cro20

This information describes the "Conservation and Restoration Opportunities" (CONSRESTOP) map, a component of the Conservation Scenario of the Willamette River Basins Alternative Futures project conducted by Pacific Northwest Ecosystem Research Consortium (PNW-ERC) (1997-2002). This project was funded by Environmental Protection Agency (US/EPA). The work described herein was performed by researchers from The Institute for a Sustainable Environment (ISE), University of Oregon (contact: David Hulse, Dept. of Landscape Architecture) with assistance from Oregon State University (Dept. of Fisheries and Wildlife). An expert working group, a citizen stakeholder group, and PNW-ERC scientists provided recommendations and background information.

The CONSRESTOP mapping describes areas within the Willamette River Basin in which habitat conservation or restoration opportunities are feasible and achievable under a futures scenario that emphasizes ecological services and habitat protection. Evaluation of the Conservation 2050 scenario indicates that conversion of these areas to a protected and restored native habitat land cover will provide a significant increase in the likelihood that terrestrial and aquatic species biodiversity will be maintained or even improved within the WRB. This was not found to be the case with Plan Trend 2050 or Development 2050, the other two alternative futures depicted and evaluated in this project. Clearly, this map represents just one possible set of opportunities, and, as discussed below, the locations of most of the opportunity areas are not absolute. However, this study does suggest that if the acreage and approximate location of each of these opportunity areas are achieved, positive environmental results are likely. For the purposes of this project and the map it produced, conservation was defined as maintaining and protecting through the year 2050 those priority habitat types which exist ca. 1990. Restoration was defined as providing, by 2050, more of those habitat types of which we have lost the most from 1850-1990.

These results and more details on the individual futures scenarios and their components can be found in "Willamette River Basin Planning Atlas: Trajectories of Environmental and Ecological Change" (Editors: Hulse, Baker and Gregory, 2002; published by OSU Press, Corvallis, OR.). Digital data sets can be obtained from the web site: <http://oregonstate.edu/Dept/pnw-erc/>.

Two levels of habitat protection are represented in this mapping: "Tier 1" and "Tier 2". Tier 1 areas are those in which conservation and restoration are the focal activities; no resource use or development occurs within these areas. Tier 2 areas are those in which resource use continues under best management practices designed to protect environmental values, and in which land cover is compatible with adjacent Tier 1 areas. Some Tier 2 areas transition to Tier 1 areas over the sixty year period of simulation (i.e. 1990 to 2050).

There are six grids that describe this "map"-one for each decade from 2000 to 2050. The starting condition for this sequence of changes is described by the "Current Conditions ca. 1990" land use/land cover map (see <http://oregonstate.edu/Dept/pnw-erc/> to download this data set).

For each time step from 2000 through 2050 the data describe the location and extent of areas that are (within the limits of the data) suitable for either conservation or restoration of selected habitat types. They are (by their data code): (1) Tier 1 oak, (2) Tier 1 prairie (wet and dry), (3) Tier 1 floodplain forest, (4) Tier 1 upland forest, (5) Tier 1 mid-elevation (Metro natural area) forest, (6) riparian protection zones inside Tier 1 forests, (7) Tier 1 wetlands, (10) restored channels of the Willamette River along with other major water bodies (including the mainstem Willamette River), (14) Tier 2 forests, (15) Tier 2 oak and prairie, (16) Tier 2 riparian protection zones, and (18) Tier 2 wetland protection zones. Please read on for a description of Tier 1 and 2 areas. Depicted conservation and restoration opportunity areas are phased in over time, as described by the time-stepped grids, simulating the time necessary to both acquire and to restore the vegetation. Each of these ecosystem types and the processes defining the opportunity areas are described further below.

In general, the habitat types chosen for definition of conservation and restoration areas are those that were once common within the WRB (as analyzed using data from 1850 pre-settlement surveys) and that are now (1990) reduced to very small areas (oak, prairie, wetlands, bottomland forest). Also included are areas that are highly valuable in that they provide significant ecological benefits as well as services to the WRB population (riparian corridors, Metro natural area forest, upland forest, and protective wetland buffers). The Conservation Scenario also included modifications to the Willamette River to increase channel complexity in accordance with historic river descriptions and the overall intentions of the Conservation 2050 scenario.

The resolution of these grids is 30 m; the format is ARC/INFO GRID. The mapping of the modeled land use/land cover within the CONSRESTOP areas can be found in each of the grids describing the time steps of the Conservation scenario (download from the data access page of the web site <http://oregonstate.edu/Dept/pnw-erc/>). For further description of land use/land cover categories used in the discussion below, see the ERC legend at <http://www.fsl.orst.edu/pnwerc/wrb/maps/ec90/lgnd_comparison.html>.

