In the 1920s, the amount of forest cover in the United States stabilized after many decades of declines, but forest fragmentation has been ongoing since the early 1900s (Sampson and DeCoster 2000, Best 2002). Also, the rate and extent of parcelization have increased in recent decades (DeCoster 1998, Best 2002). Parcelization is the division of a larger tract with a single owner into multiple, smaller parcels with multiple owners (Best 2002, Ko and He 2011). As a result of parcelization, the average parcel size decreases and the number of landowners increases (Kendra and Hull 2005, Ko and He 2011). During parcelization, forests are frequently fragmented (Best 2002). Fragmentation is the division of contiguous forest into discrete patches. These smaller patches often exhibit greater isolation, less interior habitat, and fewer ecosystem services (Groom et al. 1999, Best 2002). Parcelization can also change local economies (Harper et al. 1990). Smaller parcels may not be economically viable for timber production because of the economies of scale (Mehmood and Zhang 2001), so regional wood supplies may decrease (Wear et al. 1999), landowners may not be able to depend on this traditional source of income, and further parcelization may result (Ko and He 2011). Parcelization can also lead to further development and the conversion of previously forested land into more intense land uses, particularly residential subdivisions (Mehmood and Zhang 2001, Best 2002, Gobster and Rickenbach 2004). Finally, parcelization is associated with changes in social dynamics (Rickenback and Gobster 2003). As the number and density of landowners increase, the community experiences more diverse objectives and values (Egan and Luloff 2000, Smith and Krannich 2000, Mehmood and Zhang 2001). As the community changes, residents may experience a loss of community identity and sense of place (Cumming and Norwood 2012).
To manage parcelization, it is important to understand how owners of large, forested properties make decisions and to understand the objectives behind their decisions. If reducing parcelization is desired, ways to keep large properties intact that compliment landowners’ objectives should be identified (Best 2002). Previous studies have found that landowners may consider parcelization because of the expense of taxes, because they can make a profit when urbanization of rural areas leads to property value increases, or when they inherit property but lack the means or interest to manage it (DeCoster 1998, Mehmood and Zhang 2001, Best 2002). The roles of finances and maintenance in parcelization are evident in the literature. If associated objectives can be met, landowners may be less likely to parcelize. Because many modern forestland owners have diverse objectives and may face increasing pressure to parcelize, landowners and forest management professionals may benefit from new approaches to forest management decision making (Kendra and Hull 2005).
We propose structured decision making (SDM) as a process that can enhance the sustainability of private forests by helping landowners identify decision options that are most likely to result in outcomes that meet objectives related to forest sustainability. The SDM procedure accomplishes this through (1) identifying landowners’ multiple objectives and their relative importance and (2) modeling the probability of different outcomes following each decision option.
SDM is based on decision analysis, the use of quantitative methods to evaluate decision options (Keeney 1992, Clemen 1996, Gregory et al. 2012, Conroy and Peterson 2013). The SDM approach is beneficial because it balances multiple objectives given constraints and uncertainty, and it recognizes the distinction between value-based information and technical information while explicitly integrating both (Gregory and Keeney 2002, Wilson and McDaniels 2007, Conroy et al. 2008). In the SDM process, the objectives and values of decision makers are included in analyses, but there is no attempt to change decision makers’ objectives or values. Decisions made through a process that explicitly defines objectives, weights conflicting objectives, and incorporates uncertainty are expected to produce desirable outcomes more often than decisions made less systematically (Conroy and Peterson 2013).
The main components of SDM are a definition of the decision problem, objectives based on values, attributes to make objectives measurable, decision options that could help achieve objectives, one or more models to describe the expected outcomes of decisions, and a method to evaluate the degree to which each decision is expected to fulfill objectives (Conroy et al. 2008, Irwin et al. 2011). These components are developed through an iterative process in which stakeholders provide input and facilitators synthesize information while attempting to remain value neutral (Miller et al. 2010, Raymond et al. 2010, Irwin et al. 2011). We conducted a SDM project focused on the management of large private forests in Macon County, North Carolina, USA, that involved a series of workshops with forestland owners.
We focused on forestland owners in Macon County, North Carolina, USA, because Macon County has experienced high rates of residential development, there is a history of conflict over land use regulations, and the region is one of the most biodiverse in North America (Barnes 1991, Gragson and Bolstad 2006). The growth in Macon County has been associated with the amenity-driven development seen throughout the southern Appalachian region. During the 1990s, this region experienced 18% growth in population (Pollard 2005).
