Hard facts and software: The co-production of indicators in a land-use planning model.
Using land-use models in deliberative planning is promoted as an example of how environmental decision-making can be subject to both: 1) facts about how the interaction between human action and natural processes; and, 2) local perspectives on how land-use planning processes can incorporate normative concerns. This 'normative' input is often shaped and limited by the presentation of the modelled facts. This paper, however, shows that the selection and measurement of indicators, the primary outcomes of modelling exercises, are subject to a highly particular assessment, influenced by social factors, political choices and technical limitations. This revised understanding does not invalidate the use of models in participatory planning. But it does open up considerable space for stakeholders in deliberative contexts to question and challenge the evidence-based policy implications of so-called 'fact-based' modelling exercises.
Land-use planning; models; co-production; normative influence; public participation
Models are conventionally understood as a means of 'letting the data speak for itself', by presenting relevant and integrated facts for environmental decision-making. At the same time, there is increasing insistence that modelling become a participatory policy strategy, rather than one managed exclusively by experts and received by other stakeholders (Cinderby, 1999; Craig, Harris and Weiner, 2002; Van der Sluijs, 2002; Yearley et al., 2003). These calls often focus on how the public can be involved in decision-making based on models, rather than on the modelling process itself, thereby separating the work of experts from the work of other participants. In other words, they focus on how to add 'culture' to the 'science' and stir (Robbins 2003a). This suggests that subjective decision-making revolves around a stable set of objective, universal facts.
There is, however, a growing recognition that modelling activities are not simply, nor best described as technical activities (that albeit need to be supplemented by social concerns), but that their very construction embodies social and political influences (Garb, Pulver and VanDeveer, 2008; Turnhout, Hisschemoller and Eijsackers, 2007). This paper contributes to the debates about the normative dimensions of that which is often perceived as 'fact'. It examines one aspect of models--outcome indicators--in the context of a land-use planning project in the region of the Mbaracayu Biosphere Reserve in Paraguay. The analysis presents ways in which indicators, key model inputs and outputs, are imbued with unrecognised normative dimensions. In turn, this analysis problematises continued approaches that praise the importance of participation in environmental governance, but promote the inclusion of different perspectives only insofar as the 'evidence' produced by models is taken as the baseline for debate.
SOCIAL INFLUENCE IN FACTUAL MODELS
The assumed objectivity and neutrality of modelling in land use came under scrutiny in the mid-late 1990s with publications such as the 1995 edited volume 'Ground Truth' (Pickles, 1995). The works in this volume highlighted the role of social and political inquiry in the development and use of GIS models for decision-making support. Subsequent work has examined the politics of computer simulations in relation to climate change modelling (Demeritt, 2001), and the related technologies of remote sensing and satellite imagery (Litfin, 1999). Authors argue that these 'tools' are hardly the neutral mechanisms for increasing knowledge, and thus making more informed decisions, that modelling experts often claim. Furthermore, the social and political influence in technical applications does not only appear 'downstream'--the point at which science is applied to policy problems. These influences also exist 'upstream'--the point at which model-based knowledge is created (Demeritt, 2001). Often the development of models is based on partial or simplistic assumptions, working on the basis of a 'naive sociology' (Wynne, 1989; Yearley, 1999). For example, Taylor (1992) analysed the 1970s MIT project to create a systems dynamics model of nomadic pastoralists who had been exposed to a long drought. He illustrates that the assumptions on which the model was based could have been configured completely differently in terms of how the historical data were used, how individuals were treated, and how potential external influences may have impacted on the system. Had these assumptions been configured differently, it would have had important effects on the outputs and conclusions of the modelling exercise.
Discrete categories used in modelling (e.g. land-use type, wildlife type, vegetation cover, etc.) are important examples of how different taxonomies and understanding of similarity and difference can affect modelling outcomes. Naidoo and Hill, for example, call for the integration of 'traditional ecological knowledge' and remote sensing after finding a 'knowledge gain' from using traditional Ache categories for vegetation cover over scientific ones in the Mbaracayu Reserve (2006). Robbins goes further in his inquiry of classification, putting the modelling process itself at the centre of inquiry in order to examine the 'softness' of 'hard' tools (Robbins, 2003b). He illustrates how landscape categories, primary inputs in land-use models, 'are nothing more than a widely different set of reflectance clusters, aggregated based on the arbitrary decision-rule of the analyst...' (Robbins, 2003a: 249). It is not that one type of classification reflects knowledge that is better, worse, more scientific or more cultural than others. Rather, it is that 'the process of resource use and conflict... gives rise to the systems of meaning each community deploys in the first place' (Robbins, 2003a: 248).
Likewise, Harvey and Chrisman (1998) illustrate how GIS analyses reflect negotiations between social groups through mediating 'boundary objects'. In their analysis, even once a definition of 'wetlands' is agreed upon, four government agencies remain in disagreement about which mapped areas consist of 'wetlands'. The authors cite an impressive 90% disagreement, and even a considerable extension in the areas delineated by each agency as 'wetland' resulted in high levels of disagreement. This is explained by the different functions of each agency, invoking different 'purposes, procedures, sources, definitions, and logic... each agency's purpose delimits which methods are acceptable for fulfilling their mandate' (Harvey and Chrisman, 1998: 1689). They explain: 'The geographic boundaries of these different wetlands delineate administrative elements in the environment... The boundary object "wetlands" indicates the disciplinary and institutional boundaries of different groups' (Harvey and Chrisman, 1998: 1689). They conclude: 'GIS technology and technoscience are not monolithic autonomous edifices but the localized results of processes of negotiation that involve the construction of artifacts to fit various social perspectives' (Harvey and Chrisman, 1998: 1693).
The insights of these authors suggest that using models to depict the natural world does not produce objective accounts of landscape realities, but interpretations that are profoundly influenced by social and political milieus. This does not imply that these models have no relevance to the 'natural' world, but that they express certain dimensions of the natural world that correspond with distinct norms, values, preferences, agendas and priorities. Such analyses, therefore, can be read as co-productionist accounts of models, as sites where the socio-political order is co-produced with the creation of policy-relevant knowledge (Miller, 2004; Jasanoff, 2004a; Forsyth, 1999) for supporting decision making around environmental governance such as scenario building (Garb et al., 2008) and land-use planning. To say that environmental knowledge is co-produced means that particular ways of knowing about and understanding environmental problems are inseparable from the way in which individuals and organisations attempt to order and control environmental problems (Jasanoff, 2004a: 2-3). Environmental facts are thus co-produced with norms, values and judgements.