In performing this work, we had to draw from a large number of data sources, some of which were less accurate than others, and some of which contradicted others. In particular, descriptions of existing vegetative communities were either incomplete (not covering the entire WRB), and/or were described at a scale and legend much coarser than we needed. The ERC primary database describing current conditions (ca, 1990) was derived from LANDSAT Thematic Mapper data and depicted landuse / landcover in 65 different categories at a 30 m resolution. However, location of existing oak savanna, native prairie and wetlands areas was problematic for several reasons. In the case of oak savanna, the structure of such an ecosystem (scattered trees in a grassy matrix) proved too complex a pattern for the techniques used in analyzing the LANDSAT data. Additionally, the land cover types of "hardwoods" or "mixed forest" available from the satellite provided insufficient information to differentiate with certainty between oaks, maples, alder, ash, etc. Similarly, differentiation between exotic pasture grasses, hayfields and native grasses (all subject to a variety of management regimes) was difficult, resulting in poor confidence in locating existing high value prairie areas. Locating wetlands from our data also provided a challenge in that agricultural land use (in particular) overlaid the functional areas of the wetlands and obscured their land cover "signature".

We had thus to rely on a modeling approach in which we combined the various land cover data sets with data describing soils (hydric soils, depth of water table, etc.), historic extents of the vegetative communities, land management, etc. to select suitable areas of each ecosystem type. The approach used for defining high priority conservation and restoration opportunities for each habitat type is described below.

We attempted to locate areas that were distributed across the WRB thus forming a network that would provide connected habitats within the larger managed landscape. Clearly, only certain areas are suitable for sustaining a particular habitat type. For example, riparian corridors must be located along water bodies; bottomland forests are adapted to high water tables, and so forth. Thus, all areas of the WRB are not equal when it comes to defining conservation or restoration areas.

In addition to locating areas that possessed potential conservation value, we also considered the costs of restoration efforts (for some habitat types). For example, agricultural areas growing crops with long rotation periods and possessing expensive infrastructure (e.g., vineyards) were considered less likely to be candidates for restoration even if these areas were adjacent to areas assessed as having high conservation potential. Areas with high densities of structures were similarly avoided as they were judged to be expensive to acquire and difficult to manage with the natural disturbances (such as fire or flooding) historically required to sustain the desired vegetation types. In particular, high structure density in the northern part of the WRB made it difficult to locate oak savanna and prairie, ecosystems for which management by fire is desirable.

Minimum patch sizes of individual conservation areas and target total acreages were defined for each habitat type by a technical expert group, and then were examined for political feasibility by the citizen stakeholders (Hulse et al., 2004). This process resulted in the expert-recommended target areas being reduced by the stakeholders to levels that were thought to be politically plausible and realistically achievable. We then attempted to locate areas within the WRB that were of the minimum size or larger and that added up to the target acreages. In some cases, these targets were not achieved. One reason was a lack of sufficiently large "core" areas around which to define the conservation "patch". Another reason was dense urbanization (i.e. road networks, buildings) in some rural regions and dense networks of small parcels that made it more difficult to achieve large areas of land under single or coordinated management practices (in particular in the north end of the WRB).

When Tier 1 areas were mapped within agricultural fields, those fields with less than 5 acres remaining for agriculture were "abandoned" and the adjacent Tier 1 area was expanded to include these remnant fields. In such cases the irrigation water rights associated with these parcels were presumed converted to in-stream uses.

Table 1 (below) lists the target acres and reports the approximate acreage of each habitat type included in the Conservation 2050 scenario. Acreages exclude water and rural structures, roads, and other development that lie within the CONSRESTOP areas.

For each decade of the Conservation scenario, selected former channels of the Willamette River abandoned by the river or cut-off by revetment or other fill activity are reconnected to the river. This increase in channel complexity is guided by knowledge of historic river conditions and the documented loss of complexity since the 1850's (Hulse et al., 2002 pp. 18-25, 132-147,166-173). Since the upper sections of the river between Corvallis and Eugene have been the most simplified, greater lengths of channels are reconnected in this area. Islands are created by this process. This mapping and the analyses were carried out by Linda Ashkenas, Dept. of Fisheries and Wildlife, Oregon State University.

At each decennial time step, the more complex river configuration for that decade is introduced in place of the 1990 description. Thus, "streams/rivers" replace some other riverside land use/land cover, including agricultural crops and forests. Riparian corridors that are established in earlier decades of the CONSRESOP simulation are redefined to recognize these changing water conditions.

Note that on this map legend class (10) includes major water bodies that existed within the Willamette River Basin in 1990 as well as the reconnected channels of the Willamette River. Comparison with the water categories mapped in the "Current Conditions ca. 1990" land use/land cover map will show the sequence of changes made in and around the Willamette River. These major water bodies comprise all rivers and streams of mapped Strahler order five and greater as well as lakes and ponds. The extent of most of these areas is static throughout the sixty year simulation. They are shown in this map to provide geographic context for the riparian mapping, as described below. In addition, this map class includes all areas identified as water in satellite imagery.

