Volume No. 10, 2001
The mission of the Gap Analysis Program is to prevent conservation crises by providing conservation assessments of biotic elements (plant communities and native animal species) and to facilitate the application of this information to land management activities (Gap Analysis Program 2000). This is accomplished through the following five objectives:
1.Map actual land cover as closely as possible to the alliance level (UNESCO 1973, Federal Geographic Data Committee 1997).
2. Map the predicted distribution of those terrestrial vertebrates and selected other taxa that spend any important part of their life history in the project area and for which adequate distributional habitats, associations, and mapped habitat variables are available.
3. Document the representation of natural vegetation communities and animal species in areas managed for the long-term maintenance of biodiversity.
4. the public and those
charged with land use, research, policy, planning, and
management.
5. Build institutional
cooperation in the application of this information to state and
regional management activities.
To meet these objectives, it is necessary that GAP be operated at state or regional levels but maintain consistency with national standards. Within the state, participation by a wide variety of cooperators is necessary and desirable to ensure understanding and acceptance of the data and forge relationships that will lead to cooperative conservation planning.
In 1989, with the support of the National Fish and Wildlife Foundation, Idaho conducted the initial research and development of the Gap Analysis Project concept and developed the prototype for national GAP projects. During the past decade, the National GAP office has updated standards for GAP products. New remote sensing and GIS technology have improved our ability to map and analyze Idaho's natural resources, while state and federal land use objectives have brought new challenges to the state. These changes have prompted Idaho to revisit its original GAP project and update its findings using new land cover information, revised species-habitat data, and an up-to-date map of land stewardship practices.
This second edition of Idaho GAP varies from the first in a few significant ways. First, our land cover mapping and subsequent classification have been conducted at a finer spatial resolution. The spectral footprint of the MSS imagery used in GAP I (1989) was 4 hectares; no habitat features smaller than 4 hectares could be detected, causing a broad-brush approach to both vegetation identification and habitat modeling for vertebrates (200-ha minimum mapping unit [MMU]). The Landsat TM imagery for GAP II (1996) produced vegetation information for each 0.09-ha area (30-m pixels), allowing evaluation of vegetation at a finer scale and the identification of minor land cover species of importance to the state (2-ha MMU). The finer scale from Landsat imagery is still considered broad-brush by biologists who study species in their discrete habitats, but the Landsat resolution meets GAP's objective to visualize the state's overall biodiversity. In addition to the finer scale, GAP II's vegetation classification came with values for slope, aspect, and elevation for each 30-meter pixel. This would prove useful in refining some of the WHR models for habitat specificity. Both vegetation classification systems identified groupings of forest, shrubland, grassland, and riparian, but the finer scale of the Landsat images also allowed us to quantify unique habitats for specialized species, such as reptiles and amphibians.
Wildlife Habitat Relationship Models were built on vertebrate life history information from peer-reviewed literature. GAP II built upon the foundational references on habitat affinity in Idaho used in GAP I, and reviewed major species-specific journal articles published between 1950 and 1998 to garner additional habitat information. Unfortunately, up until the past few years, most field researchers have failed to record useful habitat information in their published reports (Karl et al. 1999). Without knowledge of a species' use of slope or scale or elevation much of the additional information available in the Landsat land cover layer went unused.
Between the GAP I and GAP II stewardship products, a greater attempt was made, in concert with Conservation Data Center, to provide detailed information on each of the ownership types and management objectives. This is an ongoing project that will improve over the coming years. As it is, ID-GAP can now refine its identification of potentially threatened environments.