Many Macon County landowners think rapid growth is detrimental, but there has not been agreement about an appropriate response (Gragson and Bolstad 2006, Cho et al. 2009, Cumming and Norwood 2012). There have been attempts to pass land use regulations in Macon County throughout the past 30 years, but they have largely failed (Cumming and Norwood 2012). Consequently, many land use decisions are made by individual landowners. Macon County is not unique in experiencing confrontational and eventually stalled land use decision making, and this phenomenon has been attributed to the lack of effective opportunities for citizens to express their perspectives, consider potential options, and learn from each other in a respectful and productive setting (Lando 2003, Stewart et al. 2004, Cumming and Norwood 2012).
The purpose of this study is to illustrate how SDM can assist landowners with land use decision making. This is important to demonstrate for a community where attempts at land use regulation have been unproductive and because SDM has a more extensive history of application to decisions that involve multiple stakeholders and a common resource. We describe the SDM process with landowners and discuss advantages and challenges of using SDM in this context. In Macon County, we applied SDM to the question of forest management on large parcels (30-ha property with 22 ha of forest), and we present the forest management options that were identified as optimal and least desirable. Additionally, we consider landowners’ perception of the SDM process and the potential for land trusts and land use planners to benefit from SDM.
The goal of the SDM project in Macon County was to address the question “What can you do to your forest to maximize the achievement of your land use objectives?” with owners of large, forested properties (at least 20 ha in total area with at least 4 ha of forest). Throughout the project, we asked landowners for their personal perspectives, but the analysis was not intended to apply to a specific property. Rather, we evaluated decision options for an average, large, forested property in Macon County using multiple scenarios of landowner values. Specifically, we modeled outcomes for a 30-ha property at 750-m elevation with 22 ha of forest, approximately the mean characteristics of properties owned by the workshop participants as determined from county records and aerial photographs. According to experts at Forest Stewards (a nonprofit corporation affiliated with Western Carolina University that promotes and implements sustainable forest management in the southern Appalachians) who have worked in Macon County, an average, large, private forest in Macon County is about 60 years old, and a timber harvest could be conducted in 10-30 years. We considered a 30-year time frame for our decision analysis because one timber harvest could occur and landownership turnover is likely after 30 years, especially because more than 25% of the Macon County population is over 65 years of age (U.S. Census Bureau 2013).
Typically when SDM is used to address public resource management, stakeholders representing the full range of interests related to the decision are invited to participate. Therefore, we sought to include landowners with diverse socio-demographic backgrounds and property characteristics, because these traits were expected to relate to land use values. Further, many scientists who have conducted social-ecological research in Macon County through the Coweeta Long Term Ecological Research (LTER) program hypothesized that multigenerational landowners and new residents have different land use values and practices.
To identify landowners for the SDM workshops, we interviewed 50 owners of large, forested properties (at least 20 ha in total area with at least 4 ha of forest) in Macon County (Institutional Review Board project number 2012108313). Interviewees were identified through a combination of snowball and random stratified sampling (Bernard 2002). Twenty landowners expressed interest in participating in the SDM workshops, and because this was a feasible size for workshops, all were invited. We scheduled 2 series of workshops and assigned 10 landowners to each series such that landowner diversity within a series was maximized (Table 1). The two series were independent because landowner composition remained constant within a series. Each series consisted of four workshops, and all workshops were moderated by P.F. in the conference room at the U.S. Forest Service’s Coweeta Hydrologic Laboratory in Otto, North Carolina. Three workshops were held in the summer of 2012, one workshop was held in the summer of 2013, and they lasted about three hours each.
The goal of the first workshop was for landowners to identify their land use objectives (Keeney 1992). Based on the landowners’ comments, we constructed an objectives network, a diagram distinguishing fundamental and means objectives (McDaniels 2000). Fundamental objectives represent the primary values that are inherently important to the decision maker, whereas means objectives highlight the path to achieving fundamental objectives (Keeney 1992, Clemen 1996). We identified some fundamental objectives that were composed of first-order and second-order fundamental objectives. Second-order fundamental objectives describe the components of a first-order fundamental objective that are important to a landowner.