Riparian protection zones intersecting Tier 1 lands are considered to be Tier 1 areas. Those riparian zones within Tier 1 CONSRESTOP areas of floodplain forest, oak savanna, and prairie, are not mapped separately as "riparian". However, because of specific policy initiatives that govern forest management, stream riparian zones within Tier 1 upland and mid-elevation forest areas are explicitly defined (CONRESTOP legend category 6). Outside of Tier 1 areas, such as in farm, rural residential, urban, or private industrial/non-industrial forest lands, riparian corridors are mapped and are considered to be Tier 2 (CONSRESTOP legend 16). Whether identified explicitly or not, land use/land cover consistent with riparian vegetation was modeled in all riparian protection zones.

The riparian corridors mapped here are defined as protection zones around almost all water bodies (see below for exceptions); a set of standard widths was adopted, as described below. We recognize that some riparian areas, as described by their vegetation and hydrology, are naturally narrow while others are quite extensive in area. The buffer width policy mapped here follows common guidelines that aim to protect against water pollution from nutrient and sediment run-off as well as to conserve riparian vegetation for stream temperature control, habitat and as a food source. The buffer widths chosen were considered by the stakeholders to be politically feasible.

Buffered water bodies consist of all water shown in each time step of the Conservation scenario (ERC land cover legend values 32 and 33 (lakes/reservoirs)), except for ponds in agricultural areas (as defined by the agricultural model) that are less than 5 acres in size. Also buffered are all streams of Strahler order 1-4 regardless of whether they are fish-bearing or not. (Note that these streams are NOT mapped on the land use/land cover grids of the Conservation Scenario due to their assumed small width). The mapped stream density in USFS and BLM lands is much greater than that mapped in other parts of the WRB, resulting in higher riparian corridor density in the Federal lands.

The width of the buffer is measured from the edge of the mapped water. Thus, a "100 ft buffer width" or "buffered at 100 ft" refers to a corridor that extends 100 ft on each side of a river with the total width being 200 ft plus the width of the water. Various buffer widths are adopted depending on whether a property is privately or publicly owned, and whether it lies inside or outside an urban growth boundary. Within the METRO UGB, all water bodies on private lands are buffered to 200 ft, following the most expansive recommendations in the METRO TITLE 3 ordinance available at the time this work was performed. In other UGBs, the buffer widths on private lands are 25 ft on small streams (orders 1-2), 50 ft on larger streams (orders 3-5), and 100 ft on all other rivers and all lakes. On all public urban lands, the buffer width is 300 ft. In a few areas, remnant agricultural fields adjacent to CONSRESTOP areas have been abandoned and incorporated into the adjacent conservation/restoration area. If this occurred adjacent to riparian buffers, then the mapped width of that part of the riparian corridor will be greater than defined above.

In all forestry areas (as defined by the forest model), private forest streams are buffered at 100 ft, while state and federal streams are buffered at 300 ft, following an upgraded Oregon Forest Practices Act, and the Northwest Forest Plan. Other public lands are buffered at 300 ft. In private lands not under forest land use (i.e., rural residential, agricultural, and other areas) a 100 ft buffer is employed.

The time sequence in creating the buffers is based on the 1995 303-d listings of streams with water quality impairments (see Federal Clean Water Act), with the total length of riparian buffering spread approximately evenly over the six decades. Buffers are first placed around all 303-d listed water bodies, and then on water bodies that lie within the watersheds of those listed streams, and finally around those remaining unlisted.

Within these riparian buffers, existing infrastructure (roads, rural structures, urban land use) are not removed in the modeling of the land cover. However, no new development is permitted. All forested land cover is retained within the buffers; all agricultural crops are converted to either a grass cover or to trees (land cover legend "forest closed mixed") depending on the adjacent crop at the time of installation of the buffer.

Restoration and conservation of existing wetlands (5 acres and larger) was prescribed for the Conservation scenario, along with a 3,000 acre expansion of emergent wetlands. All are Tier 1 lands. Restoration/conservation of all of the selected wetlands is assumed to occur over the 60 years from 1990 to 2050, with approximately equal areas being restored in each decade.

The only category of wetlands that is explicitly included in the PNW-ERC land use/land cover legend is "flooded marsh" (ERC class 89). The "signature" of other wetlands can be found in the current conditions land use/land cover map as consistent combinations or grouping of various vegetation types but such signatures are apparently not unique to wetlands. Also, since the basis of the current conditions map is LANDSAT data acquired in a relatively dry year (1992), "wetlands" are under-represented. Consequently, other sources of data were needed to define the location of existing wetlands. Since no one source comprehensively covered the full extent of the WRB, we had to combine several data sets. These were: National Wetlands Inventory (NWI), Oregon Dept. of Fish and Wildlife (ODFW) land cover mapping of the Willamette Valley, The Nature Conservancy mapping of riparian wetlands, wetland mapping by METRO within the Portland-METRO UGB, and, the "flooded marsh" and "water" areas from LANDSAT-based PNW-ERC 1990 existing conditions mapping (ERC land use/land cover legend categories 89, 32 and 33).

