practice rice-paddy and home garden intensive agriculture, and upland people such as Hani (or Akha), Jinuo, Yao, Lahu and Bulang, largely engaged in swidden agriculture, with a forest mosaic landscape. It is one of the most famous tourism destinations regionally. Xishuangbanna National Nature Reserve was established in 1958, one of the earliest reserves in China, recognized for its high endemic species and for the conservation of Asian elephants, which are threatened by the expansion of rubber monoculture (Chen et al., 2013, submitted for publication; Zhang et al., 2007). After the 1949 Revolution, the new government of China identified rubber as an important strategic national resource. Rubber plantations and production were promoted through centralised governmental planning and regulatory initiatives. Commencing in the 1950's, the Central Government established rubber plantations or state rubber farms in Xishuangbanna Prefecture under the direction of the Ministry of Land Reclamation of Yunnan (Xu, 2006). As a consequence of the first Forestry Reform in 1983 (legislated as the Forestry Law of the People's Republic of China 1985), state forest tenure was reassigned to collectives and smallholder rubber plantations, initiated by lowland farmers assisted by technical guidance from the state coupled with government funded farm subsidies (Yi et al., 2013). From 1976 to 2007, rainforest in Xishuangbanna decreased from 13,193 km2 to 8336 km2, while rubber expanded from 249 km2 to 2256 km2 (Li et al., 2007a). The area of rubber plantations continued to increase to 4242 km2 in 2010, almost doubling the area estimated in 2007 (Xu et al., 2013). The rapid expansion of land devoted to rubber plantations has increased the income of some, while eroding the traditional livelihoods of many others, in particular those dependent on non-timber forest products. Moreover, the remaining forest area is increasingly fragmented and continues to degrade further through area and edge effects (Li et al., 2009; Xu et al., 2013). Assessments of biodiversity in rubber monocultures remain very low and dominated by introduced species (Wu et al., 2001). As a consequence of increasing monoculture rubber, biodiversity, as measured by the number of species found in the landscape (Sodhi et al., 2004; Xu and Melick, 2007) and carbon stocks both above (Bunker et al., 2005; Li et al., 2007b) and below ground (Guo and Gifford, 2002) are predicted to decrease. Consequentially, associated livelihood options are likely to decline (Xu et al., 2005). Ziegler et al. (2009) contend that expanded rubber monocultures are likely to affect the regional hydrology, as surface and groundwater recharge rates are likely to deteriorate because rubber
Table 1 Chinese rubber consumption and global production for 2009. Chinese rubber consumption and production for 2009
Tonnes
% Of world production
Domestic supply quantity Import Quantity Local Production Global production 2009
2,418,556 1,813,667 618,866 9,737,674
25% 19% 6%
Source FAO 2013.
A. Smajgl et al. / Environmental Modelling & Software 69 (2015) 187e195
plantations require more water compared to endemic rainforest (Guardiola-Claramonte et al., 2008; Tan et al., 2011). Increased application of fertilizers and residues of products used in rubber processing have led to eutrophication of water bodies, and the combination of herbicide use and reduced natural filtering has contaminated drinking supplies (Zhou and Hu, 2008; Zhou et al., 2006). Reduced and contaminated water supplies have, in some cases, resulted in conflicts in water access between villages. In summary, the degradation of ecosystem services reduces the environment's capacity to support a diversity of livelihoods, the broader economy and threatens the long term viability of the rubber industry. Maintaining productive rubber production and secured livelihoods coupled with policy support to sustain ecosystem services and biodiversity conservation remain as complementary policy objectives in Xishuangbanna. These joint imperatives manifest as initiatives to design policy instruments and economic incentives to promote environmentally friendly rubber across Xishuangbanna. Crucially, the political consensus and pathway for implementing such policy changes already exist. Under pressure from both national and provincial governments to address the environmental problems correlated with expanded rubber plantations, the Xishuangbanna Prefectural government and the rubber industry established the “Leadership Group for Environmentally Friendly Rubber” (LGEFR) in 2009. The paper describes the results of an invitation to provide research insights to evaluate the degree that conservation goals can be achieved without compromising the economic reality of sustained agricultural livelihoods and incomes. Payments for ecosystem services (PES) were identified as a key incentive to reduce the area under rubber. The PES design offered financial compensation to rubber farmers if they replaced 20% of their monoculture rubber by native vegetation in altitudes below 900 m and completely eliminated rubber in areas above 900 m. However, in the initial design the financial compensation replaced less than 10% of the subsequent loss in rubber income. As a corollary, farmers did not participate in the PES scheme. The resulting political impetus was to negotiate for full compensation of losses incurred by rubber farmers participating in the PES scheme. This research was co-designed by decision makers in Xishuangbanna to assess the potential impact of a revised PES scheme with full compensation of rubber farmers. 