For ID-GAP, Idaho land cover was mapped in two sections. Redmond et al. (1996) at the University of Montana's Wildlife Spatial Analysis Lab (WSAL) mapped the northern part of the state as part of a U.S. Forest Service Region 1 land cover mapping effort. Homer (1998), at the Utah State University Remote Sensing/GIS Laboratory, mapped the southern part of the state as part of the Wyoming mapping initiative. Contracting with two different remote sensing labs, which were already mapping vegetation in adjacent states, expedited the development of Idaho's vegetation layer for gap analysis. It also created a minimally disjunctive land base on which to conduct subsequent research. Although the mapping endeavors were conducted independently, Homer's (1998) vegetation classification system was designed to compliment the earlier work of Redmond et al. (1996). Satellite imagery was acquired primarily from the growing seasons during 1992 and 1993, but some scenes were selected from other years (ranging from 1991 to 1995) to minimize cloud cover.
The Northern Idaho vegetation map was created from Landsat TM
scenes and stored in a series of seven ARC/INFO grids (one per TM
scene covering Northern Idaho).
The database was built through a two-stage classification
involving both unsupervised and supervised procedures. First,
for each TM scene, an unsupervised classification of pixels was
conducted. This pixel classification, based on Euclidean
distance calculations, was designed to maintain patterns observed
in a color composite of bands 4, 5, and 3. The resulting
spectral classes were then regrouped and merged to 2-ha MMU (>
22 pixels). Next, a raster database was constructed in
ARC/INFO where base regions (or raster polygons) were delineated,
and attributes for each region were collected. Meanwhile, 7.5
minute quadrangles were selected and field sampled in 1994-95 by
the U.S. Forest Service, Northern Region.
These ground-truth plots were combined with plots from existing
sources and passed to the WSAL, where they underwent a series of
logical and positional tests to verify their accuracy and utility
for supervised classification purposes. In all, 17,854 plots
were compiled in the ground-truth database.
Of these plots, 80% were used in the subsequent supervised
classification, and 20% were used to conduct the accuracy analysis
for the classification system. The supervised classification
system assigned cover type labels using a "Nearest Member of Group"
classifier. Decision rules were applied where necessary in
assigning labels to vegetation, size class, and canopy cover. The riparian vegetation was mapped through a separate process.
Using digital elevation data, predicted riparian zones were
delineated, then spectral classes were selected to represent
riparian vegetation within the zones at a 30 m pixel
resolution.
For southern Idaho, mapping zones were used in an effort to
optimize these criteria and gain desired resolution within
acceptable budgetary and time lines. A mapping zone was
defined as an independent mapping project area. (Vegetation
training sites and classification were applicable only to this
area.)
With mapping zones, an effort was made to contain spatially
similar ecological areas within a reasonable sample of TM
pixels. It was determined that nine mapping zones would
optimize this mapping effort. In each zone a master scene was
selected, and surrounding scenes slaved into the master
scene. A two-step approach of atmospheric standardization and
histogram adjustment was used to mosaic the TM imagery.
Cover-type class definition was based first on correlation with
previous Utah and Nevada classifications, and second, with the
classification scheme generated by the University of Montana.
Signatures in each mapping zone were classified using the ERDAS
(TM) ISODATA algorithm (Tou and Gonzalez 1974) to generate
unsupervised spectral clusters. An iterative review of the
clustering process was used to identify the optimum number of
spectral clusters needed to characterize land-cover variation in
each mapping zone. Cover-type modeling followed the protocol
developed by Homer et al. (1997) and consisted of two phases: (1)
statistical association of spectral classes with cover-types, and
(2) ecological modeling based on ancillary information.
The resulting combined land cover data set consisted of 82
classes and was the highest-resolution, continuous land cover map
yet to be produced for Idaho. Idaho's most extensive
vegetation community was Basin Big Sagebrush (Artemisia
tridentata) and Wyoming Big Sagebrush (Artemisia
tridentata wyomingensis) across southern Idaho. It
covered 34,787 square kilometers or 16.08% of Idaho's land. All sagebrush and shrub-steppe types combined constituted 33% of
the Idaho landscape.
