0529-B1
Daniel W. McKenney, Brendan Mackey, Jacques Trencia and Cameron Wilson 1
Forest management requires descriptive and predictive capacity across large landscapes with seamless, interoperable data layers. In this paper we characterize these information needs and note that they are ubiquitous across all forested nations. Understanding the distinction between primary and secondary data and the potential for economies of scale in data base development and dissemination is critical with scarce budgets. We provide complementary examples of spatial data base development, delivery and use from both Canada and Australia.
Forest management is a complex and difficult challenge due in no small part to the diversity of services forests provide and the interactions between these services (Clawson, 1975). Traditionally, managers were primarily required to consider how their actions (and non-actions) would affect wood supply. Increasingly, all forest managers, even plantation managers, must consider the consequences for other goods and services including recreation, carbon storage, water yields and quality, wildlife habitat, and other dimensions of biodiversity such as genetic diversity. Thus, the kinds of data and information forest managers need has greatly expanded beyond basic inventories of wood supply. Increasingly, managers need to know not only what currently exists in their landscape, but also what is likely or could exist over time. This means predictive as well as descriptive capabilities. Critically, the kinds of phenomena for which data and information are now required include complex biophysical and ecological processes (e.g. the meta-population dynamics of wildlife populations) and economic and social valuation and trade-off analyses. Unfortunately, no simple rules like the historically used Maximum Sustained Yield doctrine exist to guide management in this more modern complex world (Bowes and Krutilla, 1989).
At the root of informed management and policy development is the development of data and information systems that cover particular sites, stands and entire landscapes or regions. It is impossible to completely inventory all possible forest attributes of interest hence landscape and regional scale models are required. Advances in information technology (from computer based Geographic Information Systems- GIS to web services), and modeling tools, combined with historical and current field data collections offer tremendous opportunities to improve the scientific basis for forest management and policy at all scales. Information layers currently used for management decisions are often produced outside the forestry community. There are currently 63 countries involved in "Spatial Data Infrastructure" (SDI) initiatives; these infrastructures vary in scope from regional to international. International efforts include the Global Spatial Data Infrastructure (http://www.gsdi.org/), the Global Disaster Information Network (GDIN; http://www.gdin-international.org/) and the Global Forest Information Service (GFIS; http://www.gfis.net/). National initiatives include the Spatial Information Council for Australia and New Zealand (ANZLIC; http://www.anzlic.org.au/), the U.S. National Spatial Data Initiative (NSDI; http://www.fgdc.gov/) and the GeoConnections Initiative (http://geoconnexions.org). Australia, Canada and U.S. also coordinate their SDI with territorial, provincial or state directories.
In this paper we describe a SDI and a modelling framework to support forest science, management and policy. We demonstrate there is a remarkably common set of data that can be used to help understand ecological complexities and trade-offs that emerge in pursuit of sustainable forestry. We give examples of these data and how they can be accessed, developed and how they have been used in both Canada and Australia.
Figure 1 (McKenney et al., 1996) attempts to characterize the major drivers of forest systems. Sustainable management must be informed about distribution, abundance and productivity. The main influences are disturbance regimes (e.g. fire, insects and diseases, extreme weather, etc), biological regimes (e.g. inter and intra-species competition, genetics), direct human actions or "anthropogenic" regimes (e.g. timber harvesting and land use) and finally environmental drivers (e.g. temperature, moisture, energy and nutrient regimes). These drivers operate at scales ranging from a site or microsite to regional and global (Mackey et al., 1994). Disturbance regimes include insect outbreaks, wild fire and even extreme weather events. Difficulties arise from the fact that disturbances vary in intensity and hence response. The Anthropocentric regime refers to the impacts modern technological society has and continues to have on the forest landscape. This includes the inherited effects of land-use history, current land use including on-going silviculture and other explicit management activities.
The Biological regimes can refer to the genetic responses of living organisms, including their life history and interactions with each other. These regimes are often the target of intensive research programs in forestry. Examples include tree breeding, stand density management and herbicide treatments or competition control. The Primary Environmental Regimes (PERs) refer to the suite of processes that determine the distribution and availability of heat, light, water and mineral nutrients. Historically, much forest research has focused on the influence of these regimes at specific locations. However, forest management has impacts across landscapes - spatial data are needed for all of these influences.