3. Model design 3.1. Methodological choice The overarching participatory process determined the research focus of the study as outlined above. Agent-based social simulation allows for the representation of such complex social-ecological interactions (Gilbert, 2008; Smajgl and Bohensky, 2013) and is suitable when the modelling aims for improved system understanding or social learning and where disaggregated effects of interactions between individuals and the environment need to be analysed (Kelly (Letcher) et al., 2013). Agent-based modelling is often employed to simulate land use change by incorporating prevailing environmental conditions, bio-physical processes, a diversity of human behaviours, decision-making, and policy alternatives (Bohensky et al., 2013; Brown et al., 2004; Marshall and Smajgl, 2013; Matthews et al., 2007; Robinson et al., 2007). Over recent years, modelling Land use and Land Cover Change (LUCC) has developed into a stronghold for ABM (Janssen et al., 2008; Robinson et al., 2007; Smajgl and Bohensky, 2013; Verburg et al., 2002). In China, ABM has been limited to a few studies, focused on LUCC and livelihood changes through PES, e.g. the Grain to Green
189
Program (Chen et al., 2012; Sun and Müller, 2013). Chen et al. (2014) is especially relevant, reporting the modelled impacts of PES schemes, in the form of cash and electricity payments, on forest cover. In this modelling context, albeit not related to rubber farming, all PES schemes resulted in increased forest cover. Other agent-based studies focus on urbanization processes (Gu and Cheng, 2007; Zhang et al., 2010, 2004). An et al. (2005) developed an ABM focussed on deforestation and forest degradation in China, to simulate the effects on forests and panda habitats associated with growing rural populations. At the time of writing, there are no published studies on the expansion of Chinese rubber monocultures and the economic effects on local people. Outside of China, rubber expansion has been modelled in the LB-LUDAS model to explore the impacts of PES on the trade-offs between goods and services of jungle rubber in Jambi Province, Indonesia (Villamor et al., 2014, 2012). Given the many effective LUCC-type applications of agent-based methodology stakeholders and the research team agreed to develop the Mekong region simulation (Mersim) model. The description of the agent-based model Mersim follows the ODD (Overview e Design concepts e Details) protocol (Grimm et al., 2006, 2010; Müller et al., 2014) and model details including Java code can be found in Smajgl et al. (2013). 3.2. Purpose of the model In initial rounds of facilitated consultations in Xishuangbanna, decision makers identified payments for replacing monoculture rubber plantations by agroforestry rubber as the policy intervention that needed to be assessed (details for the participatory process can be found in Smajgl et al., in press). This type of payment for ecosystem service was to be combined with full compensation of losses incurred by participating rubber farmers. Decision makers identified the aggregate area planted to rubber as the key indicator of PES performance, which required the spatially explicit modelling of behavioural changes of households and environmental dynamics (Smajgl et al., in press); see Lieske (2015) for the relevance of spatially explicit modelling. 3.3. State variables The participatory process placed the stakeholder priorities at the core of the model design and determined the state variables as: rubber production, household income and poverty, land use and household livelihoods, household location and migration. 3.4. Emergence Following from the stakeholder-defined modelling goal, the spatial extent of rubber and the broader land use change were critical emergent phenomena, which are inherently linked to the livelihood and migration decisions implemented by households. Livelihood changes emerge in response to changes in environmental, economic and institutional factors. Land use patterns evolve as a consequence of socialeenvironmental interaction, including the area used for cultivating rubber. The perception of changes in the agent's environmental demands typically requires a cognitive structure that allows for the evaluation of options and their likely consequences. Several scholars argue for cognitively rich agents in agent-based social simulation (i.e. Edmonds and Moss, 2001; Sun and Naveh, 2004) and alternate learning algorithms have been developed and tested (Brenner, 2006; Fudenberg and Levine, 1998). In an empirical model, however, the characterization of the cognitive architecture of agents is a challenge as individuals display a diversity of responses to manifold
190
A. Smajgl et al. / Environmental Modelling & Software 69 (2015) 187e195
drivers, subject to variable motivations and expressed as different behavioural thresholds. This introduces high levels of uncertainty for assumptions regarding cognitive structure and parameterisation. Therefore, the cognitive reasoning regarding potential behavioural adaptation was elicited through a randomised survey of Xishuangbanna households, providing intentional data for parameterising largely reactive agents. Thereby, the cognitive reasoning is implicit to the intentional data elicited for all relevant scenarios and the parameterisation process. This is best explained through the characterisation and parameterisation process itself. 