Agriculture ranked second in land area with 29,029 square
kilometers or 13% of land cover. Grassland and meadow
vegetation communities occupied 11% of the Idaho landscape, with
Perennial Grasslands comprising 46% of that area. Douglas-fir
was the most common forest type (7%) in Idaho, and no other single
forest species or forest community occupied more than 5% of the
state landscape. The total forest area was 37% of the Idaho
landscape. Riparian, wetlands, and marshes covered 2% of
Idaho's landscape and were categorized in seven classification
codes. Shrub-dominated riparian occupied the largest area
with 0.87% of the total mapped riparian/wetland distribution. The combined sand and rock classifications occupied 2% of the
landscape with the greatest portion of that distribution seen in
exposed rock.
Assessed accuracy measures of the land cover map varied greatly between areas. Particular attention should be paid to the sample size for each cover type when interpreting the results. For the five scenes combined to create the north Idaho land cover map, producer's accuracy for those comparisons acceptable or better (3 or greater in the fuzzy matrix) ranged from 53.35% to 71.23%. Total percent correct measures for southern Idaho mapping units ranged from 65.5% to 79.3%. Overall percent correct for the southern Idaho land cover classification was 69.3%. Overall, total percent absolutely correct for the Idaho Land Cover Classification was 50.15%. Estimated kappa value for the Idaho Land Cover Classification was 0.4942.
Modeling of vertebrate distributions for ID-GAP followed a 7-step process. First, we compiled a list of species known to breed in Idaho. Second, we collected occurrence and habitat association data for each species. Third, we used the occurrence data to approximate the range boundaries of each species in Idaho. Next, we assembled the habitat association information on breeding habitats into a format acceptable by our modeling programs. Fifth, we combined the range approximation with the coded habitat associations to produce a GIS model of the predicted distribution of each species. Sixth, biologists familiar with the distribution of Idaho's wildlife reviewed the models. Finally, each model was subject to an accuracy assessment with independent occurrence data.
Of species recorded in 10 or more of the accuracy assessment areas, 93.69% of the models were assessed to have greater than 80% correct present. For those species listed in 10 or more areas, the percent correct present ranged from 81.82 to 94.44% for amphibians, 55.56 to 100% for birds, 58.82 to 100% for mammals, and 76.47 to 100% for reptiles. Appendices E through H contain comments on the accuracy of each WHR model for birds, mammals, amphibians, and reptiles, respectively.
Species richness can provide a rough assessment of the diversity
of wildlife within a given area. While species richness as an
index of conservation effectiveness is very limited (e.g., does not
account for representation or rarity, and tends to emphasize
habitat and range edges), it is generally useful for characterizing
regional biological diversity. We defined species richness as
the number of species predicted to occur within a given unit.
For ID-GAP, we investigated species richness by land cover type
and by Environmental Monitoring and Assessment Program (EMAP)
hexagon. Individual species' WHR model grids were combined
and the number of species summed over each unit area. For
calculations of richness by EMAP hexagon, we considered only native
species that were determined to not be able to sustain their
populations exclusively within human-developed landscapes.
Out of 379 species, the maximum predicted to occur in a single
cover type was 235 (62.0%). Thus, no single cover type
contained all species. Riparian cover types were predicted to
be habitat for the most species in Idaho (Table 3.7).
All of the riparian types (excluding wetland types) were predicted
to have over 200 species using them as habitat. Following
riparian areas, the next richest habitats were forested cover
types. The most species-poor cover types (3 to 73 species)
were alpine (perennial ice and snow, alpine meadow), urban, and
non-vegetated cover types.
