home

Decision Support Systems: New Tools for Data Users

Patrick Crist

National Gap Analysis Program, Moscow, Idaho

Putting digital data and information in the hands of users more efficiently and easily is the "hot" topic among data-producing agencies today. In October 1998, I presented the Gap Analysis Program’s (GAP) three-pronged approach to dissemination at the "Gateways to the Earth" workshop-an initiative of USGS. These three approaches are:

1. The GAP web site, which acts like a gateway of its own to the state projects which serve not just GAP data but, in many cases, a large collection of other state information and services.

2. CD-ROMs, which are currently GAP’s official and preferred method of data delivery to the public. While not discounting the growing importance of Internet delivery, I believe CDs currently offer a more efficient method of transmitting our very large data sets in a way guaranteed to provide the user with all documentation, including the final project report, needed for intelligent data use.

3. Decision support systems (DSS). GAP has recognized a need for biodiversity decision support systems in general, and DSS for GAP data in particular. This is because GAP data is novel in the decision-making realm-planners and managers have never before had data like GAP or a methodology for incorporating it into their decision-making processes. How they should accomplish this, though, is neither obvious nor simple.

There are many definitions of DSS, but essentially, they are systems that package information and tools that interpret that information in a form more readily integrated into decision-making processes (Zhu et al. 1998). Typically, but not always, they are comprised of a GIS for spatial data, databases of textual or tabular data, an interface designed for the user’s needs, as well as models, scripts, and programs that process the data according to the user’s specific criteria and queries.

In addition to the Gateways workshop which focused on an Internet-based interface to the breadth of USGS programs and information, I also attended two other workshops on DSS. In September, the Aurora Partnership workshop was sponsored by the Department of the Interior (DOI) and Environmental Systems Research Institute (ESRI). This was a meeting of over 100 representatives from federal agencies, private consulting and software firms, nonprofit groups, and academics from around the country interested in DSS. Its purpose was to demonstrate the state of DSS and exchange ideas on how to coordinate efforts and share knowledge. The goal of the partnership is to facilitate the intellectual and technological development of DSS. In October, the USGS Biological Resources Division likewise had a meeting to survey existing BRD DSS activities and discuss ways to strategically implement and coordinate DSS.

Jankowski et al. (1997 p. 578) identify three functions for a decision support system:

(1) help decision-makers formulate, frame, or assess decision situations by identifying the salient features of the environment, recognizing needs, [and] identifying appropriate objectives by which to measure the successful resolution of an issue; (2) provide support in enhancing the abilities of decision-makers to obtain and analyze possible impacts of alternative courses of action; and (3) enhance the ability of decision-makers to interpret impacts in terms of objectives, leading to an evaluation of alternatives and selection of a preferred option.

The goals of GAP’s pilot systems are less ambitious than the three functions described above, but their capabilities are balanced with the basic objective of making the consideration of biodiversity by nontechnical users as simple and routine as other commonly considered elements such as stormwater management or scenic views. Stormwater management is a suitable analogy because it required many years of base data collection and analysis-tool development by the federal government prior to becoming a common element of consideration at the local level (Environmental Protection Agency 1995). Negative effects of over- or improper development in the form of increased flooding and pollution created the political will to adopt regulations and tools to implement them (Environmental Protection Agency 1995). I believe that the state of biodiversity decline has reached a similar crisis and will follow a similar pattern of solution development and implementation.

To address this need, GAP currently has two ongoing DSS development projects to serve two different user groups with very different missions. The first is BEST, the biodiversity expert systems tool designed for local government land use planners to consider impacts of proposed land uses on species and plant communities (see http://www.gap.uidaho.edu/RA/View.asp?ProjectID=-2147483648). It is intended to operate at the front end of the development permit process and allows queries of impacts from specific land use types on individual biotic elements. BEST is an expert systems tool, rather than a full decision support system in the more robust definition (Zhu et al. 1998) in that it provides knowledge necessary for objective decision-making. It does not guide the user through the complete process to reach the final decision such as "permit or deny the development proposal" because of the many other nonbiotic factors that must be considered by land use planners. We define an expert systems tool in this context as one that incorporates knowledge of experts into a system that allows easy retrieval of that knowledge to support decisions by nonexperts. Such systems are applicable to situations where it is impractical or inefficient to provide direct interaction with the expert because it is either too costly or the need is too infrequent to retain continuous access to such expertise. The approach is also valuable when the expert knowledge would be inefficient to access on a case-by-case basis, such as when literature reviews or calculations are required. In those situations, it is more efficient to gather the information all at once and archive it in a database (Zhu et al. 1998).