3.5. Household data for parameterisation The parameterisation process is described based on the framework provided in Smajgl et al. (2011). Fig. 1 shows the principle parameterisation steps required in an empirical model (boxes) and which particular options were implemented for this study (arrows). The Mersim model formulation is based on theory articulated by Castellani and Hafferty (2009) that conceptualises social-ecological complexity, in particular the focus on a disaggregated systems approach that allows non-linear system components to interact and, thereby allowing for emerging phenomena (i.e. Funtowicz and Ravetz, 1994). Experts helped to identify principle agent classes, such as household agents, government agents and spatial agents. This expert-based process also identified principle agent behaviour such as the harvesting of tea and the tapping of rubber. These livelihood related activities were put into a typical calendar and linked to associated regions and altitudes where necessary. The next step involved the specification of household attributes and household behaviours. A random sample of 1000 households (20 randomly selected households from 50 randomly selected villages) across Xishuangbanna were surveyed to elicit their key characteristics (i.e. location, household size, livelihoods, production, and income), their self selected attributes of subjective wellbeing, the principle human values that guide their lives, and their adaptation intentions. Intentions represent responses to questions that frame a specified hypothetical change, see Annex A in Smajgl et al. (2013) for the questionnaire. In this case the change households were asked to imagine: - Government payments are made once rubber trees are replaced by native trees to convert rubber monoculture plantations to an agroforestry farms. - The reduction of household income by 50% for at least five consecutive years. - The availability of tourism sector employment. Households had four principle response options: either
- To maintain their livelihood activity in their current household location; - To change their livelihood in their current household location; - To migrate out but maintain their current livelihood; or - To migrate out and change their livelihood activity. In each of these categories, responses to follow-on question informed estimates of the magnitude or type of livelihood response, the impediments to adaptation and/or the location for migration. The intentional data and behavioural changes elicited from household survey responses delimited the cognitive complexity of household agents to a more parsimonious depiction of largely reactive agents in the model. The sample data for attributes and behavioural rules was mapped into the total agent population by disproportional upscaling. Proportional up-scaling refers to a technique in which the proportions of responses in the sample is maintained to parameterise the whole population by simply replicating (or cloning) each response by sample size divided by population size (in this case multiplied by 1000). Disproportional up-scaling on the other hand changes proportions as some responses are used more often than others due to some scaling factors. In this case the proportions were amended to match the number of rubber plantations by using remote sensing data to map areas under rubber into village areas and then applying an empirically based range of farm sizes. This approach provided the number of rubber farms that were initialised for each village with a realistic farm size that collectively matched the actual area under rubber in the Prefecture. Responses from rubber farmers were used to parameterise modelled rubber farms and other responses were used for other respective land uses. This GIS-based adjustment was intended to represent a more realistic spatial distribution of simulated household behaviour. 3.6. Adaptation & objective Given the way we reduced agent cognition, agents step through a simple adaptation process, which allowed a reduction in the run time of the model so that live runs were able to be performed during the participatory modelling process (Smajgl et al., in press). Fig. 2 depicts the steps for household agents. Household agents respond to income levels derived from paid labour or agricultural activities. Households' objectives are implicit to their behavioural response functions (or rules). Modelled agents respond to livelihood related changes based on intentional data derived from the household survey responses. No additional optimisation or satisficing assumption is implemented. As a corollary, household expectations and learning are not explicitly represented but implicitly captured by the empirically derived response strategies.
Fig. 1. Parameterisation Sequence for the Mersim model, adapted from Smajgl et al. (2011).
A. Smajgl et al. / Environmental Modelling & Software 69 (2015) 187e195
Fig. 2. Flowchart for household agents.
3.7. Stochasticity Most parameters are assumed to be stochastic to resemble more realistic model assumptions, including crop prices, productivity, wages, and rainfall. For instance, the range for weekly income from rubber is set at ¥140e¥350 per mu for plantations below 900 m altitude and ¥45e¥200 per mu for plantations above 900 m altitude. The ranges were developed by experts in conjunction with time series data. 3.8. Initialisation and submodels The Mersim model utilises five sets of GIS data: (1) administrative boundaries down to administrative villages, (2) soil data, (3) land cover data, (4) rainfall projections, and (5) a digital elevation model. These datasets were used to specify the artificial landscape while the household survey provided the necessary data on household attributes and behavioural responses. Household income was calculated in weekly steps as the sum of all livelihoods that all household members are engaged in,
Fig. 3. Spatial distribution of rubber-based income in Xishuangbanna.