The goals of the second workshop were to identify attributes to make the fundamental objectives measurable, brainstorm decision options, and begin to construct an influence diagram. Attributes provide the scales to measure the degree to which outcomes from a decision satisfy fundamental objectives (Failing et al. 2007, Wilson and McDaniels 2007, Gregory and Long 2009). When there was no natural scale (e.g., hectares) for a fundamental objective, the landowners created a scale with explicitly defined levels through consensus-based discussion (Keeney and Gregory 2005, Miller et al. 2010).
Determining fundamental objectives at the beginning of the decision-making process facilitates identifying creative decision options (McDaniels 2000). Landowners determined decision options through consensus-based discussion (Miller et al. 2010).
The objectives network provided the framework for an influence diagram (Marcot et al. 2001, 2006). An influence diagram consists of nodes, which represent variables, connecting the decision options to fundamental objectives, and arrows, which represent causal links between variables. Each node can take one of multiple discrete states (Marcot et al. 2001). Landowners discussed what nodes should be included and how to connect nodes so that the influence diagram realistically described how forest management decisions affect fundamental objectives. The influence diagram, and later the Bayesian decision network, was built in Netica 4.09 (Norsys Software Corp., Vancouver, British Columbia, Canada).
The goals of the third workshop were to identify the landowners’ objective weights and attribute scores. The influence diagram provides the structure for a Bayesian decision network, a model that predicts expected outcomes for each decision option and assesses how well the expected outcomes satisfy fundamental objectives (Conroy et al. 2008, Miller et al. 2010, Irwin et al. 2011). To analyze the decision options in a Bayesian decision network, objective weights and attribute scores were required. An objective weight reflects the relative importance of the objective to the landowner, with a larger weight indicating greater importance. An attribute score reflects how satisfied a landowner would be if that level in the attribute scale occurred. Landowners completed worksheets to identify their objective weights, using the swing weighting method, and to assign attribute scores (Clemen 1996; Appendices 1 and 2).
Between the third and fourth workshops, we identified conditional probabilities for the decision network, calculated expected utility values for each decision option, and compared expected utility values to determine a decision recommendation. Conditional probabilities describe the likelihood that each level in a node is realized given states of influencing nodes (Oliver and Smith 1990, Marcot et al. 2001). We searched the scientific literature for relevant studies from which probabilities of outcomes could be obtained. However, we found that papers rarely presented results in this form. Therefore, we used the available scientific literature to develop hypotheses about system dynamics, but we relied on expert opinion for probabilities (Haas 1991, 2001, Clemen 1996, Peterson and Evans 2003). See Appendix 3 for a discussion of using expert opinion in SDM. We sent worksheets to 33 experts to elicit conditional probabilities. These experts consisted of faculty at the University of Georgia (UGA), faculty at other institutions who are affiliated with Coweeta LTER, graduate students at UGA who had conducted research at Coweeta, U.S. Forest Service employees at the Coweeta Hydrologic Laboratory, a Macon County government employee, staff from the Land Trust for the Little Tenneessee, which is based in Macon County, and staff from Forest Stewards. When we received probabilities for a node from more than one expert, we made a new node for expert identity that affected the nature node. Through the expert identity node, we weighted each expert’s probabilities equally, reflecting equal belief in each expert’s contribution. Landowners provided conditional probabilities related to heritage, a topic on which they were the best qualified experts, through consensus-based discussion. Through the use of probabilities, we incorporated environmental stochasticity and partial controllability in the predictions of outcomes following decisions (Williams et al. 1996, Conroy et al. 2008, Irwin et al. 2011).
Utility functions combine the probability of outcomes and landowners’ satisfaction with outcomes such that the expected utility value indicates the relative suitability of the decision option. Expected utility values were calculated for each decision by a weighted average of the objective weights, attribute scores, and conditional probabilities (Peterson and Evans 2003). All combinations of objective weights and attribute scores, generating scenarios of landowner values, were used to calculate expected utility values. The sets of optimal and least-desirable decision options were defined as decision options with utility values within one point of the highest or lowest utility value, respectively. The frequency with which each decision option fell in the set of optimal or least-desirable decision options was recorded.
We calculated the expected utility according to equation (1) where W indicates a first-order fundamental objective weight, U indicates a second-order fundamental objective weight, and S represents an attribute score. For each of the G second-order fundamental objectives (s = 1, 2, ..., G) within a first-order fundamental objective (p = 1, 2, ..., F), we weighted the attribute score for a possible outcome (v = 1, 2, ..., H) by the probability of that outcome (Xv) given states of influencing nodes (A). Note that G may depend on p and H may depend on p and s.