These data sources do not have the same spatial extent, are of different scales, have different attributes and categories, and are dated to different years. Some of the mapped wetlands are consistently located in the different data sources, some are found in one data source and not in others. There is no information available as to the condition of these wetlands - i.e., whether they are functioning, whether they have been filled, etc. Thus, it is essential that the user of the CONSRESTOP map be especially cautious when utilizing the wetlands mapping.

The following describes the layering process used to construct the wetlands coverage, where any wetland region on an "upper layer" replaces a coincident one on a "lower" layer. ERC flooded marsh and water categories were selected as the "top" layer. ODFW categories of hawthorn riparian, willow, reed canarygrass, and cattail-bulrush wetland were then layered on top of selected NWI lacustrine and palustrine classes (see below). METRO lacustrine and palustrine classes formed the fourth layer, and TNC wetlands (aquatic bed/floating plant, emergent wetland, bog, and shrub scrub communities) that lay outside the area described by the NWI and METRO data sets were finally added as the bottom layer. Finally, wetlands less than 5 acres in size were removed.

At each 10-year time step of the CONSRESTOP modeling, vegetative land cover within randomly selected wetlands is updated to simulate the restoration process. All existing built categories (roads, railroads, rural structures, and urban built categories are retained. Trees (hybrid poplar, hardwoods, mixed forest, and closed canopy conifers (PNW-ERC classes 49-66), shrub (ERC 87, 101), flooded marsh (ERC 89), water (ERC 29, 32, 33) are retained. Grassy land cover (grass seed crops, hay, and pasture (classes ERC 67, 83 and 85)) are restored to "natural grass" (ERC class 86). The land cover in the remaining areas within the wetlands (primarily agricultural crops) is replaced by expanding the cover of any treed land cover (other than "closed conifer"), grass, shrubs and flooded marsh into those vacated areas. "Shrub" within a wetland is described as "wet shrub" (ERC 101) if the majority area of that wetland is located in an area defined by either high water table depth or lies within the 100 ft riparian corridor. Otherwise it is set to "shrub" (ERC class 87).

Additional emergent wetlands are created by selecting existing emergent wetlands and seasonal wetlands (from mapping described above with TNC emergent wetland category, ODFW reed canarygrass wetland, and selected NWI PEM and PSS categories (see White, 2000)). These are expanded by 45 m into adjacent areas of soils with seasonal high water table (less than 6" from the surface). In some instances, wetlands of other types were replaced by this expansion. This process yields approximately 3000 acres of additional emergent wetland, the target specified by the stakeholders, and occurs in six decennial increments. Land cover within these new wetlands is defined as above, except that instead of "natural grass" (ERC 86), category "flooded marsh" (ERC 89) is defined, and instead of "shrub" (ERC 87), "wet shrub" (ERC 101) is specified. All built categories are retained.

Wetlands intersecting any other Tier 1 areas (except water bodies) are mapped on top of these other CONSREST types.

Wetland protection zones are placed around all designated wetlands in CONSRESTOP in the decade in which they are selected for restoration and/or conservation. These buffers are intended to protect the wetlands from degradation due to nutrient and sediment runoff resulting from adjacent land use. All wetland buffers are designated as 100 ft in width: these stretch 100 ft out from the edge of the wetland.

As was the case with riparian protection zones, wetland buffers falling within Tier 1 areas are not displayed in these CONSRESTOP grids. Nevertheless, the land cover in these regions is modeled consistent with the rules below. Outside of Tier 1 areas, wetland buffers are considered to be Tier 2 areas.

Suitable land cover in wetland buffers adjacent to agricultural fields is dependent on the type of crop in each field. A woody crop (PNW-ERC classes 72, 73, 75, 92, 93) or pasture (ERC 85) is replaced by shrub (ERC 87 or 101, the latter in wet areas as defined above) in the designated buffer. Other crops (e.g. field crops) are edged with replacement "grass" (ERC 80) land cover. All other existing vegetative land cover (all hardwood, mixed and closed conifer tree areas, natural grass, and shrub), and water and flooded marsh are retained in the buffers. All existing built categories (roads, railroads, rural structures, and urban built categories) are retained.

The targeted area for oak savanna conservation/restoration was 50,000 acres, with a minimum patch size of 350 acres, and some patches greater than 1,000 acres scattered throughout the Willamette River basin. The objective was to find sufficiently large core areas of existing oaks around which oak savanna restoration could be feasibly planned. Since "oak" vegetation was not identified within the PNW-ERC existing conditions (circa 1990) land use/land cover map, other sources of data were used to complement the ERC mapping.