Figure 1 thus not only identifies the major drivers of forest systems it indicates primary data needs. Primary data are generic in that they have a multitude of possible applications; environmental and anthropogenic regime information share the characteristic that they are collected, generated and used by a broad diversity of institutions with local to international focus. The four general processes identified above are ubiquitous and influence several aspects of ecological response. The fact that these ecological processes are generic and ubiquitous also means that economies of scale may be captured by developing such data.
Data collection and generation approaches include field-based observations, remotely sensed (including digitized aerial photo-based forest inventories) and other spatial models. Examples of disturbance regime data in Canada include surveys of insect and disease defoliation (Bowers and Hirvonen, 1998). More recently satellite imagery are being used to help map out changes in forest condition caused by fire (http://fms.nofc.cfs.nrcan.gc.ca/firem3/index.html). Anthropogenic regimes such as land use history can have a critical role in forest succession. The collation of these data often requires the creation of standardized GIS files of historical land use activities. Examples applied to wilderness assessment can be found in Lesslie et al, (1988).
Making the data seamless, interoperable, easily searchable and accessible is important for all potential users. The current paradigm for GIS analysis is based on user-friendly search and access and on data download to a local system. Searching and accessing data sets has been greatly improved recently through standardized metadata providing information about data sets, which in turn, populate catalogues. The emerging web services paradigm uses online Internet based services to provide access and interpretation of data from heterogeneous sources. The benefits of emerging web services are (1) minimal data replication as data coming directly from the responsible/mandated organizations is authoritative (2) cost efficiency resulting from reduced, duplicated storage and (3) integration from heterogeneous sources/platforms and immediate ("plug and play") compatibility. The resulting environment can allow a web developer to customize online applications for specific users' communities without the need for costly GIS infrastructures.
In Canada, a broad spectrum of communities support distributed spatial data warehousing. The CGDI is the global national infrastructure to enable the management of national economic and social priorities including those of the forest sector. It is intended to provide "an integrated, on-line mechanism to deliver geospatial data and services and information for applications serving as well better business, policy decision-making, and value-added commercial activities." (http://geoconnections.org/english/index.html)
The five thrusts of the CGDI are:
While web based services currently allow easy access and simple data analysis, advanced modelling still requires the use of analytical software such as Geographic Information and Image Analysis Systems, statistical packages and other software.
Canada is currently working towards the development of a National Forest Information System as one element of the CGDI serving the forestry community (http://www.nfis.org/). This information system accesses and distributes data from different jurisdictions using web services that are based on national and international standards and specifications and is supporting important national programs like carbon budget modeling and biodiversity indicator reporting for Canada. The data includes topography, hydrography, land use, geo-political boundaries as well as national forest inventory data, productivity and biodiversity data. Another project called NatGRID - National Geo-referenced Information for Decision-makers (McKenney et al, 1996) is creating a nation-wide, spatially-explicit set of the primary environmental regimes through the use of existing climate, topographic and soil/parent material data and re-analyzing these through specific software (http://www.glfc.cfs.nrcan.gc.ca/landscape/index_e.html). Taken together, the availability of these and other types of data are providing increased capacity to develop predictive models of species distribution, abundance and productivity (Figure 1) supporting science, management and policy development. Canadian applications include modeling and mapping the influence of climate on a variety of plants, animals, and diseases (e.g.. Mackey and Sims 1994; McKenney et al 1996; Venier et al. 1998, 1999; McKenney and Pedlar, in press; see also http://www.cfl.cfs.nrcan.gc.ca/ECOLEAP/home.html and http://www.nrcan-rncan.gc.ca/cfs-scf/science/brochure_carbon_budget/index_e.html for other important Canadian forest productivity studies that are utilizing spatial data).
The acquisition and integration of data and information about Australian forested landscapes has been driven in recent years by two major national initiatives: (1) the Regional Forest Agreement (RFA; http://www.rfa.gov.au/) and (2) The Australian National Carbon Accounting System( NCAS; http://www.greenhouse.gov.au/ncas/index.html). Both initiatives have resulted in the generation and integration of substantial spatial data sets about Australia's forest resources within a common GIS framework.