If knowledge is co-produced with the social world, then it cannot but reflect the characteristics of the social world (Jasanoff, 2004b). This includes ways in which structural power is manifest in differential access to resources and decision-making authority. Indeed, co-production can explain why certain ways of reading and relating environmental problems (and solutions) gain authority and credibility, while others are marginalised or even silenced altogether (Jasanoff, 2004a: 4). In this sense, making co-productionist accounts of model building for land-use planning explicit has profound implications for environmental governance--particularly in light of the strong normative stance in favour of public participation in public decision-making. Indeed, many authors argue for the democratisation of knowledge that is seen as credible and legitimate in policy making (Forsyth, 2004; Jasanoff and Wynne; 1998, Backstrand, 2003). Co-productionist analyses are important means of democratising such knowledge, by destabilising the widespread understanding of its basis as factual and making the ways in which it is co-produced with socio-political order more explicit. The tenets of knowledge thus become contestable rather than unquestionable. A co-productionist approach to knowledge is not about 'disproving' evidence for policy, but rather emphasising the political aspects of policy making alongside the technical aspects.
MODELLING LAND-USE OUTCOMES IN THE MBARACAYU: BOUNDARIES OF EVIDENCE AND PARTICIPATION
In 2004, in partnership with a Paraguayan environmental NGO called the Fundacion Moises Bertoni (FMB) and the Alberta Research Council (now, Alberta Innovates--Technology Futures) undertook a project to help stakeholders improve environmental governance in Paraguay's Mbaracayu Biosphere Reserve (MBR). The project was funded by the Canadian International Development Agency (CIDA), under the auspices of the Technology Transfer Fund (TTF)--which emphasises social and technical capacity building. Developing an interactive land-use planning model was seen as complementary to building representative and deliberative environmental governance institutions.
The MBR is part of the Interior Atlantic Forest, valued as a habitat for numerous and diverse species. Many consider Paraguay's Atlantic forest a globally important source of biodiversity and the MBR is said to contain upwards of 90% of Paraguay's species classified as 'rare' or 'endangered' (Hill and Padwe, 2000). The Interior Atlantic Forest, furthermore, is widely considered among the world's most threatened ecoregions. This corresponds with Landsat data that shows that the country's forested area was reduced to 176,741[km.sup.2] from 202,202[km.sup.2] between the 1990's and the 2000's (Huang et al., 2009). Forested area in Canindeyu, the administrative department where the Mbaracayu is located, contracted from 8262.97 [km.sup.2] in the 1990's to 4903.05 [km.sup.2] between 2000 and 2010--a change in forest cover of a staggering 40.66% in roughly a decade (Huang et al., 2009).
Land use in the MBR is a mixture of conservation and agriculture. The FMB manages a 64,000 hectare reserve, protected from most human activity, called the Mbaracayu Natural Forest Reserve (MNFR). Outside the MNFR, most Mbaracayu inhabitants are predominantly agriculturalists and range from small-scale subsistence growers occupying ten hectare family farms to proprietors of large-scale soy plantations and cattle ranches covering hundreds of hectares. Over the past three decades Paraguay, among other soy producing countries, has experienced the 'soy boom' (Dros, 2004). For some the soy boom has been a financial windfall. The cultivation of soybeans, however, has become a controversial business perceived by many to occupy a central role in environmental and social problems (Altieri and Pengue, 2005; Fogel and Riquelme, 2005; Garcia-Lopez and Arizpe, 2010; Steward, 2007). Evictions of peasant and indigenous populations, pesticide-related epidemics and deaths, and even kidnappings and murders have been linked to the soybean as conflict infects the soy-producing countryside.
Under the auspices of this project a land-use planning tool called ALCES (A Landscape Cumulative Effects Simulator) was adapted to the Mbaracayu context. ALCES projects indicator outcomes relative to different land-use trajectories, allowing stakeholders to compare and assess their potential ecological and social consequences. ALCES functions using evidence established through systematic and context-specific research about relationships between land-use options and outcomes. Outcomes are represented by pre-selected indicators; measurable variables that reflect progress, or lack of progress, towards certain predefined goals. Indicator data must be available, or calculable, for different points along the trajectory. Data availability and the ability to directly and quantifiably relate these data to land-use scenarios determine the appropriateness of both potential indicators and possible trajectories or scenarios. Ideally, the number of indicators should be limited but those selected should completely and accurately represent the characteristics of planning goals in the most parsimonious way possible.
In addition to projecting indicator outcomes of different land-use trajectories, ALCES designers make explicit their intention to extend participation in land-use planning beyond the modellers, to a wide array of stakeholders including land users and regional policy makers. Advocates for the use of ALCES for facilitating participatory planning claim that 'The active engagement of stakeholders in the modelling process and the transparency of the model, in which the key processes are all under the control of the user, promotes the understanding and acceptance of the outcomes' (Schneider et al., 2003: online).
Establishing indicators and generating simulations
Using both evidence and wider participation to establish indicators for use within the model is emphasised by the ALCES model, owing that they 'should include variables that are meaningful to the local community and communicate both the ecological and socioeconomic implications of land-use' (Carlson, 2006: 6). Three indicators were initially chosen by the modeller for simulation with the ALCES software: net agricultural income; net agricultural income for small producers; and remaining 'natural' area. These indicators were based on economic, agricultural, land-use, land cover and demographic data from the Mbaracayu, from neighbouring regions believed to be comparable, and from global trends in the agricultural sector (see Table 1). Additionally, they were chosen based on their direct and quantifiable relationship with distinct land uses.
To enhance wider participation in the construction of the model, work was undertaken to establish 'community-based indicators' (See Box 1)--following other projects involving ALCES (Parlee, 1998). This was among my principle roles in the project and in the development of the model. Researchers over the past decade have claimed that community-based indicators are vital to ensuring the inclusion of public values and local knowledge in the monitoring and assessment of project outcomes (Gasteyer and Butler, 2000; Nurick and Johnson, 1998). Furthermore, compared to more conventional ecological indicators, authors claim that they are more likely to measure a wider array of social and political issues that are related to environmental management (Mitchell and Davis, 2005).