First, the potential range of oak within the Willamette River Basin was estimated by combining the pre-settlement vegetation (PESVEG categories of white and black oak, oak/pine, oak/fir), Oregon Biodiversity Project communities-at-risk mapping (categories for oak savanna and woodlands), and Willamette Valley vegetation mapped by Oregon Dept of Fish and Wildlife (categories containing oak). The resulting area was constrained to elevations below 1000 ft. Further, areas were eliminated where the use of fire as a management tool was infeasible. This included areas of high density of rural structures (more than 64 per sq. mile), rural residential zones, urban areas inside UGBs, airport campuses, areas within 1/8 mile of major arterials, and areas within 1/4 mile of Interstate 5. (These latter two constraints were based on smoke management rules from the State Fire Marshal). Oak CONSREST areas were located only within this resulting potential range mask.

A probability ranking was next described within this mask to predict the presence of oak within the land use/land cover mapping of 1990 existing conditions. This ranking was developed by combining the ODFW oak mapping with certain ERC categories (in order of probability these were "forest closed hardwood" and "forest closed mixed;" "upland forest semi-closed hardwood" and "natural shrub"; "upland forest open" and "upland forest semi-closed mixed"; and, "upland semi-closed conifer" and "conifers 0-20 years"). The resulting values varied from 0 (not present) to 10 (high certainty), and were filtered to reduce "noise" by using a moving analysis window (focal majority).

Only areas with a sufficiently large core of existing oaks were considered as potential conservation sites. These were identified by first combining contiguous areas with high probability ranking (8-10) and then retaining only those patches with cumulative area greater than 10 acres. Then, all patches within 240 meters of each other were identified as a region, and a 90 m buffer was drawn around each region. The regions were then confined (clipped) to the oak potential range mask. It was further assumed that acquisition of the Tier 1 conservation areas would be focused on larger taxlot parcels. Hence areas with a high density of small tax lots (more than 15 tax lots per sq. mile) were also eliminated. Finally, the size of each region was re-examined and those smaller than 10 acres were discarded.

The next step was to rank and map areas in which restoration was considered feasible (compared with conservation, described above). ERC existing conditions data were ranked from 10 to 0 based on a subjective likelihood of restoration (considering the ease of conversion to oak savanna, and economic constraints). The woodland categories described above plus "natural grasslands" were given a value of 10; "pasture", "hayfields" and "bare/fallow" areas were ranked as 8. Old conifers, all built areas, water and marsh were eliminated from consideration. Highly ranked areas that coincided with both the pre-settlement oak range and with the mapping by the Oregon Biodiversity Project were considered to be the most feasible restoration areas. As was done for the core conservation areas, this probability surface was smoothed using the moving analysis window (focal majority); high probability areas were grouped into patches, and only patches of 50 acres or more were retained. Areas with many small tax lots were eliminated.

Once the "core conservation" and the "feasible restoration" patches were defined, a process was undertaken to select from these the best areas for oak conservation/restoration that (1) met the patch size requirements, (2) were distributed throughout the WRB, and (3) added up to the targeted acreage. First, each of these patches was ranked as to conservation value based on ownership (public preferred over private), the number of sensitive species that occurred within the patch (based on data from the Oregon Natural Heritage Program), the number of these species that preferred oak or prairie communities, and the minimum distance to surface water (the closer the better). Highly valued conservation patches were merged with adjacent highly valued and highly ranked (feasible) restoration patches to form oak CONSREST regions. Only regions greater than 350 acres were retained.

The process above describes most of the regions defined for oak savanna restoration. One additional region was added from the list of the top 100 natural areas provided by METRO Greenspaces Program.

All regions were progressively changed from Tier 2 to Tier 1 over the course of the sixty year simulation, with approximately equal areas being reclassified as Tier 1 in each decade. The core conservation areas were first assigned to Tier 1 in order of conservation value, followed by highest valued and highest ranked restoration areas, and finally with the least valued and lowest ranked areas being transferred to Tier 1 in 2050. This simulates the concept of saving the best as soon as possible, and then working to restore surrounding areas as time permits. The temporary Tier 2 areas act as protective buffers to the core areas until they are in turn restored.

The primary land cover assigned to the Tier 1 oak areas is "oak savanna" (ERC legend 98) which is understood to represent widely-spaced oak trees in a matrix of grass and forbs with some shrub understory. This is not a blanket assignment: any conifers older than 81 years (ERC classes 60 and 61) and any areas of water (ERC classes 29, 32, 33) are retained, and in riparian areas within the oak CONSREST regions, all existing trees, "flooded marsh" (ERC 89) and "wet shrub" (ERC 101) are retained along with a portion of the existing "natural grassland" and "dry shrub" (ERC 87). All wetlands are retained. All existing built categories (roads, railroads, rural structures, and urban built categories) are retained

A similar process to that used in defining oak savanna areas was used to locate potential prairie restoration sites. "Prairie" refers to either upland or wet prairie. The total targeted area for prairie was 70,000 acres, with a minimum patch size of 100 acres. The objective was to find sufficiently large core areas around which prairie restoration could be feasibly planned. Since "prairie" vegetation was not identified within the PNW-ERC existing conditions land use/land cover map, other sources of data were used to complement the ERC mapping.