The RFA process aimed at creating jobs and protecting forests through:
Critically, all RFAs were intended to be based on scientific Comprehensive Regional Assessments of the environment, heritage, social and economic uses and values of the forests. These assessments demanded the compilation of all existing data, the systematic collection of new data, and the application of sophisticated analytical procedures.
The RFAs particularly emphasised the analysis of biodiversity values. Australia is a megadiverse country with an estimated 5 588 vertebrate animal species and 22 000 vascular plant species (Common Wealth of Australia, 1996). Detailed information about habitat requirements is absent for most species, with the exception of a few well studies species such as Leadbeaters Possum (Lindenmayer et al. 1991).
Hollow bearing trees represent a critical habitat resource in Australian temperate forest ecosystems. Many forest animals are dependent on tree hollows, which are a function of tree age as tree hollows usually do not start forming in Australian Eucalyptus trees until about 120 years of age. Hollow-dependent vertebrate species include: 13 species of arboreal marsupials; 9 species of scansorial mammals; 16 bat species; around 50 bird species; and 67 species of frogs, lizards and snakes (Gibbons 1994). Thus, a major challenge for the RFAs was to generate spatial estimates of sufficiently fine resolution data about stand age and structure, two critical attributes for mapping the distribution and abundance of hollow-dependent species.
In addition, more comprehensive habitat models were required by the RFA process. Specifically, models were needed that integrate the influence of critical biophysical parameters and processes operating at various scales (see discussion in Mackey and Lindenmayer 2002) including meso-scaled climate and terrain attributes. A major focus of the RFA analyses were the calculation of predictive habitat models. The predictive functions were generated using Generalised Linear Models and General Additive Models to identify statistically significant correlations between the presence/absence (and where data allowed, abundance) of selected taxa and a candidate set of environmental (including vegetation) variables. Spatial predictions of potential habitat were generated by coupling the correlation functions to spatial estimates of the environmental predictors. The RFA habitat modeling process was distinguished by the large number of taxa modeled, the elegant sophistication of the statistical procedures applied, and the high quality of environmental spatial data based specifically developed for the project. These environmental data included 20m resolution digital elevation models and derived terrain, hydrological and surface radiation surfaces - the same type of data noted in Figure 1.
The RFA process made a significant contribution to habitat modeling in Australia through the generation of new survey biological data, the integration of existing biological data, and the application of statistical modeling techniques that utilized GIS data on climate, terrain, land cover, land use, and substrate. The habitat modeling undertaken for the north east forest region remains among the best in Australia and indeed the world, generating spatial predictions of potential habitat for over 450 forest-dependent vertebrate species.
The Commonwealth Government of Australia established the Australian Greenhouse Office as part of its strategy to fulfill obligations under the United Nations Framework Convention on Climate Change. One of the major tasks of the AGO has been the establishment of the National Carbon Accounting System (NCAS). The aim of NCAS is to provide a complete accounting and forecasting capability for human-induced sources and sinks of greenhouse gas emissions from Australian land-base.
Critically, a spatial data base has been developed so that one model can be run across the entire Australian continent on a one hectare spatial resolution (FULLCAM - Richards, 2001). A particularly interesting component of the spatial data base development for FULLCAM is the vegetation cover project which is mapping changes in land cover using remotely sensed data sources from about 1970 to present. Each time period iteration will review areas of change in both the loss and gain of woody vegetation, producing a new woody/non-woody mask as a base for the next iteration. This vegetation cover project will provide a consistent and objective approach to identifying afforestation, reforestation and deforestation across Australia over time.
The RFAs and NCAS have required and resulted in the development of high resolution environmental spatial data bases to support high level quantitative spatial analyses at previously unavailable resolutions.
Forest management has decision-making needs at many scales, from stand level silvicultural prescriptions and allowable cut calculations to regional conservation reserve design and the establishment of baseline standards for nationally reported Criteria & Indicators of sustainable management. The emerging paradigm of web services and the concept of primary spatial data are relevant whether in a tropical, temperate or boreal forest system. Applying and improving geomatics standards and understanding the ubiquitous nature of ecological processes will enable resource management agencies to capture economies of scale in database development programs and support decision-making at many scales. As we move into the 21st century the international forestry community should increasingly recognize these needs and opportunities.