Focus groups generated debate about what sustainable development entails in the Mbaracayu, and how social and environmental sustainability might be measured by indicators (see Table 2). Themes included concerns about natural resources such as water and forest; basic needs and services such as nutrition, health and education; social concerns such as sense of community, language and religion; economic concerns about production and income, technical assistance and infrastructure; and land availability and tenure. These themes translated into a variety of potential indicators, such as: quality and quantity of and access to water; availability and access to forest products; quantity and variety of personal consumption; levels of agricultural production and agricultural income; and land distribution. It was made clear in the focus groups, that indicating sustainability means addressing a variety of issues from a range of perspectives.
Ultimately, no new indicators were added to the original suite as a result of the focus groups. Discussion among project team members around the exclusion of the community-based indicators was limited, even non-existent. In fact, this exclusion was not posed as a 'decision' entailing judgement, but rather an obvious outcome of both circumstance and inherent qualities of the indicators vis-a-vis the modelling program. First, many of the community-based indicators were excluded on the basis of a lack of data, making it impossible to characterise the relationship between indicator and outcomes. Secondly, many were excluded due to an inherent incompatibility with the ALCES program based on a lack of quantifiable, direct relationships with land use. Each of these rationales for excluding the established community based indicators from the land-use planning model is unsurprising within the logic of modelling. These exclusions, therefore, remained unproblematised by some project stakeholders as the model was consolidated and disseminated, precisely because they remained congruent with the logic of modelling.
Other stakeholders, however, found these exclusions very problematic indeed. For example:
The indicators presented for ALCES differ in good measure from those that were identified by the FMB staff in (indicator development) workshops; on this point I am very sorry that during the ALCES presentations the priority was to show the program and convince us of the importance and capacity of it, and not to work on key aspects such as developing appropriate indicators--those presented were identified and presented as the necessary ones (Comment made by an FMB Manager at an ALCES workshop in Asuncion).
Such comments evoke critique of how participatory development is often symbolic rather than effectual--conceptualised and implemented as political 'theatre', where despite public involvement 'on stage', the real decision-making happens 'off-stage' (Cornwall and Brock, 2005; Lazarus, 2008).
In order to generate the scenarios for the subsequent 50 years, baseline data for each indicator, and indicator responses to three land-use trajectories were established through available data (see Table 1) and assumptions regarding the rates of soy expansion, population growth and smallholder agricultural expansion, and soil fertility over time. In 2006 a first draft of the report detailing the analysis was released, and in March of the same year scenarios were presented to stakeholders in the CARJ. (1) In 2007, the final report was published on the ALCES website (Sorrenson, 1997).
Indicator responses to the three land-use trajectories were assessed as follows: (2)
* Undercurrent practices, land used for shifting cultivation increases by 1.5% which accounts for the population growth. Soy production experiences a 6% growth, reflecting the growth in the industry. Furthermore, 22% of the biosphere is protected by private reserves (including the MNFR) or indigenous reserves, leaving only 78% of the total land area available for agricultural expansion. Within 50 years all available land will have been converted to agriculture, and rendered increasingly unproductive, making agricultural production unfeasible. Natural habitat shrinks to the minimum guaranteed by reserves, and incomes for smallholders vanishes as production cannot cover the costs of cultivation(Sorrenson, 1997).
* Under a 4% increase in protected areas in the CARJ, the scenario is practically identical to the current practices scenario except for that agriculture can only expand to cover a maximum of 74% of the land, as an additional 4 is protected as reserve. Thus, the natural habitat indicator grows slightly to 26% while smallholder income shrinks similarly to the first scenario, with a slightly steeper decline because there is less land available for expansion.
* With the adoption of sustainable agriculture (defined in this case as notill cultivation) by all producers in the CARJ, soil quality is maintained, productivity increases and agricultural expansion slows to reflect only increases in soy production. No land needs to be cleared to replace degraded land as soil quality is maintained and the need for abandonment is eliminated, thus the decline in natural habitat is considerably slower. Furthermore, productivity is not merely maintained but increased under sustainable agriculture and thus smallholder incomes also rise.
Implications of the simulations: Fact-based policy recommendations
Implementation of sustainable agricultural practices, according to interpretations of the ALCES simulations, plays a large role in averting social and ecological disaster. Under current practices, reads the 2007 ALCES report, 'the simulation predicts that in 50 years the Cuenca will be a region of severe poverty and ecological degradation' (Sorrenson, 1997). The report continues: 'Fortunately, the scenario analysis indicates that economic and ecological ruin need not occur. Sustainable agricultural practices, in particular, have the potential to support the Mbaracayu program's goal of supporting both biodiversity and the well-being of local inhabitants' (Sorrenson, 1997). Overall, it is assumed that by implementing sustainable agriculture, environmental degradation, and deforestation in particular, is more likely to be mitigated because: a) the simulator has shown people what they have to do; and, b) through technical assistance they have been shown how to do it. Non-compliance is attributed to a lack of the understanding and sophistication required to see the local, regional and global significance of the goal of reducing deforestation, and the vision to understand the impact of specific behaviours on the end goal.
CO-PRODUCING FACTS AND NORMS IN INDICATORS
Data were used to create ALCES simulations, which then became evidence for policy choices. The roles of the expert and the participants were clearly separated: the expert was to establish the facts by generating the simulations; the public were then invited to debate normative land-use trade-offs and planning goals, in light of these facts. But to what extent are the outcomes depicted in the scenarios unequivocal? Upon closer inspection, policy inputs cannot be neatly compartmentalised into norms on the one hand and facts on the other; the simulations do not provide plain facts about the social and ecological trajectory of the Mbaracayu. They provide, rather, one perspective that is deeply influenced by a variety of caveats, including the choices and preferences of the modeller and requirements of the model. Five such caveats are discussed here.