At the fourth workshop, we presented the completed decision network and discussed the optimal and least-desirable decision options. We also asked landowners to complete questionnaires addressing their experience with the SDM project (Appendix 4). One questionnaire was distributed before we presented results and one was distributed after.
A small set of fundamental objectives were identified by the landowners: maximizing forest health, safety, heritage preservation, and net income, but the landowners in series 2 included maximizing aesthetic enjoyment. For some of these fundamental objectives, landowners also defined second-order fundamental objectives, which described components of a first-order fundamental objective while remaining fundamental objectives themselves (Fig. 1).
Because objective weights and attribute scores are based on values, we did not consider responses from landowners right or wrong, but they had to be logically consistent; outcomes that were considered better needed a higher attribute score. We found many inconsistencies in the worksheets used to elicit objective weights and attribute scores, so we eliminated those responses and collated all of the logically consistent responses across all landowners. If a landowner completed all components of the worksheet correctly, their responses were used to calculate a set of objective weights or attribute scores (Appendices 5-6). The consistent responses from landowners who did not complete the entire worksheet correctly were used to calculate a set of mean objective weights or mean attribute scores. Therefore, for series 1 there were 4 scenarios of objective weights and attribute scores, and for series 2 there were 24 scenarios. The expected utilities of the decision options were compared under each scenario.
The landowners identified 11 decision options: no modification of the forest, personal use of the forest (e.g., collecting firewood, building and using recreational trails), crown-thinning harvest through the Present-Use Value (PUV) program, group selection harvest through the PUV program, shelterwood harvest with residual trees through the PUV program, each of the above with a conservation easement, and sell 1 ha (approximately 5% of the forest) with personal use of the remaining forest (Figs. 2 and 3). Details about conservation easements and the PUV program can be found in Appendix 7.
Eight experts provided conditional probabilities (Appendix 8). On average, two experts contributed conditional probabilities for each node other than those for heritage or aesthetic objectives, for which landowners provided probabilities. See Appendices 9 and 10 for more details about the decision network.
The decision option with the largest or smallest utility value varied depending on the scenario of landowner values (Appendix 11). The decision most often included in the set of optimal decision options was crown thinning in the PUV program, and the consistently least-desirable decisions were selling 1 ha and personal use of the forest with or without a conservation easement (Table 2).
Under some landowner value scenarios, the optimal decisions were no modification of the forest with or without a conservation easement, shelterwood harvest in the PUV program with a conservation easement, and selling 1 ha. In certain landowner value scenarios, the least-desirable decisions were shelterwood harvest in the PUV program with or without a conservation easement and group selection harvest in the PUV program with a conservation easement.
The forest management practices that the decision network indicated were optimal were often different from those anticipated by the landowners. Only 33% of landowners correctly identified an optimal decision option, and 53% expected decision options to be optimal when the decision network actually identified them as the least-desirable decision option (Table 3). However, many landowners (67%) currently use the optimal practices according to the decision network (Table 3). Many landowners said they would consider management options other than those they currently use, and the SDM project influenced this opinion. After discussing the decision network results, the number of landowners who indicated that, in general, they would reconsider what they are currently doing to manage their forest increased by 46%. (Tables 3 and 4).
Eighty-five percent of landowners had a good experience participating in the project, and 69% said they understood most of the material presented during the project. All understood at least half of the material. Some aspects of the project that landowners found beneficial included “meeting others with similar interests in forest conservation,” “group discussions of individual management practices and what things participants value,” “objectively evaluating our property and values,” “watching the decision network grow,” and “encouragement to do something beneficial.”
A landowner’s efforts to avoid parcelization can be supported by a decision tool that includes the diverse factors contributing to a landowner’s decision to manage or subdivide a large, private forest (Best 2002). The importance of net income, heritage preservation, and aesthetics to landowners has been prominent in the literature and was discussed by Macon County landowners (Birch 1997, DeCoster 1998, Mehmood and Zhang 2001, Best 2002, Rickenback and Gobster 2003, Kendra and Hull 2005).
Each Macon County landowner held diverse objectives, and there was little variability in objectives among landowners. As opposed to the working hypothesis of many scientists at Coweeta LTER, multigenerational landowners and new residents did not appear to have different objectives. This unexpected pattern was also found by other researchers concurrently conducting social-ecological research in Macon County (S. Evans, personal communication).