The second GAP DSS under development is Refuge-GAP, intended to aid the U.S. Fish and Wildlife Service in using GAP and other data for refuge planning (see sidebar). This system has some similarity to BEST in that it allows the user to specify a parcel, in this case an existing refuge or user-defined area, and to identify those elements that may occur on the parcel. It is also hierarchical so that the user can query for elements with different levels or categories of risk. Unlike BEST, it contains a decision-making process for scoring lands for acquisition potential using an established system LAPS (Land Acquisition Priority System). The combination of functions provided in Refuge-GAP supports both an established agency process and also the ability to explore and query the information in a more flexible manner.

These two systems characterize the two basic, though not exclusive, types of biodiversity applications: short-term impact assessment and long-term planning. BEST is an example of a system designed to address short-term local retention of biodiversity. This is accomplished by predicting impacts to individual biota from specific land uses at the level of individual parcels. Refuge-GAP is a planning system that provides a broad-scale, longer-term view of elements in need of further protection by the refuge network. The former predicts risk and aids in mitigation of that risk before the impact has occurred, while the latter identifies elements that could be currently at risk (conservation gaps) and reduces their level of risk in the future through refuge planning.

When the concepts underlying these two systems are coupled, a system emerges that addresses biodiversity conservation across the gradient of scales from large-area, long-term planning to single-parcel, short-term decision-making and design. This coupling is critical to moving systematic conservation forward because too often large-area plans fail for lack of attention as to how the plan will be implemented by site-level decisions. Likewise, site-level based activities conducted without the context of a larger plan result in fragmented, ineffective, or destructive actions.

The recent workshops have called for integration of DSS tools to create more robust systems that can address, for example, biodiversity, natural hazards, and socioeconomics in project planning. Integration of tools is complicated, however, by their creation in separate institutions for different user groups. It suffers from the same difficulties of integration that GIS data itself suffered prior to standards for metadata and interoperability-problems not yet entirely resolved. The difficulty of integrating DSS modules composed of custom scripts, models, and interfaces will, I believe, be far more difficult than problems of GIS data interoperability. Institution of top-down directives and funding systems to facilitate integration similar to those of FGDC and NBII programs for GIS data may be the only way to address this need, but such approaches are not currently viewed as desirable or feasible. For these reasons, I suggested to BRD that the focus of integration be on the output of DSSs. That is, that the output from one system be readily readable and usable by another system. Outputs are already fairly standardized in the form of GIS, text, tabular, and graphic image formats. The dream of some may be a suite of "plug-and-play" DSS modules, but without substantial funding, standards, and agency directives, that dream may never be realized.

Moving Forward

Technologically, development of a DSS has become a relatively simple task with off-the-shelf tools for interface development and better integration of GIS, database, and graphics software. The real work comes in the interpretation of the user’s mission or decision-making process to queries that can be answered by a DSS. In the simplest of systems, this requires converting the data and documentation to a format that is directly informative to the user’s questions and making documentation accessible and friendly. In a manuscript for Landscape Ecology (Crist et al. in press) I explore the science needs in the development of biodiversity decision support systems. However, the discussion is far from exhaustive. The science needs fall under three categories that must be integrated in a biodiversity DSS: biological science, GIS/technological science, and decision process science. The discipline of the user group for which the DSS is designed has to be incorporated with all of these. Each of these categories is discussed briefly below.

Biological Science Needs:

Given our desire to create biodiversity decision support systems (or BDSS as I will refer to it from here on), the most important need in this regard is for distribution and status information for all biotic elements. GAP follows an element-based approach in mapping and analysis. I advocate the same for a DSS because I believe there are not yet, and may never be, surrogates appropriate for multiple phyla, umbrella species, or ecoregion approaches that can effectively conserve all biodiversity. This is because such approaches cannot incorporate all of the requirements of every species. Given the unlikelihood that all elements will be mapped and assessed before many of them are extirpated, however, I acknowledge the need for the use of surrogates where appropriate. These surrogates should be applied at the level of the guild, in sufficient detail to conserve the diversity of niches for species with the narrowest niche requirements. I also recognize the need for more comprehensive ecosystem assessment and conservation methods to deal with the nonbiotic components of biodiversity conservation, such as ecosystem processes. Some specific biological science needs are:

(a) Species information database: In constructing BEST, qualitative information on the biotic elements had to be drawn from GAP, Heritage databases, literature, and personal communications. Gathering and interpreting sufficient information to designate species sensitivities was a substantial effort that required approximately four months for 40 species. This experience has led me to advocate a central, national database (or linked distributed databases centrally accessed) of our sum knowledge of all species. The content would include not only the taxonomy and habitat association information, but the life history, known sensitivities, and of course, known and predicted distribution. This information, collected according to uniform protocols and a stable set of data fields, could be accessed for use in the development of land use and management plans and individual project impact assessments. Standardized databases of biological elements are taking shape through The Nature Conservancy (http://www.consci.tnc.org/src/zoodata.htm), Natural Heritage Programs, Gap Analysis Program, and the Integrated Taxonomic Information System (ITIS) (http://biology.usgs.gov/cbi2/programs/itis.html), but they generally lack the data required for the kind of biotic impact prediction that many users need. Another benefit of such an approach is that it could foster archiving the vast amount of information gathered every year by academia, agencies, private consultants, naturalists, and programs such as NatureMapping (http://salmo.cqs.washington.edu/~wagap/nm/).

(b) Impact guilds: Despite the need to address biodiversity conservation at the individual species level (Noss and Cooperrider 1994), many plant and animal species may face extinction before they can be adequately studied for sensitivities to human impacts. This suggests that we must generalize existing knowledge of studied species to provide guidelines for related species in the form of "impact guilds" (Truett et al. 1994, O’Neill et al. 1986). Because some genetically related species show high degrees of variation in response to the same disturbance, impacts guilds should be based upon similar response to the same categories of human land uses, e.g., "elimination of the herbaceous layer impacts foraging" (Arcese et al. 1997). The use of this approach can allow assessments of species not previously studied by relating them to other studied species with which they share similar requirements and therefore sensitivities. Individual species assessments, therefore, will not be absolutely necessary to take actions to help conserve those species while they are still viable.

GIS/Technological Science Needs:

As noted above, many of the technological problems of constructing a DSS have been solved with off-the-shelf tools suited to easy system development; however, there are still some difficult challenges. Among these are:

• Interoperability among platforms. Most systems, including ours (developed for ArcView), are not easily ported to other platforms.

• Adaptability to changing and future technologies

• Data exchange-how easy will it be for a nontechnical user to upgrade or add data to the system?

Decision Process Science Needs:

This science is probably the least developed, but much work is being conducted (Jankowski et al. 1997, Pressey et al. 1993). The most important and obvious need is to understand the user’s decision process, what data is needed for that process, and how that data must be packaged, filtered, transformed, etc., to answer user queries. I have found through our pilot projects that simply asking users what they need is not effective; they often have no experience with DSS or GAP data and cannot envision how to incorporate it into their decision-making process. Therefore, an iterative approach to DSS development is required. This begins with a pilot system to demonstrate system ability to deal with some core queries, followed by user feedback on necessary changes and additions to the system. I provide some more thoughts on DSS implementation later; following are some categories of decision process science needs related to a BDSS.

A starting point in the development of a BDSS is a categorization scheme for human activity impacts. In the development of BEST, which "cans" biological expertise for instant access by the system user, we had to build a bridge between the languages of planners and biologists. This led us to develop a system for categorizing land uses in terms of ecological impacts rather than the standard planning descriptions such as "single family subdivision" that have no ecological meaning. This system provides a biologist with the information to assess biotic sensitivities to land uses without direct knowledge of the exact land use or specific design. The system also had to take the biological knowledge and report it back to the planner in terms meaningful to him or her. Putting aside the real need for conservation plans for every species, we should conduct specific impact assessments for each biotic element and each proposed human activity at the local level. This would require a substantial effort. Therefore, a categorization scheme that can translate human activities into categories of impacts is necessary to provide a more generalized approach.

There is discomfort with the idea that a biologist will assess potential impact to a biotic element without knowing the specific land use and design (M. Scott, pers. comm.; also Truett et al. 1994), but some authors readily acknowledge both the necessity and workability of such an approach (Duncan et al. 1986, Blair 1996). Currently, for individual land use project assessments, the biologist and land use planner must educate each other about both the nature of the land use, and nature of the habitat and biotic elements, and integrate the information into knowledge that can be used to determine biotic impacts (C. Whiteside, pers. comm.). Although such an arrangement would be beneficial, few jurisdictions can afford it. BEST uses a priori categorization of land uses to levels of impact, therefore capturing the knowledge in a single effort to be used for all future assessments (with periodic updates as needed). Emlen (1974) gives support to this approach, noting that the effects of urbanization on birds can usually be readily identified in the form of increase in perch sites that support territorial bird species, cavity and arboreal nest sites, water availability, ground predation, and food. Therefore, one should be able to examine species’ habits and resource needs, compare these to categories of urbanization and land use, and predict if the species will do better or worse after development. Knight and Cole (1995) likewise identify characteristics of recreation activities that can be used to predict species reactions in combination with information on the species. They also note that the similarity of the human disturbance to natural disturbance can be used to predict species reactions.

New types of land uses appear all the time (e.g. paint ball, jet skis); therefore, it is important that the impact categories be generalizable and do not require an individual study of every new land use type before predicting possible impacts. The concept is familiar to land use planners and managers when developing land-use zones or management units where the capabilities and desired types of uses are identified with enough flexibility to accommodate unforeseen uses. The approach has also been used by conservation planners to denote zones of differing degrees of use and management activities (Foreman et al. 1997). The scheme used in BEST (Table 1) is an attempt to categorize land use impacts; it has known and unknown inadequacies. I believe, however, that it is an improvement over others (e.g., Blair 1996, Knight and Cole 1995) that categorize land use from a human-social rather than ecological perspective. To make the categorization useful for biologists to predict impacts on biodiversity, the latter perspective is critical (Clemmons and Buchholz 1997). Adoption of this approach would allow any of the hundreds of different human land uses to be satisfactorily and objectively placed into categories of habitat impact and biologists to relate the categories of impact to biotic sensitivities. While a small group of informal reviewers composed of both planners and biologists found the pilot scheme satisfactory, I sense this is only evidence of the lack of previous consideration of such a need.

Developing an ideal scheme would first require identification of the characteristics of human activities applicable to assessing species sensitivities. In an attempt to balance simplicity with sufficient detail in BEST, I divided land use impacts into two categories: land cover alteration and human presence. I believe that balance will remain critical for a functional system, but the characteristics should be established through rigorous examination. The quantitative requirements of the categories require a substantial research effort to identify meaningful levels of physical habitat disturbance and human presence for each category. These measures need to be easily obtainable by the planner or manager from the land use characteristics as well as meaningful to the biologist assessing the species’ response. A national standardized scheme will be essential for consistent application and distribution of the knowledge of human activity impacts throughout the nation. Development of the scheme could be done through the FGDC similar to the way the National Vegetation Classification (FGDC 1997) was established. Experts on all taxonomic groups would have to be involved to develop one or more categorization schemes that would be suitable for all taxa; not a trivial task to say the least.

Coupled with the human activity impacts scheme is a standardized biotic sensitivity ranking scheme. The biotic rankings provide the information to the planner as to the degree of anticipated impact to the species from the proposed land use. I developed a five-level ordinal scheme for BEST, ranging from 1) beneficial to 4) severe, and a fifth level (dangerous to humans) applicable to animals (see Table 2). However, at least one other scheme (and probably more) exist, such as that of the National Marine Fisheries Service (1996). While I believe that scheme is overly complex and not sufficiently explicit, it could be modified for a standardized national scheme by including the "dangerous" ranking (level 5) and "beneficial" (level 1) from BEST. These two categories are important for avoiding hazardous conflicts and predicting creation of nuisances from land uses that benefit human-adapted species. Numerous schemes are possible (for a more extensive discussion see Westman 1985), but what is required is a scheme that is simple enough to quickly complete species assessments a priori and not require biologists to use a level of precision inappropriate to the available knowledge. At the same time the scheme must provide sufficient information to the planner to indicate what behaviors and habitat characteristics are being impacted and the degree of severity of the impact. In the end the information must be suitable for making the decision as to whether the activity should be relocated, denied, or mitigated and how. It is also very important that the scheme produce consistent results when tested by a group of biologists applying the scheme to the same group of species (M. Scott, pers. comm.).

Implementation

User groups are not homogeneous entities, and, even if we could develop one system for each user group, it simply would not suffice. For instance, county governments range from minimal service providers of fire protection and road maintenance to those as sophisticated and complex as that of a large city (Johnson 1998). The goal of DSS development in general, and GAP’s initiatives specifically, should not be to try and serve all potential users, but to produce models for a breadth of application types. These can then inspire and facilitate further development and distribution of systems within those user communities. For example, our approach with USFWS is to develop an initial product and then facilitate the production model and training in its use within USFWS itself.

The commitment to DSS must be long and patient. Many institutions are only now committing resources to the creation or use of spatial data, and it will be years before institutions routinely provide resources for DSS development and implementation. GAP will continue to work with interested cooperators on pilot projects and coordinate our efforts with those of Aurora, DOI, BRD, and others.

Conclusion

The interest and enthusiasm that our pilot project BEST has generated has been gratifying but also surprising, given the small amount of funding and time devoted to its development as well as the subsequent uncertainties in its output. This indicates to me that a) there is tremendous need and desire for this type of system and b) given the fact that human impacts continue daily with little biological consideration, people are willing to live with large uncertainties in the decision process. With undeniable and staggering losses of biodiversity occurring, it will be critical for the potential user groups of these systems to provide the resources needed not only to implement them but also conduct the research necessary to develop systems that deliver scientifically sound and defensible results.

Literature Cited

Arcese, P., L.F. Keller, and J.R. Cary. 1997. Why hire a behaviorist into a conservation or management team? In J.R. Clemmons and R. Buchholz, editors. Behavioral approaches to conservation in the wild. Cambridge University Press. Cambridge, United Kingdom.

Blair, R.B. 1996. Land use and avian species diversity along an urban gradient. Ecological Applications 6(2):506-519.

Clemmons, J.R. and R. Buchholz. 1997. Linking conservation and behavior. In J.R. Clemmons and R. Buchholz, editors. Behavioral approaches to conservation in the wild. Cambridge University Press. Cambridge, United Kingdom.

Crist, P., T. Kohley, J. Oakleaf. In press. BEST: an expert systems tool for assessing land use impacts on biodiversity. Landscape Ecology.

Duncan, D.K., E.E. Johnson-Duncan, and R.R. Johnson. 1986. Urban environments as avian habitat. In K. Stenberg and W.W. Shaw, editors. Wildlife conservation and new residential developments: Proceedings of a National Symposium on Urban Wildlife. School of Renewable Natural Resources, University of Arizona.

Emlen, J.T. 1974. An urban bird community in Tucson, Arizona: Derivation, structure, regulation. Condor 76(2):184-197.

Environmental Protection Agency. 1995. Storm water discharges potentially addressed by phase II of the National Pollutant Discharge Elimination System Storm Water Program. Report to Congress. EPA 833-K-94-002.

FGDC, Vegetation Subcommittee. 1997. FGDC vegetation classification and information standards, June 3, 1996 draft. Federal Geographic Data Committee Secretariat, Reston, Virginia. 35 pp.

Foreman, D., A. Holdsworth, and J. Humphrey. 1997 (unpublished). Draft Sky Island/Greater Gila Nature Reserve Network proposal. Sky Island Alliance, Tucson, Arizona.

Jankowski, P., T.L. Nyerges, A. Smith, T.J. Moore, and E. Horvath. 1997. Spatial group choice: a SDSS tool for collaborative spatial decision-making. Int. J. Geographical Information Science 11:577-602.

Johnson, R. 1998. Presentation to USGS Gateways to the Earth workshop. October 14-15, 1998. Sterling, West Virginia.

Knight, R.L. and D.N. Cole. 1995. Wildlife responses to recreationists. In R.L. Knight and K.J. Gutzwiller, editors. Wildlife and recreationists: Coexistence through management and research. Island Press, Washington, D.C.

National Marine Fisheries Service. 1996. Making Endangered Species Act determinations of effect for individual or grouped actions at the watershed scale. NMFS, Environmental and Technical Services Division, Habitat Conservation Branch.

Noss, R., and A. Cooperrider. 1994. Saving nature’s legacy. Island Press, Washington, D.C.

O’Neill, R.V., D.L. DeAngelis, J.B. Waide, and T.F.H. Allen. 1986. A hierarchical concept of ecosystems. Monographs in Population Biology, #23, Princeton University Press, Princeton, New Jersey.

Pressey, R.L., Humphries, C.J., Margules, C.R., Vane-Wright, R.I., Williams, P.H. 1993. Beyond opportunism: Key principles for systematic reserve selection. Trends in Ecology and Evolution 8:124-128.

Truett, J.C., H.L. Short, and S.C. Williamson. 1994. Ecological impact assessment. Pages 607-622 in T.A. Bookhout, editor. Research and management techniques for wildlife and habitats. Fifth edition. The Wildlife Society, Bethesda, Maryland.
Westman, W.E. 1985. Ecology, impact assessment, and environmental planning. John Wiley & Sons. New York.

Zhu, X., R.G. Healey, and R.J. Aspinall. 1998. A knowledge-based systems approach to design of spatial decision support systems for environmental management. Environmental Management 22:35-48.