First, selection of the indicators favoured those with a short term, quantifiable relationship with land use. Indicators are vital elements in models; they signal the changes in outputs (such as income or forest cover) that correspond with changes in inputs (such as land-use practices and effective protection policies). The signal is both in vector (direction of change, be it positive or negative; desirable or undesirable) and strength (the degree to which the change is effected). Because of this assumedly 'predictive' capacity, indicators must have a causal, quantifiable and incremental relationship with the outcome.
These indicator requirements effectively eliminate those variables and issues that may be highly relevant to land-use outcomes, but that are not easily quantifiable or exhibit a less clear or direct relationship with land use. The requirements exclude variables that cannot be linked to land-use outcomes via an inherently causal relationship. Indeed, often understandings and available evidence about causal relationships become problematic in conjunction with complex systems and socio-political processes. This, however, does not invalidate the long term, complex and non-incremental, or non-fixed incremental connection between land use and many aspects of wellbeing.
For example, the 'well-being of local communities' is an important goal of the land-use plan. The ability of 'measurable indicators that closely reflect land-use changes', to monitor progress towards this goal, however, are unlikely to be sufficient. Concerns around health and education are important to local people, but related indicators are inappropriate for integration into the model because of their unclear links to land-use options and changes. For example, education is likely to have dramatic impacts on land use through many intermediate processes, such as literacy and alternative employment opportunities. This relationship, however, is better described as a correlation, and is not easily quantified (x number of years of education will result in reduced deforestation in the order of y number of hectares). Such important local issues are at risk of being disregarded either because of unavailable data, or more importantly, because they are considered inappropriate as indicators of land use (lacking a direct link with land use).
Secondly, the ALCES indicators express implicit preferences for the importance of certain indicator dimensions over others. These preferences affect the way in which the indicator is measured, and ultimately, policy implications. For example, when the model refers to natural area and local people refer to forest resources, the object is the same. The perspective, however, changes and this has implications for identifying the indicator. The existence and maintenance of water and forest are, in some respects, encapsulated by the ALCES indicator of 'natural area'. The relevance of natural area to standing forest and the existence of forest products is apparent (though not all forests are equal in their production of all forest products, and not all so-called 'natural area' is forest; it also includes savannah grasslands). The relationship between natural area and water quality and quantity is also implied, assuming that the larger the portion of land base accounted for by 'natural habitat', the more likelihood that riparian zones will exist, and be sufficiently large to protect water resources.
In focus groups, local resource users recast these particular ecological issues as issues of access to resources. Participants drew attention to an aspect of both forest cover and water not addressed directly by the ALCES indicators, but of tantamount importance to local people: access and distribution of resources. While the existence of clean water and forest products is relevant to resource availability, an issue of equal importance is accessibility. The degree to which resources are protected, or even the degree to which this protection contributes to maintaining high standards of quality, says little about who has access to and control over those resources. Thus, the existence of forest cover, while an obvious prerequisite to access to forest, is an incomplete measure of access to forest products. Several focus group participants commented that the indicator 'forest area' needs to be accompanied by one which measures 'forest area available for public use' in order to be an adequate measure of sustainable development.
The different motivations behind establishing such indicators can have ramifications for how they become manifest in policy. For example, ALCES indicators captured reduced forest cover and acute deforestation specifically because of the link with reduced biodiversity. Community-based indicators also capture reduced forest cover, but because deforestation can have negative implications for local people who use the forests for firewood, food and cash income. An overt emphasis on forest protection as a means of protecting biodiversity, without paying attention to local peoples' dependence on forest resources and existing property rights, may cause hardship for local people such as displacement and restrictions on forest use.
Thirdly, the data requirements for indicator calculations may be more complex and nuanced than meets the eye. Often, social, cultural and political context determines what kind of data is required for a particular indicator. For example, the ALCES indicator of agricultural income is measured with several data sources, accounting for productivity, production inputs and other costs, and the rate of smallholder agricultural expansion. Intuitively, these data seem sufficient to arrive at a reasonable estimate of agricultural income for small producers. However, the community-based indicators suggest many other factors that contribute to the ability of small producers to profit from agricultural production. For example, commercialisation support, improved transportation, infrastructure and technical assistance were emphasised in the focus groups as important factors relating to agricultural income (see Figure 1). Indeed, at existing levels of production, producers often encounter difficulties selling, and heavily depend on intermediaries for commercialisation and transportation. If the price is low, the weather is bad, or the intermediaries have just simply moved on, producers may get stuck with crops and no buyer at all. Mountains of produce can be laid to waste before any buyer has made an appearance (Anonymous interview).
This example illustrates that contextual factors have a fundamental impact on how, and whether, farmers are able to market their production. This means that the seemingly reasonable assumption that agricultural income can be calculated with several core data sources renders the accuracy of the indicator to measure actual income questionable. Improved incomes in the watershed may depend somewhat on increased production. But, incomes would be much more responsive to other factors such as even marginal improvements in infrastructure. Thus, modelling indicators may overlook mid points in development processes, and have significantly more complex data requirements than meets the eye.
The fourth factor that confounds the idea of factual indicators is that indicators are limited by data availability, and missing data can be the result of concerted and deliberate decisions. Obviously, a model has a high level of data dependence. Without quantitative information about land-use patterns and relationships between land-use and ecological, social and economic indicators, the program cannot run. 'Avoidable error' and the quality of data have been posited as the fundamental issue in the successful implementation of ALCES. Data availability was the most forthcoming limitation to using the community-based indicators in cumulative effects modelling. In this case study, data availability largely drove the design of the simulations. Once data sources were identified, the scenarios were constructed around them. This is not particularly surprising, especially in developing countries where there is little funding for research.
While not wanting to exaggerate the importance of insufficient or incorrect data, it is important to recognise how much of a shortcoming this can potentially be for using database-oriented tools for land-use planning, such as ALCES, in Paraguay. Data availability is neither a politically neutral phenomenon nor does it have politically neutral consequences. It is important to look not only at the 'missing-data' issue, but also to examine and reflect on which data are missing, why they are missing (aside from the obvious 'lack of research and funding for research') and how this might redirect focus from certain activities to others. This redirection may be a product of false assumptions, created by the elimination of a necessary part of the picture.