Our findings are consistent with the notion behind SDM that stakeholders often do not have drastically different objectives; rather, they may assign objective weights and attribute scores differently (Keeney et al. 1990, Gregory and Keeney 1994). Conflict can arise in the decision-making process when this distinction is not recognized and stakeholders feel as if they have to defend their objectives. Instead, building models that incorporate multiple objective-weight and attribute-score scenarios can abate conflicts and facilitate decision making.
Our decision network produced reasonable results. It makes sense that crown thinning in the PUV program would be the optimal decision option because it causes a relatively low level of disturbance, landowners receive income from timber harvesting, and property taxes are reduced. Selling 1 ha and personal use of the forest with or without a conservation easement were the least-desirable decisions because although personal use of the forest causes a relatively low level of disturbance, there is no financial return. The property would be taxed at full market value, and establishing a conservation easement is expensive to the landowner. Although there may be financial benefits from selling property, it is detrimental to ecological and heritage objectives.
The robustness of results from the decision network influences how results should be interpreted and applied. In our project, landowners’ objective weights and attribute scores influenced which decision was optimal; and in some landowner value scenarios, there was little difference among the utility values for the various decision options. We did not find a single best management practice for large, private forests. When decision options have similar utility values, an optimal decision is less apparent, and it suggests that utility values may not be robust to objective weights, attribute scores, or probabilities. This emphasizes the need to tailor the decision network to an individual landowner and their property.
Use of expert opinion is an established practice in SDM applications when a decision must be made given the current best technical information, however incomplete (Appendix 3). We found that the conditional probabilities we obtained from experts were consistent with patterns indicated in the literature (Appendix 12). The effects of conditional probabilities on utility values can be investigated by varying the weights assigned to experts’ opinions. Because in our decision network there are 34,992 combinations (37*42 because there are 7 expert nodes with 3 possible combinations of weights and 2 expert nodes with 4 possible combinations of weights) in which weights might be uniform across the experts in a node or 100% on one expert in a node, we focused on the decision network from series 1 and landowner value scenario 1, and did not do an exhaustive analysis. When we varied the weights on experts’ opinions, the order of the top few decision options changed. This has implications for a goal of identifying the single decision with the greatest utility value, but decision networks do not need to be considered authoritative. They are meant to be decision support tools, so decision makers could identify the few decision options that most consistently have high utility values and engage in additional decision-making strategies to arrive at a final decision.
Because SDM can effectively integrate diverse objectives and multiple scientific models, and in the process reduce unproductive conflict, it has potential application to broad land use questions in Macon County and elsewhere. For a population that has been resistant to land use regulation, it is notable that 85% of the landowners had a good experience during the SDM project and 54% indicated that they might want a decision network made for their property. These results highlight a potential way to improve natural resource management, even in areas where past regulatory attempts have been unproductive.
For this project, our goal was to evaluate the potential utility of SDM to assist a landowner with a decision about their natural resources when the community has been resistant to regulation. Therefore, we held workshops including diverse landowners and analyzed a hypothetical property. This procedure allowed us to test SDM with a variety of landowners and minimize the appearance of being prescriptive about someone’s property. In actual application, an individual landowner’s objectives, decision options, and weights would be used in the decision network.
Besides individual landowners, conservation and land use planning organizations can benefit from using SDM to guide decisions about their operations or to support clients’ decision making. In some of these cases, it may be appropriate for a group of stakeholders to participate in a SDM process. A group SDM process is suitable when there are multiple stakeholders, such as conservation organization members, that share a common resource, such as the conservation organization’s operating procedures (Peterson and Evans 2003, Miller et al. 2010).
SDM also could be a way to make inroads into county-level land use or residential development decision making. There may be greater success in decision making when the process involves stakeholders throughout, explicitly incorporates diverse stakeholder values, addresses uncertainty, and is transparent. If county land use questions can be addressed in a SDM process that allows landowners to feel represented and respected, effectively integrates value-based and technical information, and avoids political tension, future decision making may be more successful than past attempts.
Some benefits of SDM that apply to both group and individual applications include (1) after systematic analysis realizing the optimal decision is different than the initially favored decision option and (2) identifying and reducing uncertainty in decision making. The first benefit was seen in this project: Landowners (40%) initially expected personal use to be the optimal decision, but the SDM analysis indicated that crown thinning in the PUV program was actually the optimal decision. The second benefit occurs because not all uncertainties have an equal effect on decision making. The most influential model components can be identified, and additional data can be collected to reduce uncertainty in key model components. Finally, SDM fits well with adaptive resource management. After a decision is made, the system is monitored, decision outcomes are compared with predicted outcomes from multiple models, uncertainty about system dynamics is reduced, and subsequent decisions incorporate this new information (Conroy and Peterson 2013).