The area in which prairie restoration was considered biologically feasible was defined by the combination of the historic range of prairie in the WRB (from the pre-settlement vegetation (PESVEG) mapping categories of "upland" and "wet" prairies), and by the area of soils with seasonal high water tables (hydrogroup D with water table depths less than 6" from the surface). In addition, this feasibility mask was extended to include areas mapped circa 1990 by ODFW and by TNC as "prairie." The resultant mapping was filtered, and long narrow strands (lying along rivers) were removed, yielding a smoother, less fragmented pattern.

Existing land cover conditions were next evaluated to assess the likelihood of successful restoration. Three different descriptions of land cover were used within the feasibility mask described above - ERC land use/land cover, TNC wetland mapping, and ODFW Willamette Valley vegetation mapping. Within each map, each vegetation category was ranked based on a subjective assessment of likelihood of conversion to "prairie" community (considering existing conditions and economic constraints). From the ERC existing conditions ca. 1990, LU/LC categories of "natural grass", "flooded marsh", "natural shrub" and "pasture" were ranked highest, followed next by "closed hardwood" (assumed to be predominately ash in the wet prairie feasibility area, and oak in the upland prairie areas), and then grassy crops ("grass seed", "hayfield", "bare/fallow", "turf grass", "burned grass" categories). Other land use/land cover categories were ranked very low to infeasible.

Within ODFW vegetation mapping of the Willamette Valley, "hairgrass prairie" was ranked as the best opportunity for prairie conservation/restoration, with "ash-cottonwood-bottomland pastures" selected as next desirable. Within the TNC plant communities coverage, "prairie" areas were identified as the best existing habitats, followed next by areas of abandoned, shrubby agricultural lands ("EXOFOR"), ash/oak areas ("FRQUNA"), and oak communities ("QUEREX", "QUERNA").

These three ranked maps were combined and the scores renormalized to provide a ranked spatial description of potential prairie restoration sites. All potential restoration areas were constrained to the biologically feasible area; all built areas and water bodies were excluded; and the pattern was filtered to reduce the "noise" created by isolated pixels of contrasting LU/LC. Areas of highest potential were those where all three data sources indicated highest probability of existing prairie vegetation.

The next step was to select CONSRESTOP areas from the set of potential prairie restoration sites. The strategy was to locate core areas of the "best" habitat (most highly ranked and most probable areas for successful prairie restoration) and expand outward around these into the lesser ranked potential prairie areas until patches of acceptable size were formed (100 acres or greater). In areas not in public or NGO ownership, all restoration/conservation areas had to be suitable for management using fire (i.e., density of structures had to be low; areas could not be close to busy highways). Further, areas with many small taxlots (more than 15 per sq. mi.) were assumed to imply many private owners which in turn was assumed to present considerable difficulties in either management or acquisition of restoration sites. These areas were therefore excluded from consideration.

Exceptions to these criteria were permitted on existing public and NGO lands after it became obvious during the analysis that these lands contained almost all of the known native prairie, areas of which lie within or close to 1990 UGBs. Precedent exists for managing prairies with fire inside the Eugene UGB on the West Eugene Wetlands and thus these exceptions were considered reasonable given the value of rare native prairie to the conservation network. Public and NGO reserves included a BLM RNA north of Fern Ridge, Willow Creek, Kingston Prairie, Finley Wildlife Refuge, Fisher Butte RNA, Amazon RNA, Jackson-Frazier Wetlands, West Eugene Wetlands, and Mt Pisgah. Other State and National Wildlife Refuges, and RNAs with potential prairie restoration sites were added to this property list. Two suitable areas in the Metro Top 100 Natural areas (targets for public acquisition) were also included.

Since insufficient prairie restoration sites were found within public and NGO lands to meet the targeted acreage, restoration areas on privately owned lands were next considered for conservation/restoration. Most of the remaining areas of known prairie remnants were located in private lands adjacent to Jackson-Frazier Wetlands and Kingston Prairie, and clustered within the Philomath and Corvallis UGBs. These were included in the CONSRESTOP inventory by expanding the boundaries of the two public and NGO reserves (simulating protection through purchase or conservation easements) and by setting up a Philomath/Corvallis reserve.

Potential prairie areas that abutted or intersected previously defined oak conservation or estoration areas were next considered. By selecting contiguous prairie - oak savanna areas larger conservation sites were defined: this was assumed to enhance the conservation value of both habitat types and improve management efficiencies. For most oak CONSRESTOP areas, the associated prairie was "assembled" by locating the contiguous highest ranked areas that together exceeded 100 acres in size. In a few cases involving areas close to urban areas or public lands, the 100 acre limit was relaxed -the open space and recreational value together made up for the diminished habitat value of the smaller area.