For example, lack of data, research gaps and lack of resources to conduct research are serious obstacles to 'feeding' simulation models. However, the case may be that data is unavailable for more covert reasons, such as illegal activity. For example, illegal logging has an observable impact on the landscape in the MBR. One only has to be present to see logs streaming out of the region on the truck beds, in ox-carts and on the backs of men. However, because specific information on the impact of illegal logging is not likely to be offered or asked for, this aspect of land use is omitted from the cumulative effects analysis. Meanwhile, the ecological burden of this activity is attributed to factors that are recognised by the cumulative effects simulator: one of these being the unsustainable agricultural techniques used by smallholders. Thus, the excluded data and 'missing explanations' may be inadvertently compensated for by exaggerating others.
Furthermore, many social indicators lack agreed-upon thresholds and quantifiable cause-effect trends backed by research. It is more likely for data to be available for expert-led indicators as the issues are more standardised and predictable, due to existing research on their potential causal effects relationships. Integrating expert-led with community-based indicators may be desirable, but research and data, particularly for contextually specific community-based indicators, are not likely to exist short of designing and implementing expensive surveys for the region in question.
Fifth and finally, the ALCES indicators privilege certain elements of diverse livelihoods over others. For example, the ALCES indicator 'agricultural income' resonated deeply with focus group participants. With few opportunities to engage in wage labour, selling cash crops continues to be the most economically significant and stable way to earn income. The design of the ALCES indicator was well executed in that it captured differences between the high earners and the low earners. This was done by splitting the agricultural income indicator into two: income from soy production, carried out nearly exclusively on large properties in the Mbaracayu region, and income on small properties. To a large extent this helps to capture the most significant income distribution issue in the watershed: the differential in earnings between large and small landholders. This also reflects the difference in political leverage vis-a-vis a small but highly influential number of large landowners over the majority of poor small-scale producers.
However, while an important part of a complex web that comprises overall livelihood strategies, selling cash crops is not the only concern, nor is it particularly the most important for local livelihoods. Focus group participants emphasised other facets of the rural livelihoods in the watershed, attaching major importance to levels of subsistence crop production for the status of nutrition and overall well-being. Opportunities for employment income, though indeed less emphasised than agricultural income, were nonetheless considered an increasingly vital part of livelihood strategies.
CHALLENGING MODELS AS PRIVILEGED PARTICIPANT
The previous section offered accounts of how the simulations developed within the ALCES model were based on social and political preference, convenience and even necessity--rather than scientific prowess. Although the expert-driven ALCES simulations are presented as evidence, the analysis suggests that those involved in land-use planning processes must also engage with the normative aspects of evidence; that facts are co-produced with particular preferences, assumptions and norms that often go unrecognised.
Participatory modelling or model-based participation?
Combining participatory and evidence-based approaches is often more about using the model to shape participation than using participation to shape the model. The idea of ALCES is to use the facts about the outcomes of land uses to reorient public values, norms and attitudes towards sustainable practices. By showing people future impacts of land uses, the modellers encourage the adoption of the land-use practices or systems to which the indicators respond most favourably. The rationale is that if people can see the future implications of current land use, they will be more likely to tailor their behaviours in a way that produces desired outcomes. Fallibility in the scenarios is attributed to flawed or incomplete data and the decisions that people make about land use. Therefore, to improve the reliability and validity of the scenarios, improved data quality, and improved influence of the models over the decisions people make are prescribed.
But improving the evidence, and the influence that evidence has over peoples' behaviour is not so straightforward. Rather than being 'fed' with facts and 'producing' facts, models are imbued with values, priorities and perspectives that are linked to particular perspectives that are associated with social identity. Furthermore, in the case of experts, this social identity is most often hidden (Van der Sluijs, 2001) as the model outcomes are presented as facts rather than as products of a particular interpretation. This spells the end of any potential for deliberation, because the model is taken for uncontestable truth, or at least passed for an objective view. Despite the emphasis on ALCES as a tool for 'participatory modelling' the scenarios are used as an unquestioned baseline that shapes and limits deliberative input. Avoiding this orientation toward 'model-based participation' requires an understanding that it is not simply that more knowledge is needed to get ALCES right. It is, rather, that the fundamental notion of what it means to 'get it right' is itself a matter of debate, as social facts influence environmental norms differently (Wynne, 1996).
Some critics of the ALCES modelling exercise may be tempted to blame the incompleteness and inaccuracies on the fact that the modeller himself was an outsider. The analysis in this paper has shown that exaggerating the importance of such criticism is to miss more important aspects of the issues of modelling that go beyond the identity of the modeller, to the issue of 'naive planning'. Drawing on the concept of naive sociology (Wynne, 1989) naivety in planning is illustrated when the analytical findings of the process make sense only in light of assumptions which are largely closed off to closer empirical examination. An example of this is how the data requirements for indicators are underappreciated and insufficiently nuanced. For instance, the indicator of smallholder income is profoundly affected by external supports such as technical assistance and infrastructure, but these are not (and cannot be) considered in projected smallholder incomes because they are neither quantitative nor are they likely to have a proportionate effect on income.
From modelling landscapes to 'model' landscapes
This brings us to the question of whether or not a model can ever be 'purely factual'. Modelling enterprises do not produce neutral snapshots of a landscape (Harvey and Chrisman, 1998; Robbins, 2003b). They provide representations of a landscape that correspond with particular views of the world, priorities and values. Contrary to the conventional view of the map as an objective representation, the well-known critical geographer/cartographer Brian Harley's careful analyses reveal the 'textuality of maps, including their metaphorical and rhetorical nature... (and) the dimensions both of external power and of the omnipresence of internal power in the cartographic representation of place' (Harley, 1989: 1). The ALCES program is nonspatial; it does not produce spatially explicit depictions of outcomes. Rather, outcomes are predicted at an aggregated, landscape level. However, despite the absence of a map, a landscape is still produced by the ALCES analyses. And the 'textuality' of this landscape is clear: rapidly disappearing forests, are eaten away by unsustainable small producers as a benevolent NGO fights to save forests and the cultivation techniques of large producers set the sustainable example for all. Blame for land degradation and deforestation lands squarely on the smallholder producers, and their rationality is further compromised by the 'ALCES-generated fact' that if they don't change their behaviour, they confound their own well-being and even assure their own demise. Simplicity and aggregation fulfil the requirements of ALCES, and make scenarios useful and operational, to enable managers to understand and intervene in clear ways. Such rewriting of physical and social landscapes, described by Scott as 'legibility', persists and is insisted upon, despite that 'the data from which such simplifications arise are, to varying degrees, riddled with inaccuracies, omissions, faulty aggregations, fraud, negligence, political distortion and so on' (Scott, 1998: 49).