However, there are challenges associated with SDM. There may be linguistic uncertainty and miscommunication. In our project, 69% of landowners said they understood most of the material. Nevertheless, there were many inconsistencies in the objective weights and attribute scores reported on the worksheets. We recommend that researchers assess each participant’s understanding and contributed information immediately and throughout the project, but this may be difficult given the researcher-to-participant ratio or time constraints. Additionally, obtaining conditional probabilities for the decision network can be challenging, given how results are often presented in the scientific literature and possibly limited transferability of results from previous studies to the SDM decision context. If expert opinion is the most suitable source of conditional probabilities, communication with consulted experts also may be challenging. Scientists often have not been trained in the distinction between and proper roles of value-based information and technical information (Failing et al. 2007).
Owners of large, private forests often have multiple ecological, economic, and social objectives motivating their land use decision making. A decision-making process, such as SDM, that helps landowners identify creative decision options and evaluate them in light of their objectives and uncertainty can help prevent parcelization and forest fragmentation. By participating in SDM, decision makers may benefit by reflecting on their values, learning technical information, and identifying decision options that are most likely to meet their objectives. Because SDM is participatory, transparently incorporates value-based and technical information, and includes uncertainty, it is an effective way to rigorously evaluate options for decision problems that are controversial or that have incomplete data. Our SDM project with owners of large, forested parcels in Macon County, North Carolina, found that crown thinning in the PUV program was the optimal forest management decision, and selling 1 ha of forest or personal use of the forest with or without a conservation easement were the least-desirable decisions for an average, large, forested property in Macon County. We have demonstrated that SDM can be effective in many challenging decision contexts.
We thank all of the Macon County landowners who participated in this project and colleagues who helped us meet landowners, including B. McRae, B. McLarney, C. Smalling, D. Shure, G. Wein, J. Love, J. Meador, J. Hunnicutt, K. McLean, P. Moore, R. Regnery, S. Guffey, and W. Swank. We appreciate feedback on this project from R. Cooper, J. Maerz, S. Evans, T. Gragson, N. Heynen, and the Coweeta Listening Project. Expert information was provided by D. Green, D. Desmond, J. Hatt, J. Frisch, K. Servidio, K. Cecala, M. Cline, R. Lamb, and T. Allen. Funding was provided by the Coweeta Long Term Ecological Research Project (NSF grant DEB-0823293), the USDA CSREES McIntire-Stennis Project (GEOZ-0159-MS), the Georgia Ornithological Society, the Warnell School of Forestry and Natural Resources at the University of Georgia, the University of Georgia Graduate School, and the Georgia Museum of Natural History.
Barnes, B. V. 1991. Deciduous forests of North America. Pages 219-344 in E. Röhrig and B. Ulrich, editors. Ecosystems of the world. Temperate deciduous forests. Volume 7. Elsevier, New York, New York, USA.
Bernard, H. R. 2002. Research methods in anthropology: qualitative and quantitative methods. AltaMira, Walnut Creek, California, USA.
Best, C. 2002. America’s private forests: challenges for conservation. Journal of Forestry 100(3):14-17.
Birch, T. W. 1997. Private forest-land owners of the southern United States, 1994. Resource Bulletin NE-138. U.S. Department of Agriculture Forest Service, Northeastern Forest Experiment Station, Washington, D.C., USA.
Cho, S.-H., S. G. Kim, R. K. Roberts, and S. Jung. 2009. Amenity values of spatial configurations of forest landscapes over space and time in the southern Appalacian Highlands. Ecological Economics 68:2646-2657. http://dx.doi.org/10.1016/j.ecolecon.2009.04.017
Clemen, R. T. 1996. Making hard decisions: an introduction to decision analysis. Duxbury, Pacific Grove, California, USA.
Conroy, M. J., and J. T. Peterson. 2013. Decision making in natural resource management: a structured, adaptive approach. Wiley-Blackwell, Hoboken, New Jersey, USA. http://dx.doi.org/10.1002/9781118506196
Conroy, M. J., R. J. Barker, P. W. Dillingham, D. Fletcher, A. M. Gormley, and I. M. Westbrooke. 2008. Application of decision theory to conservation management: recovery of Hector’s dolphin. Wildlife Research 35:93-102. http://dx.doi.org/10.1071/WR07147
Cumming, G., and C. Norwood. 2012. The community voice method: using participatory research and filmmaking to foster dialog about changing landscapes. Landscape and Urban Planning 105:434-444. http://dx.doi.org/10.1016/j.landurbplan.2012.01.018
DeCoster, L. A. 1998. The boom in forest owners—a bust for forestry? Journal of Forestry 96(5):25-28.
Egan, A. F., and A. E. Luloff. 2000. The exurbanization of America’s forests:research in rural social science. Journal of Forestry 98(3):26-30.
Failing, L., R. Gregory, and M. Harstone. 2007. Integrating science and local knowledge in environmental risk management: a decision-focused approach. Ecological Economics 64:47-60. http://dx.doi.org/10.1016/j.ecolecon.2007.03.010
Gobster, P. H., and M. G. Rickenbach. 2004. Private forest parcelization and development in Wisconsin’s Northwoods: perceptions of resource-oriented stakeholders. Landscape and Urban Planning 69:165-182. http://dx.doi.org/10.1016/j.landurbplan.2003.09.005
Gragson, T. L., and P. V. Bolstad. 2006. Land use legacies and the future of southern Appalachia. Society & Natural Resources 19:175-190. http://dx.doi.org/10.1080/08941920500394857
Gregory, R., L. Failing, M. Harsotne, G. Long, T. McDaniels, and D. Ohlson. 2012. Structured decision making: a practical guide to environmental management choices. Wiley-Blackwell, Hoboken, New Jersey, USA. http://dx.doi.org/10.1002/9781444398557
Gregory, R., and R. L. Keeney. 1994. Creating policy alternatives using stakeholder values. Management Science 40(8):1035-1048. http://dx.doi.org/10.1287/mnsc.40.8.1035
Gregory, R., and G. Long. 2009. Using structured decision making to help implement a precautionary approach to endangered species management. Risk Analysis 29(4):518-532. http://dx.doi.org/10.1111/j.1539-6924.2008.01182.x
Gregory, R. S., and R. L. Keeney. 2002. Making smarter environmental management decisions. Journal of the American Water Resources Association 38(6):1601-1612. http://dx.doi.org/10.1111/j.1752-1688.2002.tb04367.x
Groom, M., D. B. Jensen, R. L. Knight, S. Gatewood, L. Mills, D. Boyd-Heger, L. S. Mills, and M. E. Soule. 1999. Buffer zones: benefits and dangers of compatible stewardship. Pages 171-197 in M. E. Soule and J. Terborgh, editors. Continental conservation: scientific foundations of regional reserve networks. Island Press, Washington, D.C., USA.
Haas, T. C. 1991. A Bayesian belief network advisory system for aspen regeneration. Forest Science 37:627-654.
Haas, T. C. 2001. A web-based system for public-private sector collaborative ecosystem management. Stochastic Environmental Research and Risk Assessment 15:101-131. http://dx.doi.org/10.1007/s004770000062
Harper, S. C., L. L. Falk, and E. W. Rankin. 1990. The northern forest lands study of New England and New York. U.S. Department of Agriculture Forest Service, Rutland, Vermont, USA.
Irwin, B. J., M. J. Wilberg, M. L. Jones, and J. R. Bence. 2011. Applying structured decision making to recreational fisheries management. Fisheries 36(3):113-122. http://dx.doi.org/10.1080/03632415.2011.10389083
Keeney, R. L. 1992. Value-focused thinking: a path to creative decision making. Harvard University Press, Cambridge, Massachusetts, USA.
Keeney, R. L., and R. S. Gregory. 2005. Selecting attributes to measure the achievement of objectives. Operations Research 53(1):1-11. http://dx.doi.org/10.1287/opre.1040.0158
Keeney, R. L., D. van Winterfeldt, and T. Eppel. 1990. Eliciting public values for complex policy decisions. Management Science 36(9):1011-1030. http://dx.doi.org/10.1287/mnsc.36.9.1011
Kendra, A., and R. B. Hull. 2005. Motivations and behaviors of new forest owners in Virginia. Forest Science 51(2):142-154.
Ko, D. W., and H. S. He. 2011. Characterizing the historical process of private forestland ownership parcelization and aggregation in the Missouri Ozarks, USA, from 1930 to 2000. Landscape and Urban Planning 102:262-270. http://dx.doi.org/10.1016/j.landurbplan.2011.05.004
Lando, T. 2003. The public hearing process: a tool for citizen participation, or a path toward citizen alienation? National Civic Review 92(1):73-82. http://dx.doi.org/10.1002/ncr.7
Marcot, B. G., R. S. Holthausen, M. G. Raphael, M. M. Rowland, and M. J. Wisdom. 2001. Using Bayesian belief networks to evaluate fish and wildlife population viability under land management alternatives from an environmental impact statement. Forest Ecology and Management 153:29-42. http://dx.doi.org/10.1016/S0378-1127(01)00452-2
Marcot, B. G., J. D. Steventon, G. D. Sutherland, and R. K. McCann. 2006. Guidelines for developing and updating Bayesian belief networks applied to ecological modeling and conservation. Canadian Journal of Forest Research 36:3063-3074. http://dx.doi.org/10.1139/x06-135
McDaniels, T. L. 2000. Creating and using objectives for ecological risk assessment and management. Environmental Science & Policy 3:299-304. http://dx.doi.org/10.1016/s1462-9011(00)00108-8
Mehmood, S. R., and D. Zhang. 2001. Forest parcelization in the United States:a study of contributing factors. Journal of Forestry 99(4):30-34.
Miller, T. J., J. A. Blair, T. F. Ihde, R. M. Jones, D. H. Secor, and M. J. Wilberg. 2010. FishSmart: an innovative role for science in stakeholder-centered approaches to fisheries management. Fisheries 35(9):424-433. http://dx.doi.org/10.1577/1548-8446-35.9.422
Oliver, R. M., and J. Q. Smith. 1990.Influence diagrams, belief nets and decision analysis. John Wiley & Sons, New York, New York, USA.
Peterson, J. T., and J. W. Evans. 2003. Quantitative decision analysis for sports fisheries management. Fisheries 28(1):10-21. http://dx.doi.org/10.1577/1548-8446(2003)28[10:qdafsf]2.0.co;2
Pollard, K. M. 2005. Population growth and distribution in Appalachia: new realities. Demographic and Socioeconomic Change in Appalachia. Population Reference Bureau, Washington, D.C., USA.
Raymond, C. M., I. Fazey, M. S. Reed, L. C. Stringer, G. M. Robinson, and A. C. Evely. 2010. Integrating local and scientific knowledge for environmental management. Journal of Environmental Management 91:1766-1777. http://dx.doi.org/10.1016/j.jenvman.2010.03.023
Rickenback, M. G., and P. H. Gobster. 2003. Stakeholders’ perceptions of parcelization in Wisconsin’s Northwoods. Journal of Forestry 101(8):18-23.
Sampson, N., and L. DeCoster. 2000. Forest fragmentation:implications for sustainable private forests. Journal of Forestry 98(3):4-8.
Smith, M. D., and R. S. Krannich. 2000. “Culture clash” revisited: newcomer and longer-term residents’ attitudes towards land use, development, and environmental issues in rural communities in the Rocky Mountain West. Rural Sociology 65(3):396-421. http://dx.doi.org/10.1111/j.1549-0831.2000.tb00036.x
Stewart, W. P., D. Liebert, and K. W. Larkin. 2004. Community identities as visions for landscape change. Landscape and Urban Planning 69:315-334. http://dx.doi.org/10.1016/j.landurbplan.2003.07.005
U.S. Census Bureau. 2013. QuickFacts: Macon County, NC. U.S. Census Bureau, Washington, D.C., USA. [online] URL: http://www.census.gov/quickfacts/table/PST045214/37113,13193,00
Wear, D. N., R. Liu, J. Foreman, and R. M. Sheffield. 1999. The effects of population growth on timber management and inventories in Virginia. Forest Ecology and Management 118:107-115. http://dx.doi.org/10.1016/S0378-1127(98)00491-5
Williams, B. K., F. A. Johnson, and K. Wilkins. 1996. Uncertainty and the adaptive management of waterfowl harvests. Journal of Wildlife Management 60(2):223-232. http://dx.doi.org/10.2307/3802220
Wilson, C., and T. McDaniels. 2007. Structured decision-making to link climate change and sustainable development. Climate Policy 7:353-370. http://dx.doi.org/10.1080/14693062.2007.9685661