A total of 317 native, non-anthropogenic vertebrates were considered for analyses of hexagon richness in Idaho (Map 3.4). Of those, 254 were the most predicted to occur within a single hexagon (79.9%) and 80 were the least. Average number of species predicted to occur per hexagon was 184.6 with a standard deviation of 39.8 species. Areas of highest species richness (more than 233 species) occurred in southern Idaho along the Snake River Plain. These areas have many lakes, reservoirs, and wetlands and thus provide a wide variety of habitats for many species. Lowest species richness was observed in the subalpine-forested uplands and alpine areas of northern and central Idaho, the shrub-steppe habitats of Owyhee County, and the largely nonvegetated lava fields of southern Idaho. While species richness is lower in these regions, they provide unique habitats to some species that are found nowhere else in the state (e.g., northern bog lemming [Synaptomys borealis] in northern Idaho, rock squirrel [Spermophilus variegates] in Owyhee County). This highlights one of the shortcomings of assessing conservation status using species richness.
To fulfill the analytical mission of GAP, it is necessary to compare the mapped distribution of elements of biodiversity with their representation in different categories of land ownership and management. We use the term "stewardship" in place of "ownership" in recognition that legal ownership does not necessarily equate to the entity charged with management of the resource, and that the mix of ownership and managing entities is a complex and rapidly changing condition not suitably mapped by GAP. At the same time, it is necessary to distinguish between stewardship and management status in that a single category of land stewardship such as a national forest may contain several degrees of management for biodiversity. The purpose of comparing biotic distribution with stewardship is to provide a method by which land stewards can assess their relative amount of responsibility for the management of a species or plant community, and identify other stewards sharing that responsibility. This information can reveal opportunities for cooperative management of that resource, which directly supports the primary mission of GAP to provide objective, scientific information to decision makers and managers to make informed decisions regarding biodiversity.
After comparison of biotic occurrences to stewardship, it is
also necessary to compare with categories of management
status. The purpose of this comparison is to identify the
need for change in management status for the distribution of
individual elements or areas containing high degrees of
diversity. Such changes can be accomplished in many ways that
do not affect the stewardship status. GAP currently uses a
scale of 1 to 4 to denote relative degree of maintenance of
biodiversity for each tract. A status of "1" denotes the
highest, most permanent level of maintenance, and "4" represents
the lowest level of biodiversity management, or unknown status.
In reality, there exists a gradient of human impacts on the land
with no landscape unmodified to some extent by human activities,
but this categorization is useful for analytical purposes.
Stewardship map data were assembled from two sources. Data at 1:100,000 scale were carried forward from previous work at the Idaho Gap Analysis Lab completed from 1989-1991 (Caicco et al. 1995). That data set included major administrative land units including those under federal, state, tribal, and private ownership.
Status 1 and 2 polygons, digitized at 1:24,000 scale, were provided by the Idaho Conservation Data Center (CDC) and were combined with existing 1:100,000 data. Sliver polygons, resulting from the discrepancy between parcel boundaries digitized at disparate scales, were removed, as were those polygons smaller than 2 hectares, the minimum mapping unit (MMU) for Idaho Gap Analysis. Polygons in the land stewardship coverage were assigned protection status values from 1 to 4 based on their owner and management status tracked by the Idaho Conservation Data Center.
Public lands (federal and state) comprised approximately 14,980,800 ha (69.31%) of Idaho. State lands accounted for approximately 1,109,400 ha (5.13%) of Idaho. Private lands made up 6,448,100 ha (29.83%) of Idaho. Of this amount, 11,200 ha (0.174%) is in status 1 management. The Nature Conservancy owns and manages 94.53% of all private status 1 lands in the state (Table 4.2).
The area of Idaho land in status 1 and 2 was 321,500 ha (1.49%)
and 2,229,500 ha (10.32%), respectively. Protection status 3
lands covered 12,442,600 ha (57.57%) of Idaho, and 6,437,000 ha
(29.78%) were in status 4. The majority of status 2 lands
were contained in Idaho's wilderness area complex, managed by the
USFS (1,556,900 ha, 69.83% of status 2 lands).
Other major status 2 land managers were the Department of Energy
(Idaho National Engineering and Environmental Laboratory [INEEL]
231,600 ha, 10.39%), wildlife protection areas and wildlife refuges
managed by the U.S. Fish and Wildlife Service (33,000 ha, 1.48% of
status 2 lands) and Idaho Department of Fish and Game (Wildlife
Management Areas, 119,500 ha, 5.36%).
The primary objective of GAP is to provide information on the distribution and status of several elements of biological diversity.
Intersecting the land stewardship and management map with the
distribution of the elements resulted in tables summarizing the
area and percentage of total mapped distribution of each element in
different land stewardship and management categories. The
data were formatted to allow users to query the representation of
each element in different land stewardship and management
categories, as appropriate to their own management objectives.
This formed the basis of GAP's mission to provide landowners and
managers with the information necessary to conduct informed policy
development, planning, and management for biodiversity
maintenance.
Although GAP seeks to identify habitat types and species not
adequately represented in the current network of biodiversity
management areas, it is unrealistic to create a standard definition
of "adequate representation" for either land cover types or
individual species (Noss et al. 1995).
A practical solution to this problem is to report both percentages
and absolute area of each vegetation type or vertebrate species in
biodiversity management areas, as described above, and allow the
user to determine which types are adequately represented in natural
areas based on detailed studies of the ecology, population
viability assessments, as well as studies of the spatial and
temporal dimensions of ecological processes. Clearly,
opinions will differ among users, but this disagreement is an issue
of policy, not scientific analysis. We have, however,
provided a breakdown along five levels of representation (0-<1%,
1-<10%, 10-<20%, 20-<50%, and >=50%). The <1%
level indicates those elements with essentially none of their
predicted distribution in protected areas. Levels 10%, 20%
and 50% have been recommended in the literature as necessary
amounts of conservation (Odum and Odum 1972, Specht et al. 1974,
Ride 1975, Miller 1984, Noss 1991, Noss and Cooperrider 1994),
although biologically defensible goals may be much higher
(Soulé and Sanjayan 1998).
Of Idaho's 71 mapped natural vegetation cover types (excluding 1000s, 2000, 3102, 5000, 9800, 9900), five had less than 1% of their total area represented in the combined protected statuses of 1 and 2. Twenty-six cover types had between 1% and 10% of their total area in status 1 and 2 lands. Nine cover types identified by the ID-GAP project had more than 50% of their total area in status 1 and 2 lands.
For the analysis of vertebrate distributions against land stewardship, we evaluated only those species that were not introduced or considered strongly associated with human-developed habitats (317 of 379 total vertebrate species modeled). We found 123 vertebrate species (38.8% of all 317 vertebrate species considered) with less than 10% of their predicted habitat on status 1 and 2 lands. This included 61 bird species (31.6% of all bird species considered), 38 mammals (42.2% of all mammal species considered), 16 reptiles (76.2% of all reptiles species considered), and 8 amphibians (61.5% of all amphibian species considered). The Clark's grebe (Aechmophorus clarkii) was the only species to have greater than 50% of its predicted habitat in status 1 and 2 lands.
At least 43.7% of natural land cover types and 38.8% of native, non-anthropogenic terrestrial vertebrates have been identified by ID-GAP as having levels of occurrence on lands managed for the long-term maintenance of biological diversity below what may be required for maintenance of viable populations. These underprotected (or underrepresented) land cover types and vertebrate species occur mostly at lower elevations under a variety of land stewardships including substantial areas of private ownership.
This project has provided Idaho with the most spatially refined and thematically detailed statewide compilation of information on Idaho's land cover types, vertebrate distributions, and land conservation status. These data should be considered an update to any previous information created as part of the ID-GAP program, and while more accurate and detailed data may exist for localized parts of Idaho, the data presented here is an enhancement over other conservation data sets currently being used statewide. Using these data, a myriad of research opportunities now exist.
To increase the utility of these data layers and their useful life span, continuing research needs to be directed toward three main areas: (1) further assessing the quality, appropriate uses, and limitations of the existing data layers; (2) refining the existing data based on continuing research, new data, and identified errors; and (3) developing methods to apply the data to real-world problems and applications affecting land use planning, management and conservation. There is much work yet to be done to refine the ID-GAP products and develop them into an indispensable tool for conservation planning in Idaho. Along these lines, we make the following suggestions for initial steps to improve the quality and usability of ID-GAP data:
1. further accuracy assessment of existing data layers,
2. periodic updates to the Idaho land cover map,
3. continual updating of the vertebrate habitat models,
4. continual updating of the Idaho land stewardship layer,
5. development of a system to disseminate ID-GAP data and support users.
Caicco, S.L., J.M. Scott, B. Butterfield, and B. Csuti. 1995. A gap analysis of the management status of the vegetation of Idaho (U.S.A.). Conservation Biology 9:498-511.
Federal Geographic Data Committee, Vegetation Subcommittee. 1997. FGDC Vegetation Classification and Information Standards-June 3, 1996 Draft. FGDC Secretariat, Reston, Virginia. 35 pp.
Gap Analysis Program. 2000. A handbook for conducting Gap Analysis. USGS Gap Analysis Program, Moscow, Idaho. http://www.gap.uidaho.edu/handbook.
Homer, C.G. 1998. Idaho/Western Wyoming landcover classification. Remote Sensing/GIS Laboratories, Utah State University, Logan, Utah.
Homer, C.G., R.D. Ramsey, T.C. Edwards, Jr., and A. Falconer. 1997. Landscape cover-type modeling using a multi-scene thematic mapper mosaic. Photogrammetric Engineering and Remote Sensing 63:59-67.
Karl, J.W., N.M. Wright, P.J. Heglund,
and J.M. Scott.
1999. Obtaining environmental measures to facilitate
vertebrate habitat modeling. Wildlife Society
Bulletin 27:357-365.
Miller, K.R. 1984.
The natural protected areas of the world. Pages 20-23 in
J.A. McNeely and K.R. Miller, editors. National parks,
conservation and development: The role of protected areas in
sustaining society. Smithsonian Institution, Washington,
DC.
Noss, R.F. 1991. Report to the Fund for Animals in Washington, DC.
Noss, R.F., and A.Y. Cooperrider. 1994. Saving nature's legacy: Protecting and restoring biodiversity. Island Press, Washington, DC.
Noss, R.F., E.T. LaRoe III, and J.M. Scott. 1995. Endangered ecosystems of the United States: A preliminary assessment of loss and degradation. Biological Report 28, National Biological Service, Washington, DC.
Odum, E.D., and H.T. Odum. 1972. Natural areas as necessary components of man's total environment. Transactions of the North American Wildlife and Natural Resources Conference 39:178-189.
Redmond, R.L., Z. Ma, T.P. Tady, J.. Winne, J. Schumacher, J. Troutwine, and S.W. Holloway. 1996. Mapping existing vegetation and land cover across western Montana and northern Idaho. Wildlife Spatial Analysis Laboratory, Missoula, Montana.
Ride, W.L.D. 1975. Towards an integrated system: A study of selection and acquisition of national parks and nature reserves in Western Australia. In F. Fenner, editor. A national system of ecological reserves in Australia. Australian Academy of Science, Canberra.
Soulé, M.E., and M.A. Sanjayan. 1998. Conservation targets: Do they help? Science 279:2060-2061.
Specht, R.L., E.M. Roe, and V.H. Boughlon. 1974. Australian Journal of Botany Supplement Series. Supplement No. 7.
Tou, J.T., and R.C. Gonzalez. 1974. Pattern recognition principles. Addison-Wesley Publishing Company. Reading, Massachusetts.
UNESCO. 1973. International classification and mapping of vegetation. United Nations Educational, Scientific and Cultural Organization, Paris, France.
Return to Table of Contents