Models: Use them or lose them?
It is unsurprising that the matter of whether, and to what extent, modelling should be used in a participatory context is a contentious one. But what does the present analysis mean for the role of models in environmental governance? Are planners held hostage to either using models and surrendering deliberation on the one hand, or surrendering models for the sake of salvaging deliberation on the other? Critical advocates suggest there is a middle ground, where 'complex simulations are no longer touted as predictive models but as heuristic devices to explore the logical implications of certain assumptions' (Peters, 1991: 116). Indeed others seem to agree that models can be useful, but their limitations must be fully recognised by stakeholders before they can be implemented into the planning process without usurping public participation and stakeholder debate altogether. For example, Van der Sluijs promotes the incorporation of 'uncertainty management' with the use of models that emphasise full awareness of 'the limitations and caveats' (Van der Sluijs, 2001: 327). Similarly, the present research suggests that a more explicit engagement of co-produced facts and norms in models may harness their usefulness, at the same time as opening the claims based on them to greater deliberative debate among stakeholders.
Working with models necessitates use of a technical discourse of indicators, outcomes and evidence-based relationships. The expertise used to generate scenarios, the very nature of the computer program itself, and the specific policy recommendations to which the scenarios give rise, often overlook more normative, political aspects of the sustainability problem. This leads to dramatic simplifications and the direction of blame for ecological degradation toward those whose concerns are not, and cannot be, integrated into the simulations. But this is not to say that the models are factual as opposed to normative. Indeed, as we have seen, the facts to which the model gives rise cannot be separated from normative positions about what assumptions are acceptable, how indicators can and should be constructed and what policy recommendations are reasonable. The analysis showed how models give preference to indicators that express: short-term, direct and quantifiable relationships between land use and outcomes; certain indicator dimensions over others (for example, conservation over access to resources); over-simplified data requirements; data availability that may be socially and politically determined; and an over-emphasis on certain livelihood dimensions over others.
Three main conclusions may be drawn from this analysis. The first is that public deliberation can provide insights into sustainability dynamics that often are not, and cannot be, captured by conventional, 'expert-led' indicators, because of technical requirements, data availability and outcome priorities. Secondly, public deliberation can effectively challenge the political and empirical legitimacy of expert-led, model-based indicators by opening for debate the ways and extent to which they are considered factual. Deliberative insights can point to model-based indicators as context dependent, socially and politically conditioned, and thus contingent and contestable as opposed to factual and necessarily accepted by all stakeholders. Thirdly, despite these contributions, public deliberation about model-based land-use planning is often shaped and constrained by what is considered evidence. Emergent norms are often considered rational policy inputs only insofar as they are compatible with that which is presented as the evidence. This convention is fundamentally challenged by a co-productionist account of facts and norms in indicators.
These findings become increasingly important as 'new' modes of governance based on participation and deliberation are promoted, but often not reflected in environmental decision making(Dressler et al., 2010, Fischer, 2009). Even in the context of an emphasis on participation, the ALCES model was treated unproblematically as the 'best' approach to defining and solving environmental problems. A recognition of the normative commitments of facts, however, has the potential to reposition participation and deliberation as a more legitimate and credible contribution to environmental governance and policy debates--contributions to be taken seriously.
Altieri, M.and M. Pengue. 2005. Roundup Ready Soybean in Latin America: A Machine of Hunger, Deforestation and Socio-Ecological Devastation. Biosafety Information Centre.
Backstrand, K. 2003. 'Civic science for sustainability: Reframing the role of experts, policy-makers and citizens in environmental governance'.Global Environmental Politics 3: 24-41. CrossRef
Bickel, U. and J. M. Dros. 2003. The Impacts of Soybean Cultivation on Brazilian Ecosystems. Washington: World Wildlife Fund.
Carlson, M. 2006. Planificacion Del Uso De La Tierra: Comparando Posibles Escenarios Para La Reserva De Biosfera Del Bosque Mbaracayu. Edmonton: Alberta Research Council.
Cinderby, S. 1999. 'Geographic information systems (GIS) for participation: The future of environmental GIS?' International Journal of Environment and Pollution 11: 304-315. CrossRef
Cornwall, A.and K. Brock. 2005. 'What do buzzwords do for development policy? A critical look at "participation", "empowerment" and "poverty reduction"'. Third World Quarterly 26: 1043-1060. CrossRef
Craig, W. J., T. M. Harris and D. Weiner. 2002. Community Participation and Geographic Information Systems. London: Taylor and Francis.
Demeritt, D. 2001. 'The construction of global warming and the politics of science'. Annals of the Association of American Geographers 91: 307-337. CrossRef
Dressler, W., B. Buscher, M. Schoon, D. Brockington, T. Hayes, C. A. Kull, J. McCarthy and K. Shrestha. 2010. 'From hope to crisis and back again? A critical history of the global CBNRM narrative'. Environmental Conservation 37: 5-15. CrossRef
Dros, J. M. 2004. Managing the Soy Boom: Two Scenarios of Soy Production. Amsterdam: AIDEnvironment.
Fischer, F. 2009. Democracy and Expertise: Reorienting Policy Inquiry. New York: Oxford University Press. CrossRef
Florentin, M. A., M. Peoalva, A. Calegari and R. Derpsch. 2001. Abonos Verdes Y Rotacion De Cultivos En Siembra Directa. Pequenas Propiedades. Proyecto Conservacion De Suelos MAG-GTZ, DEAG San Lorenzo, Paraguay. MAG-GTZ, DEAG.
Fogel, R. and M. Riquelme. 2005. 'Introduction'. In Enclave Sojero: Merma De Soberania Y Pobreza, R. Fogel and M. Riquelme (eds.), pp.9-14. Asuncion: Centro de Estudios Rurales Interdisciplinarios.
Forsyth, T. 1999. 'Environmental activism and the construction of risk: Implications for NGO alliances'. Journal of International Development 11: 687-700. CrossRef
Forsyth, T. 2004. 'Social movements and environmental democratization in Thailand'. In S. Jasanoff and M. L. Martello (eds.), Earthly Politics: Local and Global in Environmental Governance, pp.195-215. Cambridge: MIT Press.
Fundacion Moises Bertoni (FMB). 2005. Management Plan for the Mbaracayu Reserve. Asuncion: Fundacion Moises Bertoni.
Garb, Y., S. Pulver and S. D. VanDeveer. 2008. 'Scenarios in society, society in scenarios: Toward a social scientific analysis of storyline-driven environmental modeling'. Environmental Research Letters 3: 1-8. CrossRef
Garcia-Lopez, G. A. and N. Arizpe. 2010. 'Participatory processes in the soy conflicts in Paraguay and Argentina'. Ecological Economics 70: 196-206. CrossRef
Gasteyer, S. and F. C. Butler. 2000. 'Measuring ppm with tennis shoes: Science and locally meaningful indicators of environmental quality'. Society and Natural Resources 13: 589-597. CrossRef
Harley, J. B. 1989. 'Deconstructing the map'. Cartographica: The International Journal for Geographic Information and Geovisualization 26: 1-20. CrossRef
Harvey, F. and N. Chrisman. 1998. 'Boundary objects and the social construction of GIS technology'. Environment and Planning A 30: 1683-1694. CrossRef
Hill, K. and J. Padwe. 2000. 'Sustainability of Ache hunting in the Mbaracayu Reserve, Paraguay'. In J. G. Robinson and E. L. Bennett (eds.), Hunting for Sustainability in Tropical Forests, pp.79-105. New York: Columbia University Press.
Huang, C., S. Kim, K. Song, J. R. G. Townshend, P. Davis, A. Altstatt, O. Rodas, A. Yanosky, R. Clay, C. J. Tucker and J. Musinsky. 2009. 'Assessment of Paraguay's forest cover change using Landsat observations'. Global and Planetary Change 67: 1-12. CrossRef
Jasanoff, S. 2004a. 'The idiom of co-production'. In S. Jasanoff (ed.), States of Knowledge: The Co-Production of Science and Social Order, pp.1-12. London: Routledge. CrossRef
Jasanoff, S. 2004b. 'Ordering knowledge, ordering society. In S. Jasanoff (ed.), States of Knowledge: The Co-Production of Science and Social Order. New York: Routledge. CrossRef
Jasanoff, S. and B. Wynne. 1998. 'Science and decisionmaking'. In S. Rayner and E. Malone (eds.), Human Choice and Climate Change, Vol.1: The Societal Framework, pp.1-87. Batelle.
Lange, D. 2005. Economics and Evolution of Smallholdings Conservation Agriculture in Paraguay: Mid-Term Experiences. Asuncion, Paraguay: FAO GTZ.
Lazarus, J. 2008. 'Participation in poverty reduction strategy papers: Reviewing the past, assessing the present and predicting the future'. Third World Quarterly 29: 1205-1221. CrossRef
Litfin, K. 1999. 'Environmental remote sensing, global governance, and the territorial state'. In M. Hewson and T. J. Sinclair (eds.), Approaches to Global Governance Theory, pp.73-96. New York: State University of New York Press.
Miller, C. A. 2004. 'Climate science and the making of a global political order'. In S. Jasanoff (ed.), States of Knowledge: The Co-Production of Science and Social Order, pp.46-66. New York: Routledge.
Mitchell, R. E. and C. Davis. 2005. Social Indicators in Cumulative Landuse Effects Modelling. Edmonton: Alberta Research Council.
Naidoo, R. and W. Adamowicz. 2006. 'Modeling opportunity costs of conservation in transitional landscapes'. Conservation Biology 20: 490-500. CrossRef
Naidoo, R. and K. Hill. 2006. 'Emergence of indigenous vegetation classifications through integration of traditional ecological knowledge and remote sensing analyses'. Environmental Management 38: 377-387. CrossRef
Nurick, R. and V. Johnson. 1998. 'Towards community based indicators for monitoring quality of life and the impact of industry in South Durban'. Environment and Urbanization 10: 233-250. CrossRef
Parlee, B. 1998. A guide to community-based monitoring for northern communities. In Northern Minerals Program Working Paper No. 5. Canadian Arctic Resources Committee.
Peters, R. H. 1991. A Critique for Ecology. Cambridge: Cambridge University Press. Pickles, J. (ed.) 1995. Ground Truth: The Social Implications of Geographic Information Systems. New York and London: The Guilford Press.
Robbins, P. 2003a. 'Beyond ground truth: GIS and the environmental knowledge of herders, professional foresters, and other traditional communities'. Human Ecology 31: 233-253. CrossRef
Robbins, P. 2003b. 'Fixed categories in a portable landscape: The causes and consequences of land cover categorization'. In K. S. Zimmerer and T. J. Basset (eds.), Political Ecology: An Integrative Approach to Geography and Environment-Development Studies, pp.181-200. New York and London: The Guilford Press.
Schneider, R. R., J. B. Stelfox, S. Boutin and S. Wasel. 2003. 'Managing the cumulative impacts of land uses in the western Canadian sedimentary basin: A modeling approach'. Conservation Ecology 7: Article 8.
Scott, J. C. 1998. Seeing Like a State: How Certain Schemes to Improve the Human Condition Have Failed. New Haven and London: Yale University Press.
Sorrenson, W. J. 1997. Financial and Economic Implications of No-tillage and Crop Rotations Compared to Conventional Cropping Systems. In TCI Occasional Paper Series. Rome: FAO.
Steward, C. 2007. 'From colonization to "environmental soy": A case study of environmental and socio-economic valuation in the Amazon soy frontier'. Agriculture and Human Values 24: 107-122. CrossRef
Taylor, P. 1992. 'Reconstructing socioecologies: Systems dynamics modeling of nomadic pastoralists in Sub-Saharan Africa'. In A. Clark and J. Fujimura (eds.), The Right Tools for the Job. London: Princeton University Press.
Turnhout, E., M. Hisschemoller and H. Eijsackers. 2007. 'Ecological indicators: Between the two fires of science and policy'. Ecological Indicators 7: 215-228.
UNDP. 2003. Human Development Report, 2003. New York: UNDP.
Van der Sluijs, J. P. 2001. 'Integrated assessment modeling and the participatory challenge: The case of climate change'. In M. Hisschemoller, R. Hoppe, W. N. Dunn and J. R. Ravetz (eds.), Knowledge, Power, and Participation in Environmental Policy Analysis, pp.317-347. New Brunswick and London: Transaction Publishers.
Van der Sluijs, J. P. 2002. 'A way out of the credibility crisis of models used in integrated environmental assessment'. Futures 34: 133-146. CrossRef
Wynne, B. 1989. 'Frameworks of rationality in risk management: Towards the testing of naive sociology'. In J. Brown (ed.), Environmental Threats: Perception, Analysis and Management, pp.33-47. London: Belhaven.
Wynne, B. 1996. 'May the sheep safely graze? A reflexive view of the expert-lay knowledge divide'. In S. Lash, B. Szerszynski and B. Wynne (eds.), Risk, Environment and Modernity: Towards a New Ecology, pp.44-83. London: Sage Publications Inc.
Yearley, S. 1999. 'Computer models and the public's understanding of science: A case study analysis'. Social Studies of Science 29: 845-66. CrossRef
Yearley, S., S. Cinderby, J. Forrester, P. Bailey and P. Rosen. 2003. 'Participatory modelling and the local governance of the politics of UK air pollution: A three-city case study'. Environmental Values 12: 247-262. CrossRef
Social Science and Policy Studies
Worcester Polytechnic Institute
Salisbury Laboratories 100 Institute Road Worcester, MA 01609-2280, USA
BOX 1. Focus Groups and Deliberative Community-based Indicators (Source: Author)
Between August and October, 2005, I conducted 2 workshops in each of 8 CARJ communities in order to establish community-based indicators along with assistant facilitators, in a deliberative context. Each focus group began with a presentation to introduce the concept of indicators, particularly, community-based ones, with a linkage made to the notion of well-being. I explained to participants that I was seeking to establish community-based indicators so that local opinions, priorities and knowledge would be captured in monitoring and evaluating watershed changes. I asked the participants to think in a general sense about what well-being meant for them. Comments were captured on a large sheet at the front of the room, or on the floor, depending on where the workshop was being held. As concepts emerged, the facilitators used probes to extend and deepen the dialogue so as to arrive at specific indicators.
When participants were satisfied that all of their important points had been documented, I posted the list of indicators and each participant was given five 'votes' and asked to stick round stickers next to five indicators considered priorities. The participants were encouraged to use more than one of their votes to emphasise any indicator they considered as very high priority. The number of votes for each indicator was tallied at each focus group to determine the priority ranking of that indicator.
(1.) The scenarios that were presented were preliminary analyses, and this was made explicit at the outset. Further analysis was awaiting feedback and input from stakeholders, and identification of additional potential data sources. The report was published by the ALCES group in 2007, and can be found on the ALCES website www.ALCES.ca.
(2.) This analysis is based on the ALCES scenarios generated for discussion with Paraguayan stakeholders in 2006. Based on this ALCES work, a Scenario Analysis was subsequently published. In this publication, an additional indicator--carbon storage--and an additional scenario--sustainable forestry as defined by sustained yield timber production--were modelled.
TABLE 1. Data required, sources of data and source details for selected ALCES indicators (Source: Author, based on information from Carlson, 2006; 2007) Indicator Data Required Source Study Net Agricultural Income--conventional (Sorrenson, 1997) Income--Soy and conservation agriculture Inputs (Bickel and Dros, 2003) Rate of expansion (Dros, 2004) Net Agricultural Productivity (Florentin et al., Income--Small 2001) Producers Inputs and other (Lange, 2005) costs Rate of expansion (UNDP, 2003) Natural Habitat Land cover of (FMB, 2005) protected area Rate of soy (Dros, 2004) expansion Rate of smallholder (UNDP, 2003) expansion Probability of (Naidoo and conversion by Adamowicz, 2006) landscape type Indicator Source Details Net Agricultural Over 10 years--net income under Income--Soy conventional and conservation agriculture on 135 Ha farms in San Pedro and Itapua Litres/Hectare pesticide application Projections of the government and soy industry Net Agricultural 20 year crop productivity--small Income--Small farmers in San Pedro Producers Production cost and crop price--from 2-7 farms in San Pedro and Edelira in 1998 and 2003, for each of corn, cotton and mandioca Includes pesticide inputs Based on population growth Natural Habitat Based on GIS mapping and measurement Based on growth estimates--industry and government Based on population expansion Based on historical patterns of expansion TABLE 2: Community-based Indicators from Focus Groups (Source: Author) Theme Indicator examples Agricultural production Cash crops Subsistence crops Commercialisation Quantity of products commercialised Sources of support for commercialisation (those known and those accessible) Culture Level of use of the Guarani language Participation in religious events Education Levels of formal education Accessibility of education (cost/location) Literacy rates Opportunities to continue education beyond basic levels Opportunities for training/work Forest cover Forest products available for use Satisfaction with availability of forest products Health Accessibility of medications through social or private pharmacies Accessibility of a health centre Income Cash crops Employment income Distribution of income by household Infrastructure Quantity and accessibility of means of transportation (for produce and people) Property rights Incidence of land title Quantity of land owned by producer households Quantity of land used per household Technical Number and type of workshops assistance for men and women Water Quantity of available water sources Quality of available water sources Existence/sufficiency of riparian zones Accessibility/distance from water sources households Theme Priority ranking Agricultural production 1 1 Commercialisation 1 1 Culture 4 4 Education 2 2 1 2 2 2 Forest cover 1 1 Health 2 3 Income 1 2 1 Infrastructure 2 Property rights 4 4 1 Technical assistance 3 Water 1 1 2 1
|Printer friendly Cite/link Email Feedback|
|Date:||Dec 1, 2013|
|Previous Article:||Sharing the earth: Sustainability and the currency of inter-generational environmental justice.|
|Next Article:||The Rediscovery of the Wild.|