At this stage in the modeling the desired area of prairie as specified by the stakeholders had not yet been achieved. Prairie sites were next selected by focusing on core areas of 10 acres or more of the areas with high prairie potential. These "cores" were ERC land cover classes "natural grass," "shrub," "flooded marsh," and "pasture," coincident with any ODFW and TNC classes indicative of wetland conditions (as described above). Lesser valued contiguous sites (of both upland and wet sites) were then added out a maximum distance of 700 m from each core. (Note that this process differs from the first stage of defining prairie CONSRESTOP areas where existing mapped native prairie core areas of any size were used, as described earlier). Resulting areas that did not meet the minimum patch size or that were fragmented by dwellings or other incompatible land uses, were excluded.

Further, if the density of prairie species (plant or animal) exceeded 2 spp/sq. mile within any potential prairie zone, then that zone was added to the prairie CONSRESTOP map (regardless of its proximity to oak restoration areas or high value core areas).

When the search for restoration areas was concluded, the target area for prairie habitat in the CONSRESTOP network was not achieved: just under 38,000 acres were defined compared with the target of 70,000 acres. The data available did not warrant the making of further selections.

All prairie regions were progressively changed from Tier 2 to Tier 1 in approximately equal areas over the course of the sixty year simulation in order of rank, the existing prairie areas ca. 1990 being the first to be assigned Tier 1 status. By 2020 at least some part of each region was under restoration, i.e. was in Tier 1.

"Prairie" areas are generally defined by the ERC legend category "natural grassland." Forest land cover types "closed hardwood forests" (ERC class 53, conifers older than 80 years (ERC classes 60 and 61), and all water bodies (ERC classes 29, 32 and 33) were retained. This simulates the removal of all agricultural crops and young conifers and their replacement with native grasses. Note that in the process of building the complete conservation network with all habitat types, marsh areas (ERC class 89) and wetlands (see above) coincident with the "prairie" areas were preeminent.

The CONSRESTOP areas of this community type were derived from forested areas specified in the 1998 mapping by the Metro Green Space Program of the top 100 natural areas outside the METRO UGB. One region was designated as oak habitat, and was assigned to the oak savanna CONSREST type (see above); several others were prairie habitat. The rest were classified as "mid-elevation forest." Overlapping rural residential zones were removed.

All regions were progressively changed from Tier 2 to Tier 1 over the course of the 60 year simulation, with approximately equal areas being reclassified as Tier 1 in each decade. The largest tax lots intersecting the mapped area were chosen first, followed by adjacent tax lots in order of size.

The following land cover classes were retained within each conservation region: water bodies (ERC classes 29, 32, 33), all tree classes (ERC classes 49, 51-62), natural grass, (ERC 86), shrub (ERC 87, 101), and flooded marsh (ERC 89). All other classes were replaced by natural grassland (ERC 86). This simulates a decision to provide for open large fields within the natural areas to accommodate Metro residents in recreational pursuits.

Upland forest conservation opportunity areas are derived almost exclusively from federally owned forests. Their size and spatial distribution were deemed adequate for habitat maintenance of dependent species by a technical group convened for this purpose. Wildlife reserves, Research Natural Areas, sensitive habitats, Late Successional Reserves, and other areas from which commercial harvest is excluded or restricted comprise this category.

Federal forests are managed toward older average age by restricting harvest to stands 60 years of age or older. Trees in older age classes are aged forward as modeled time progresses in ten-year time steps. Three hundred foot riparian protection zones are maintained around streams, and non-coniferous vegetation remains unchanged.

Achieving by 2050 the target area for this habitat type, 58,700 ha., was considered politically infeasible by the stakeholder's Possible Futures Working Group responsible for defining the assumptions for the Conservation 2050 scenario of which the CONSRESTOP map is a major component. This directed the delineation effort toward a primarily biophysical prioritization rather than an approach based on a strict distance to water measurement. Data including pre-settlement vegetation, depth to water table based on SSURGO soils classification, public ownership, recent (1996) flood extent, and distance to major stream confluences were combined to establish the candidate space for floodplain forest. This resulted in a total area of 27,519 ha. allocated to this habitat type by 2050, about 47% of the target value. DIGITAL DATA SETS:

Ownership: PUBLANDX grid - created by this project from tax-assessor data, data acquired from the State Service Center GIS download site, and from a forest ownership grid created by Atterbury and Associates. Streams: RIV_REACH2 line coverage (see <http://oregonstate.edu/dept/pnw-erc> ) and stream coverages provided by Willamette National Forest, Mt Hood National Forest, and BLM.

303-D listings: The 1998 edition of the Oregon Dept. of Environmental Quality digital map and tabular listings, obtained from the Oregon Spatial Data Library (see <http://www.sscgis.state.or.us/data/alphalist.html>).