Bowers, W. and H. Hirvonen, 1998. Forest Insect and Disease Conditions in Canada Hall, J.P., Nat. Res. Canada. Canadian Forest Service, Ottawa 72 p.
Bowes, M.D. and J. Krutilla, 1989. Multiple-use Management: the economics of public forestlands. Resources for the Future, Washington, D.C.
Clawson, M. 1975. Forests for what and for whom. John Hopkins University Press, Resources for the Future. Washington.
Commonwealth of Australia (1996). Australian State of the Environment. An Independent Report Presented to the Commonwealth Minister for the Environment by the State of the Environment Advisory Council
ISBN 0 643 05830 3.
Gibbons, P. 1994. Sustaining key old-growth characteristics in native forests used for wood production: retention of trees with hollows. Chapter 5 In. Ecology and Sustainability of Southern Temperate Ecosystems. Edited by T.W. Norton and S.R. Dovers. CSIRO Publishing. pp. 59-84.
Lesslie, R., Mackey, B.G. and Preece, K.M. 1988. A computer-based method for evaluation of wilderness, Environmental Conservation 15: 225-232.
Lindenmayer, D.B., Cunningham,R.B., Tanton, M.T., Smith, A.P. and Nix, H.A. 1991. The conservation of arboreal marsupials in the montane ash forest sof the Central highlands of Victoria, south-east Australia. III. The habitat requirements of Leadbeater's Possum, Gymnobelideus leadbeateri and models of the diversity and abundance of arboreal marsupials. Biological Conservation 56: 295-315.
Mackey, B.G., McKenney, D.W. and Sims, R.A. 1994. Pages 23-50 In: D.W. McKenney, R.A. Sims, M.E. Soule, B.G. Mackey and K.L. Campbell (eds.) Towards a set of Biodiversity indicators for Canadian Forests: Proceedings of a Forest Biodiversity Indicators Workshop1994. Sault Ste. Marie, ON., Canada Nov. 29-Dec.1, 1993.
Mackey, B.G. and Lindenmayer, B.G. 2002. Towards a hierarchical framework for modelling the spatial distributions of animals. Journal of Biogeography 28:1147-1166.
Mackey, B.G., and Sims, R.A. 1993. A climatic analysis of selected boreal tree species and potential responses to global climate change, World Res. Rev. 5(4): 469-487.
McKenney, D.W., Mackey, B.G, Sims, R.A. 1996. Primary databases for forest ecosystem management - examples from Ontario and possibilities for Canada: NatGRID. Environmental Monitoring and Assessment 39: 399-415.
McKenney, D.W., Mackey, B.G., Bogart, J.P., McKee, J.E., Oldham, M.J., Chek, A. 1998. Bioclimatic and spatial analysis of Ontario reptiles and amphibians. Ecoscience 5(1) 18-30.
McKenney, D.W. and Pedlar, J. 2002. Spatial models of site index based on climate and soil properties for two boreal tree species in Ontario, Canada. Forest Ecology and Management, In Press.
Richards, G.P. 2001. The FullCAM Carbon Accounting Model: Development,
Calibration and Implementation for the National Carbon Accounting System. National Carbon Accounting System Technical Report No. 28. The Australian Greenhouse Office, Canberra.
Venier, L.A., Hopkin, A., McKenney, D.W. and Wang, Y. 1998. A spatial, climate-determined risk rating for Scleroderris disease of pines in Ontario. Can. J. For. Res. 28: 1398-1404.
Venier, L.A., McKenney, D.W., Wang, Y. and McKee. J. 1999. Models of large-scale breeding-bird distribution as a function of macro-climate in Ontario, Canada. Journal of Biogeography, 26: 315-328.
Figure 1: Drivers of forest ecosystems and primary data needs (adapted from McKenney et al, 1996)
1 Natural Resources Canada, Great Lakes Forestry Centre, 219 Queen St. E. Sault Ste. Marie, Ontario, Canada P6A 2E5. [email protected]