191
192
A. Smajgl et al. / Environmental Modelling & Software 69 (2015) 187e195
including the monetisation of subsistence production to avoid a misleading, under-estimated quantification of poverty. Fig. 3 shows how income generated from the cultivation of rubber was distributed across Xishuangbanna after the model initialisation. Dark markers show the location of villages with very high levels of rubber-based income, largely located in the lowlands. The income from rubber drops with decreasing green pigment. Initial model validation was largely based on the spatial distribution of rubber production, the total production of rubber and the distribution of household income. 3.9. Rubber related household rules The above described characterisation and parameterisation translates into the following behavioural rules for households engaged in farming rubber. In weekly time steps the income of households below 900 m is calculated by multiplying the area under rubber (in Mu ¼ 1/15 ha) with the profit per Mu, estimated by experts to range between ¥140 to¥ 350 per Mu and per week. In areas above 900 m productivity is lower and estimated to range between ¥45 and ¥200 per Mu. The logic for estimating the income of rubber farming households is explained thus:
The second critical logic for rubber farming is the response to the PES scheme. Once initialised, each agent refers to a survey response that corresponds to livelihood activity and the land use. Depending on the survey response the agent will perceive the PESbased incentive change and execute the change reported during the survey. In the case of a change, the previous livelihood activity is replaced by the response the representative household gave during the survey when asked “what would you do differently if the debated PES scheme was introduced” (see also Fig. 2). 4. Simulation results Decision makers engaged during the participatory research process identified three principle scenarios requiring simulation and comparison: The impact of government payments on the spatial extent of rubber; the PES scenario The impact of new regulation (combined with effective monitoring) that prohibits rubber above 900 m and reduced rubber below 900 by 20% to create ecological buffer zones along water bodies; the regulation scenario and The impact of additional off-farm employment options, in particular in the tourism industry; the tourism employment scenario. The PES and Regulation scenarios allow for the actual assessment of market-based and the regulatory approaches and their comparative evaluation. Based on options debated among decision makers, the market-based scenario assumes that payments are
made if below 900 m altitude 20% of rubber trees are replaced by native trees and above 900 m altitude 100% trees are replaced by native trees. The modelled economic incentive was calculated to fully compensate income loss due to rubber replacement. For the regulatory mechanism we assume that the reduction of rubber was not a choice but mandatory and is not accompanied by compensation payments. The PES scenario, if implemented outright, would assume effective monitoring and enforcement. To test the current state of rubber expansion a remote sensing based assessment was undertaken which indicated that the 1) the area under rubber is approximately 39% greater than the area listed in the official government census and 2) substantial areas of rubber extended into protected areas (Xu et al., 2013). In the respective scenario a lack of monitoring can be interpreted that a PES is paid even if the previous cover was not monoculture rubber. Given the lack of actual effective monitoring, we implemented two steps into agents' behavioural responses. First agents activate the survey response to the PES related survey question (i.e. “Yes, I would reduce the number of rubber trees by 20%”). Second, those agents representing households that indicated in the survey their capacity and interest to expand rubber, test this strategy as their income related to rubber (independent from the PES) is actually declining. This approach allows approximating a fictitious play-type learning process without the computationally expensive cognitive architecture. We introduced an additional scenario to address the lack of monitoring. The PES scenario with effective monitoring assumes that the current plantation area is constrained and no additional areas can be planted to rubber. All losses in rubber-based income are fully compensated. The regulation approach is similar to the monitored PES scenario but without compensation. The Tourism employment scenario aimed to test how effectively off-farm income could replace rubber as a livelihood. 4.1. Area under rubber Table 2 depicts the average impacts of the four policy scenarios on the area under rubber. Results are shown as per cent changes compared to the baseline and are calculated as averages over 200 simulation runs. The market-based incentive was adopted by over 90% of the rubber farming agents, which is higher than reported in Villamor et al. (2012) for rubber farmers in Jambi in Indonesia. However, the government payments without reinforced monitoring lead to a further increase of areas under rubber by 19.4% as the additional payments create an incentive sufficient for landowners to expand rubber as a mixed crop. Thereby, the PES scheme (without effective monitoring and enforcement) translates as a subsidy scheme for rubber. The expanded area of rubber production combined with the decrease in rubber density per ha results in a mean reduction of prefecture rubber production of 0.44%. If, however, institutional changes were implemented to enforce current and new land use plans, including this PES scheme, the results would be as expected with the area under rubber being eliminated above 900 m and rubber farmers below 900 m altitude are reducing the area under rubber by 20% (see Fig. 4). Table 2 Simulation results for area in Xishuangbanna under rubber and for average household income for 2029 as an average of 200 model runs. In % compared to baselines
Regulation
PES
Tourism employment
Area under rubber Average household income
61% 32%
þ19.4% þ0.1%
0% þ1.8%
A. Smajgl et al. / Environmental Modelling & Software 69 (2015) 187e195
193
Table 3 Farm income and livelihood activity proportions of 1000 randomly sampled households in Xishuangbanna. Income source
Off farm income Total farm income Subsistence income Total median income
Altitude 550e700 m
701e900 m
901e1200 m
Median (RMB)
Median (RMB)
Median (RMB)
4769 52,397 1828 58,994
865 22,064 3386 26,315
1403 22,377 3313 27,093
Table 4 Livelihood activity proportions of 1000 randomly sampled households in Xishuangbanna. Livelihood activity Fig. 4. Changes in area under rubber for 2029 under the “PES & Monitoring” scenario, with squares indicating location of villages and their colour indicating the change in area under rubber. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
Fig. 4 shows the spatial distribution of changes in rubber plantations for the PES scenario with effective monitoring and enforcement. The map shows Xishuangbanna with markers for each village location. Villages without rubber are white, largely located in the above 900 m altitudes. Red markers represent villages where rubber production ceases following the 100% reduction rule of the PES scheme. The yellow markers show the location of villages where production drops by 20%. The modelled legislative solution reduces the area under rubber by 61%, revealing that nearly a quarter of all rubber is currently located at altitudes above 900 m (see Fig. 3), land considered as marginal. The modelling assumes a 20% reduction in lowland rubber and 100% reduction in upland districts to comply with the legislation. Therefore 76.25% of the reduction occurs in lowland areas (ie 80% of 61%), revealing that 23.75% is currently located in upland districts. The provision of additional tourism employment has no impact on the total area planted to rubber. 5. Discussion The results indicate that the proposed PES scheme is likely to create perverse incentives if not accompanied with effective monitoring and enforcement. This is likely to trigger a further expansion of land planted to rubber in Xishuangbanna. If one assumes perfect monitoring and perfect compliance then the area under rubber would drop according to the 20%/100% rule. However, the macro-economic price-tag is substantial with annual compensation payments exceeding ¥34b. In contrast, regulatory instruments effectively achieve the goal of limiting rubber plantations. However, the environmental success comes at a cost as the ¥34b would be incurred by rubber farming households in the form of decreasing income levels. The lack of attractiveness of alternative land uses to households should not be a surprise as the difference between the high incomes derived from rubber and most other livelihoods is substantial (Table 3 and Table 4). Off-farm income relates to non-farm related income sources, subsistence income refers to the monetised values of household agricultural production consumed by the household, total farm income refers to the aggregate of subsistence income and production sold in markets and total income is the aggregate of all reported income sources. The column % main activities refers to percentage of respondents engaged in the main livelihood activities across the three altitude zones. The expansion
Growing rice Plantation (rubber) Plantation (tea) Growing maize Livestock Tourism
9.0% 70.6% 0.8% 0.4% 4.1% 0.4%
10.0% 57.5% 3.2% 11.8% 5.0% 0.5%
12.7% 8.0% 62.0% 0.8% 1.7% 0.4%
of alternative high-income livelihoods such as employment in tourism seems ineffective for reducing the cultivation of rubber, although this instrument results in relatively small increases in average income. The research results do not indicate that PES is an ineffective conservation mechanism in any institutional setting. In contrast, the results emphasise the need to design monetary incentives very carefully to avoid perverse incentives and ensure cost effective outcomes (Connor et al., 2008; Vatn, 2010). Conservation goals could be achieved without compromising income levels if existing initiatives are combined with an additional regulation that prohibits the establishment of new rubber cultivations and is aligned with effective field-based monitoring of land use change. The success of a regulatory alternative to PES is based on the assumption that statutory requirements are actually implemented. Considering that rubber is also currently cultivated in protected areas (Xu et al., 2013) statutory compliance is most likely an unrealistic assumption. This makes the comparisons between the payment schemes and regulation difficult. 6. Conclusions From a policy perspective four recommendations emerged to manage rubber expansion in Xishuangbanna. First, the currently proposed PES design has the potential to expand the area planted to rubber instead of meeting the reduction objective. Second, regulation could be successful but nearly 100,000 households are likely to experience a substantial reduction in income, demanding a carefully sequenced policy implementation that aligns income loss with some form of temporary compensation measures. Third, both the regulatory and PES options will require an effective improvement of field monitoring coupled with enforceable sanctions and penalties. Fourth, the development of alternative sectors is unlikely to achieve conservation goals. In synthesis, conservation goals require an effective combination of regulation, household payments and monitoring to achieve the land use change desired by national and local prefecture governments without either worsening the environmental degradation or creating social distress. Methodologically, the combination of fieldwork and agentbased social simulation within a participatory process has proven
194
A. Smajgl et al. / Environmental Modelling & Software 69 (2015) 187e195
very effective in revealing the unexpected emergence of potentially counter-productive effects of PES. The dominant expectation of participating decision makers was that the proposed PES design would be a cost effective instrument to achieve the balanced conservation productivity objective. Exposure to the participatory and simulated modelling process facilitated an effective learning experience for decision makers and decision influencers. It is critical to emphasise that this work was implemented in a learningfocused process to challenge the existing beliefs held by stakeholders and decision makers. The research focus distinguishes this work from modelling enterprises focused on predictive outcomes and estimates and promoted as validated truths (Bankes, 1993). The modelling results were presented to stakeholders to seed doubt as to the efficacy of PES schemes that do not account for localised context. Decision makers engaged in the present round of discussions concerned with biodiversity conservation and rubber production in Xishuangbanna perceived PES schemes as a panacea for achieving conservation goals. To the surprise of all stakeholders participating in the engagement process, the modelling revealed the potential for perverse incentives and shifted the focus to important design details of PES and the relevance of effective monitoring and enforcement mechanisms. Acknowledgements This research was funded by the DFAT-CSIRO Research for Development Alliance. The authors wish to thank also Beth Fulton for her constructive comments. References An, L., Linderman, M., Qi, J.G., Shortridge, A., Liu, J.G., 2005. Exploring complexity in a human-environment system: an agent-based spatial model for multidisciplinary and multiscale integration. Ann. Assoc. Am. Geogr. 95 (1), 54e79. Bankes, S., 1993. Exploratory modeling for policy analysis. Oper. Res. 41, 435e449. Bohensky, E., Smajgl, A., Brewer, T., 2013. Patterns in household-level engagement with climate change in Indonesia. Nat. Clim. Change 3, 348e351. Brenner, T., 2006. Agent learning representation: advice on modelling economic learning. In: Tesfatsion, L., Judd, K.L. (Eds.), Handbook of Computational Economics. North Holland, Amsterdam, pp. 895e947. Brown, D.G., Page, S.E., Riolo, R., Rand, W., 2004. Agent-based and analytical modeling to evaluate the effectiveness of greenbelts. Environ. Model. Softw. 19 (12), 1097e1109. Bunker, D.E., DeClerck, F., Bradford, J.C., Colwell, R.K., Perfecto, I., Phillips, O.L., Sankaran, M., Naeem, S., 2005. Species loss and aboveground carbon storage in a Tropical Forest. Science 310 (5750), 1029e1031. Castellani, B., Hafferty, F., 2009. Sociology and Complexity Science: a New Field of Inquiry. Springer, Berlin, Heidelberg. Chen, S., Yi, Z.-F., Campos-Arceiz, A., Chen, M.-Y., Webb, E.L., 2013. Developing a spatially-explicit, sustainable and risk-based insurance scheme to mitigate humanewildlife conflict. Biol. Conserv. 168 (0), 31e39. Chen, S., Yi, Z.F., Campos-Arceiz, A., Chen, M.Y., Webb, E., 2015. Fair and Sustainable Insurance-based Compensation to Mitigate Human-wildlife Conflict. submitted for publication. Chen, X., Vina, A., Shortridge, A., An, L., Liu, J., 2014. Assessing the effectiveness of payments for ecosystem services: an agent-based modeling approach. Ecol. Soc. 19 (1). Chen, X.D., Lupi, P., An, L., Sheely, R., Vina, A., Liu, J.G., 2012. Agent-based modeling of the effects of social normas on enrollment in payments for ecosystem services. Ecol. Model. 229, 16e24. Connor, J., Ward, J., Bryan, B., 2008. Exploring the cost effectiveness of land conservation auctions and payment policies. Aust. J. Agric. Resour. Econ. 52 (3), 303e319. Edmonds, B., Moss, S., 2001. The importance of representing cognitive processes in multi-agent models. In: Dorffner, G., Bischof, H., Hornik, K. (Eds.), Artificial Neural NetworkseICANN'2001. Springer, Heidelberg, pp. 759e766. FAO, 2013. FAOSTAT. Food and Agriculture Organisation of the United Nations (FAO). Fudenberg, D., Levine, D.K., 1998. The Theory of Learning in Games. MIT Press, Cambridge. Funtowicz, S.O., Ravetz, J.R., 1994. Emergent complex systems. Futures 26 (6), 568e582. Gilbert, N., 2008. Agent-based Models. SAGE Publications, Los Angeles. Grimm, V., Berger, U., Bastiansen, F., Eliassen, S., Ginot, V., Giske, J., Goss-Custard, J., Grand, T., Heinz, S.K., Huse, G., Huth, A., Jepsen, J.U., Jørgensen, C., Mooij, W.M., Müller, B., Pe'er, G., Piou, C., Railsback, S.F., Robbins, A.M., Robbins, M.M.,
Rossmanith, E., Rüger, N., Strand, E., Souissi, S., Stillman, R.A., Vabø, R., Visser, U., DeAngelis, D.L., 2006. A standard protocol for describing individual-based and agent-based models. Ecol. Model. 198 (1e2), 115e126. Grimm, V., Berger, U., DeAngelis, D.L., Polhill, J.G., Giske, J., Railsback, S.F., 2010. The ODD protocol: a review and first update. Ecol. Model. 221 (23), 2760e2768. Grumbine, R.E., Xu, J., 2011. Mekong hydropower development. Science 332 (6026), 178e179. Gu, L., Cheng, C.Q., 2007. Research on simulation of Wuhan land-use change based on GIS-agent Models (in Chinese). Urban Stud. 14 (6), 47e51. Guardiola-Claramonte, M., Troch, P.A., Ziegler, A.D., Giambelluca, T.W., Vogler, J.B., Nullet, M.A., 2008. Local hydrologic effects of introducing non-native vegetation in a tropical catchment. Ecohydrology 1 (1), 13e22. Guo, L., Gifford, R., 2002. Soil carbon stocks and land use change: a meta analysis. Glob. Change Biol. 8 (4), 345e360. Janssen, M., Goldstone, R.L., Menczer, F., Ostrom, E., 2008. The effect of rule choice in dynamic interactive spatial commons. Int. J. Commons 2 (2), 288e312. Kelly (Letcher), R.A., Jakeman, A.J., Barreteau, O., Borsuk, M.E., ElSawah, S., Hamilton, S.H., Henriksen, H.J., Kuikka, S., Maier, H.R., Rizzoli, A.E., van Delden, H., Voinov, A.A., 2013. Selecting among five common modelling approaches for integrated environmental assessment and management. Environ. Model. Softw. 47, 159e181. Li, H., Ma, Y., Aide, T.M., Liu, W., 2008. Past, present and future land-use in Xishuangbanna, China and the implications for carbon dynamics. For. Ecol. Manag. 255 (1), 16e24. Li, H., Mitchell Aide, T., Ma, Y., Liu, W., Cao, M., 2007a. Demand for rubber is causing the loss of high diversity rainforest in Southwest China. Biodivers. Conserv. 16, 1731e1745. Li, H.M., Ma, Y.X., Liu, W.J., Liu, W.J., 2009. Clearance and fragmentation of tropical rain forest in Xishuangbanna, SW, China. Biodivers. Conserv. 18, 3421e3440. Li, W., Dickinson, R.E., Fu, R., Niu, G.-Y., Yang, Z.-L., Canadell, J.G., 2007b. Future precipitation changes and their implications for tropical peatlands. Geophys. Res. Lett. 34 (1), L01403. Lieske, D.J., 2015. Coping with climate change: the role of spatial decision support tools in facilitating community adaptation. Environ. Model. Softw. 68 (0), 98e109. Mann, C.C., 2009. Addicted to rubber. Science 325, 564e566. Marshall, N.A., Smajgl, A., 2013. Understanding types of cattle graziers in the Burdekin River region, Australia, based on their adaptive capacity to climate variability and level of resource dependency. Rangel Ecol. Manag. 66 (1), 88e94. Matthews, R., Gilbert, N., Roach, A., Polhill, J., Gotts, N., 2007. Agent-based land-use models: a review of applications. Landsc. Ecol. 22 (10), 1447e1459. Müller, B., Balbi, S., Buchmann, C.M., de Sousa, L., Dressler, G., Groeneveld, J., Klassert, C.J., Le, Q.B., Millington, J.D.A., Nolzen, H., Parker, D.C., Polhill, J.G., Schlüter, M., Schulze, J., Schwarz, N., Sun, Z., Taillandier, P., Weise, H., 2014. Standardised and transparent model descriptions for agent-based models: current status and prospects. Environ. Model. Softw. 55 (0), 156e163. Robinson, D.T., Brown, D.G., Schreinemachers, P., Janssen, M.A., Huigen, M., Wittmer, H., Gotts, N., Promburom, P., Irwin, E., Berger, T., Gatzweiler, F., Barnaud, C., 2007. Comparison of empirical methods for building agent-based models in land use science. J. Land Use Sci. 2 (1), 31e55. Smajgl, A., Bohensky, E., 2013. Behaviour and space in agent-based modelling: poverty patterns in East Kalimantan, Indonesia. Environ. Model. Softw. 45, 8e14. Smajgl, A., Brown, D.G., Valbuena, D., Huigen, M.G.A., 2011. Empirical characterisation of agent behaviours in socio-ecological systems. Environ. Model. Softw. 26 (7), 837e844. Smajgl, A., Egan, S., Kirby, M., Mainuddin, M., Ward, J., Kroon, F., 2013. The Mekong Region Simulation (Mersim) Model - Design Document. CSIRO Climate Adaptation Flagship (Townsville). Smajgl, A., Foran, T., Dore, J., Ward, J., Larson, S., 2015. Visions, beliefs and transformation: exploring cross-sector and trans-boundary dynamics in the wider Mekong region. Ecol. Soc. in press. Sodhi, N.S., Koh, L.P., Brook, B.W., Ng, P.K.L., 2004. Southeast Asian biodiversity: an impending disaster. Trends Ecol. Evol. 19 (654e660). Stone, R., 2008. Showdown looms over a biological treasure trove. Science 319, 1604. Sun, R., Naveh, I., 2004. Simulating organizational decision-making using a cognitively realistic agent model. J. Artif. Soc. Soc. Simul. 7 (3), 5. Sun, Z., Müller, D., 2013. A framework for modeling payments for ecosystem services with agent-based models, Bayesian belief networks and opinion dynamics models. Environ. Model. Softw. 45 (0), 15e28. Tan, Z.H., Zhang, Y.P., Song, Q.H., Liu, W.J., Deng, X.B., Tang, J.W., Deng, Y., Zhou, W.J., Yang, L.Y., Yu, G.R., Sun, X.M., Liang, N.S., 2011. Rubber plantations act as water pumps in tropical China. Geophys. Res. Lett. 38 (24), L24406. Vatn, A., 2010. An institutional analysis of payments for environmental services. Ecol. Econ. 69, 1245e1252. Verburg, P.H., Soepboer, W., Veldkamp, A., Limpiada, R., Espaldon, V., Mastura, S.S.A., 2002. Modeling the spatial dynamics of regional land use: the CLUE-S model. Environ. Manag. 30 (3), 391e405. Villamor, G.B., Le, Q.B., Djanibekov, U., van Noordwijk, M., Vlek, P.L.G., 2014. Biodiversity in rubber agroforests, carbon emissions, and rural livelihoods: an agentbased model of land-use dynamics in lowland Sumatra. Environ. Model. Softw. 61 (0), 151e165. Villamor, G.B., Noordwijk, M.V., Troitzsch, K.G., Vlek, P.L., 2012. Human Decision Making for Empirical Agent-based Models: Construction and Validation.
A. Smajgl et al. / Environmental Modelling & Software 69 (2015) 187e195 International Environmental Modelling and Software Society (iEMSs). In: http://www.iemss.org/society/index.php/iemss-2012-proceedings. Wu, Z.L., Liu, H.M., Liu, L.Y., 2001. Rubber cultivation and sustainable development in Xishuangbanna, China. J. Sustain. Dev. World Ecol. 8, 337e345. Xu, J., 2006. The political, social, and ecological transformation of a landscape: the case of rubber in Xishuangbanna, China. Mt. Res. Dev. 26, 254e262. Xu, J., Grumbine, R., Beckschaefer, P., 2013. Landscape transformation and use of ecological and socioeconomic indicators in Xishuangbanna, Southwest China, Mekong Region. Ecol. Indic. 36, 749e756. Xu, J., Ma, E.T., Tashi, D., Fu, Y., Lu, Z., Melick, D., 2005. Integrating sacred knowledge for conservation: cultures and landscapes in southwest China. Ecol. Soc. 10 (2), 7. Xu, J.C., Melick, D.R., 2007. Rethinking the effectiveness of public protected areas in southwestern China. Conserv. Biol. 21, 318e328. €fer, P., Swetnam, R.D., 2014. Can Yi, Z.-F., Wong, G., Cannon, C.H., Xu, J., Beckscha carbon-trading schemes help to protect China's most diverse forest ecosystems? A case study from Xishuangbanna, Yunnan. Land Use Policy 38 (0), 646e656. Yi, Z.F., Cannon, C.H., Chen, J., Ye, C.X., Swetnam, R.D., 2013. Developing indicators of economic value and biodiversity loss for rubber planations in Xishuangbanna, Southwest China: a case study from Menglun township. Ecol. Indic. 36, 788e797.
195
Zhang, H.H., Zeng, Y.N., Bian, L., Yu, X.J., 2010. Modelling urban expansion using a multi agent-based model in the city of Changsha. J. Geogr. Sci. 20 (4), 540e556. Zhang, J., Cao, M., 1995. Tropical forest vegetation of Xishuangbanna Southwest China and its secondary changes, with special reference to some problems in local nature conservation. Biol. Conserv. 73, 225e238. Zhang, J.M., Wu, B., Shen, T.Y., 2004. Research on dynamic simulation of Beijing land covering and changing by applying agent modeling (in Chinese). J. East China Inst. Technol. 27 (1), 80e83. Zhang, L., Ma, L.C., Feng, L.M., 2007. New challenges facing traditional nature reserve: Asian elephant (Elephas maximus) conservation in China. Integr. Zool. 1 (4), 179e187. Zhou, Z., Hu, S., 2008. Study on impacts of rubber industry on ecological environment of Xishuangbanan. Environ. Sci. Surv. 27, 73e75. Zhou, Z., Hu, S., Tan, Y., 2006. Ecological envrionment impact from large-scale rubber planting in Xishuangbanna. Yunnan Eviron. Sci. 25, 67e69. Zhu, H., Xu, Z.F., Wang, H., Li, B.G., 2004. Tropical rainforest fragmentation and its ecological and species diversity change in Southern Yunnan. Biol. Conserv. 13, 1355e1372. Ziegler, A.D., Fox, J.M., Xu, J., 2009. The RubberJuggernaut. Science 324, 1024e1025.