WETLANDS:TNC wetlands: aquatic bed/floating plant, emergent wetland, bog, shrub scrub communities. Downloaded from State Service Center GIS.

METRO wetlands: lacustrine and palustrine categories from GIS file WET_FILL.SHP from RLIS Lite CD (Metro Regional Govt).

ODFW: Willamette Valley vegetation; landuse categories 20, 22, 30, 31. Downloaded from <http://www.nwhi.org>; dated 1993

NWI wetlands: <http://www.nwi.fws.gov>, dated from early 80's; selected palustrine and lacustrine classes - see table 4, White (2000).

METRO top 100 natural areas outside UGB: data set obtained from Metro Regional Government, Green Space Program.

Oregon Biodiversity Project - at risk communities; data CD from Defenders of Wildlife.

ODFW - Willamette Valley Vegetation. Downloaded from <http://www.nwhi.org>; dated 1993

Oregon Natural Heritage Program: sensitive species occurrence data base. Provided by ONHP.

Soils: interim SSURGO (with STATSGO data base in areas not defined by SSURGO). Hydric soils, seasonal water table.

METRO top 100 natural areas outside UGB: data set obtained from Metro Regional Government, Green Space Program. PESVEG - Historic vegetation circa 1850, extended from Willamette Valley GLO mapping to the entire basin; see <http://oregonstate.edu/dept/pnw-erc>

Oregon Biodiversity Project - at risk communities; data CD from Defenders of Wildlife.

ODFW - Willamette Valley Vegetation. Downloaded from <http://www.nwhi.org>; dated 1993

Oregon Natural Heritage Program: sensitive species occurrence data base. Provided by ONHP.

Soils: interim SSURGO (with STATSGO data base in areas not defined by SSURGO). Hydric soils, seasonal water table.

Tax assessor parcel data bases: acquired from all counties (excluding Yamhill and Columbia).

METRO top 100 natural areas outside UGB: data set obtained from Metro Regional Government, Green Space Program.

Rural residential zones: acquired from State Service Center GIS download site, as part of the general zoning coverage.

Table 1. Conservation and Restoration Opportunity (CRO) areas defined for Conservation 2050 scenario, showing expert group targets

and the results from CRO modeling.

		Conservation and Restoration Opportunity Areas (CRO)
Habitat designated for conservation and restoration	Expert group: target area in 2050 (ha.)	Tier 1 : hectares in 2050	% Tier 1 in 1990 public ownership	% Tier 1 in 1990 agricultural use	Tier 2: hectares in 2050
Floodplain Forest	58,700	23,300	16	29	25,300
Oak Savanna	20,200	22,300	7	21	4,200
Wet/Dry Prairie	28,300	15,300	29	53	5,800
Emergent Wetlands	1,200	1,200	2	78	0
Conifer Forests	40,500
Upland (1)		310,200	98	0	0
ÝÝÝÝÝÝÝÝÝÝÝÝÝÝ Mid-elevation		21,500	4	6	3,600
Riparian Protection Zone (2)
within Tier 1 forests (1)		199,200	99	0
within other forests (3)			90 (5)	0 (5)	236,200
other lands (4)			15 (5)	31 (5)	64,700

(1) The public "upland" forests are Federal forest reserves created under the Northwest Forest Plan and were present in 1990; no additional Tier 1 areas were

ÝÝÝÝ added in public lands; no additional riparian reserves were added.

(2) Riparian protection zones within Tier 1 floodplain forest, oak savanna, prairie and wetlands are not included.

(3) "Other forests" are private industrial forests and public forest matrix lands.

(4) "Other lands" include areas in agriculture, urban and rural development, private non-industrial forests, and other Tier 2 areas.

(5) These are percentages within the Tier 2 area component of the 'within other forests' and 'other lands' categories.

HULSE, D., S. GREGORY, J. BAKER. (Eds). (2002) Willamette River Basin Planning Atlas: Trajectories of environmental and ecological change. (2nd edition), Oregon State University Press, Corvallis, Oregon 97333. 180 p.

HULSE, D., A. BRANSCOMB and S. PAYNE. (2004). Envisioning Alternatives: using citizen guidance to map spatially explicit assumptions about future land and water use. Journal of Ecological Applications as part of an Invited Feature titled "Willamette Basin Alternative Futures Analysis".

WHITE, D. 2000. "Terrestrial Vertebrate Species of the Willamette River Basin: Species-Habitat Relationships Matrix", 25 July 2000, <http://oregonstate.edu/dept/pnw-erc>

Data sets &References: Titus, J.H., et al, 1996. Native wetland, riparian, and upland plant communities and their biota in the Willamette Valley, Oregon. Report to Environmental Protection Agency, Region X, Seattle, Wa. Willamette Basin Geographic Initiative Program, Nov. 1996.

cro20

Metadata: