Michael R. Kunzmann, Susan M. Skirvin, Peter S. Bennett, and Craig A. Wissler
Professional management requires accurate and comprehensive information of park resources. Perhaps the most basic information for managing natural-resource areas is a thorough knowledge of vegetation, an undisputed resource of scenic beauty, wildlife habitat, and overall ecosystem function. Rapid vegetation changes in Southeastern Arizona ecosystems have been well documented (Nichol, 1937; Reynold and Bohning, 1956; Hastings and Turner, 1965; Bahre, 1991). To assess vegetation change at Chiricahua National Monument, an ArcInfo-based GIS has been developed as well as a revised Brown, Lowe, and Pase vegetation classification system. The GIS and vegetation classification systems in conjunction with numerous remote sensing methodologies and computer-based automated field mapping techniques, have been indispensable tools in the process of creating an up-to-date vegetation map. The new vegetation map coupled with GIS analysis provides a means for making informed management decisions about the relative extent and nature of vegetation changes between 1939 and the present. The new vegetation map will also (1) serve as a baseline dataset for monitoring the effects of natural fires and prescribed burning programs, (2) serve as an aid in the evaluation of potential wildlife habitat, and (3) be used to assess the possible impacts of human activities.
Southeastern Arizona's Chiricahua National Monument was established by Presidential Proclamation on April 18, 1924, to preserve scenic rock formations commonly referred to as the "Wonderland of Rocks" (Picture of Wonderland). Until 1934, when the Monument was transferred to the National Park Service, the Monument was under the jurisdiction of the U.S. Forest Service. The Monument contains 11,985 acres, of which 10,290 acres are designated as a wilderness area. Temperatures are moderate, with a January mean of 40 degrees F. and a July mean of 74 degrees F. Annual precipitation is about 18 inches, mostly in July and August, with occasional snow during winter months. The Monument's topography is very rugged. Two deep canyons and their tributaries cut into high rhyolitic tuff plateaus. Erosion controlled by vertical fractures through the tiff formation has created numerous giant columns, slender spires, and an array of unusually shaped rocks, some of which appear balanced precariously on small pedestals. Local outcrops of gneiss, schist, and limestone create varied soil types. The Monument is located on the northwest flank of the Chiricahua Mountains, which are the southern extension of the Basin and Range Province, extending between the Colorado Plateau and the Sierra Madre Mountains of Mexico. The Monument ranges in elevation from 5,160 ft. to 7,825 ft. while the highest point in the range is 9,792 ft. Biogeographic location and terrain roughness combine to produce high plant species richness (Bennett and Kunzmann, 1992). The number of species in the monument are approximately 680, of which 30 are trees, 92 are shrubs and 317 are herbaceous perennials.
Professional management requires accurate and comprehensive information about park resources. Perhaps the most basic information for managing natural-resource areas is a thorough knowledge of vegetation, an undisputed resource of scenic beauty, wildlife habitat, and overall ecosystem function. This vegetation project is listed as one of the "highest priority" items in the Monument's Resource Management Plan. The priority assignment reflects the primary importance of "natural/native" vegetation and concerns over the extent to which human activities may have altered the natural vegetation. Throughout time, such activities have included overgrazing, interruption of natural fire cycles (Bennett 1992; Swetnam et al. 1991), initiation of prescribed burning programs, and a wide range of recreational uses. The Monument may be experiencing additional influences from population increases in surrounding towns, regional air pollution, and possible impacts from climate change. Rapid vegetation changes in Southeastern Arizona ecosystems have been well documented (Nichol, 1937; Reynold and Bohning, 1956; Hastings and Turner, 1965; Bahre, 1991).
The degree and relative influences of some human activities may be adequately ascertained with a detailed map of actual vegetation resources. The best detailed vegetation map for the Monument, which used a variety of traditional land surveying techniques, was done in 1939 by Roseberry and Dole. One of our first project priorities was to transform this historic (paper) vegetation map (and other maps) to a standardized Brown, Lowe, and Pase-compatible ArcInfo GIS format (historic 1939 Map) . To assess the vegetation changes at the Monument, a multi-disciplinary approach would be used. It was determined that the following steps would be initiated to better understand the vegetation of the monument: (1) normalize various vegetation classifications systems to a Brown, Lowe, and Pase standard using available historic and GIS datasets, (2) evaluate the Roseberry and Dole 1939 vegetation map along with associated vegetation transects and photo points, (3) evaluate the usefulness of existing Landsat Thematic Mapper imagery and commonly used remote sensing processing techniques and, (4) collect additional vegetation-community information in the field as necessary, using GPS/GIS technologies to delineate and ground-truth vegetation classes and evaluate vegetation change. Some of the final products would include: (1) a natural-color, ortho-rectified, digital air-photo image of the Monument, (2) a "terrain corrected" set of satellite imagery, (3) BLP classified vegetation maps complete with polygon vegetation-plot summaries, (4) an evaluation and comparison of 1939 and 1996 vegetation changes, and (5) an evaluation of which remote sensing Thematic Mapper classification techniques, if any, would work best in such a complex area. The products created by this project will (1) serve as a tools for making informed management decisions, (2) serve as baseline datasets for future studies, (3) be used for monitoring the effects of natural fires and prescribed burning programs, (4) serve as an aid in the evaluation of potential wildlife habitat, and (5) be used to delineate observable vegetation changes. GIS and GPS mapping technologies have been critical tools in achieving these goals in a timely and cost-effective way.
The vegetation of the area is diverse, and is a consequence of a broad range of factors including elevation, topography, temperature, precipitation, geology, soil, fire, and an assortment of anthropogenic effects. Examples of prominent vegetation biomes by BLP number include: Madrean Montane Conifer Forest (122.6), Madrean Evergreen Forest and Woodland (123.3), Relict Conifer Forest and Woodland (123.5), Interior Chaparral (133.3), Semidesert Grassland (143.1), and Interior Southwestern Riparian Deciduous Forest and Woodland (223.2) (Brown, Lowe, and Pase, 1979). Typical vegetation series (biome subdivisions) found are: Grama Grass-Scrub Series (Bouteloua spp., Hilaria belangeri), Manzanita Series (Arctostaphylos pungens, Quercus toumeyi), Encinal Series (Q. emoryi, Q. arizonica, Juniperus deppeana), Pine-oak Series (Pinus leiophylla, P. engelmanii, P. cembroides), Douglas Fir-Mixed Conifer Series (Ponderosa Pine, Douglas Fir), and Cypress Series (Cupressus arizonica). Detailed descriptions of the vegetation within the Monument can be found in Roseberry and Dole (1939), Moir (1974), and Reeves (1976). Madrean Evergreen Woodland evolved during the Miocene in southern United States and the Sierra Madre of Mexico. Although the Madrean Woodland arose at approximately the same time period as the northern conifer forest and California scrublands, it had a very different developmental history (Bennett and Kunzmann, 1992). Madrean woodland communities are about 65% of the Monument's vegetation cover.
The Civilian Conservation Corps (CCC) prepared vegetation maps for the national parks and monuments of the west. This work was begun about 1935 and continued until the outbreak of World War II. The Roseberry and Dole 1939 fieldwork was undertaken to meet an ever increasing need for a map showing the composition, extent, and character of local vegetation.
The fieldwork at Chiricahua National Monument was done in the spring of 1937 by R.D. Roseberry and N.E. Dole, Jr., CCC Junior Foresters. Their "type" map was constructed by sketching in the apparent vegetation boundaries on a topographic map as seen from a vantage point where a good view of the adjoining country could be obtained. Topographic details and features were used as horizontal and vertical controls. Aerial photographs were not employed, and all community classification was made by direct observation.
Vegetation plot information was recorded on 23 sites within the Monument. These plots listed most of the common cover species, and were used to determine the presence and extent of understory species not appearing on the community "type" map. Twenty-three sample plots and 31 photographs were located on the reference map.
The final map was a remarkable achievement, especially considering the meager time and resources available to these scientists. Our comments are not intended to denigrate their work. However, their map has some shortfalls in comparison with modern vegetation maps. Only such vegetation as could be seen from above or recognized from a distance was counted in determining the types. The 23 sample plots are too few to adequately represent the vegetation monument-wide. Not noted on the map are the understory species that cannot be seen from above. In spite of these concerns, the historic map, vegetation plots, and photographs were adopted as the historic vegetation-community baseline for the derived GIS model.
The Roseberry and Dole vegetation classification was strongly influenced by utilitarian rather than theoretical concerns. Thus, they sharply distinguished between stands of trees large enough to harvest and those that were not. The vegetation was divided by general physiognomic types; i.e., grassland, shrubs, and trees, or combinations of these assigned into vegetation classes by less than or greater 20% cover. The Roseberry and Dole classification is non-evolutionary and not based on biomes, since trees of the Madrean Biome (Pinus arizonica (Arizona Pine), P. engelmannii (Apache Pine) and P. leiophylla (Chihuahua Pine) are indiscriminately included with trees from the Rocky Mountain Biome P. ponderosa (Western Yellow Pine) and Pseudotsuga menziesii (Douglas-fir). Cupressus arizonica (Arizona Cypress), included in their Timber Type, is evolutionarily unrelated to either the Rocky Mountain or Madrean biomes. The other major Roseberry and Dole classification units are more satisfactory, by current standards. The chaparral types correspond to the Interior Chaparral biome, and the woodland type is now recognized as oak woodlands of Madrean origin.
No system of classification is perfect but the Brown, Lowe, and Pase (1979) ("BLP") vegetation classification system has features desirable for GIS vegetation themes and for the GAP Program. Since it is benchmarked on the biome, it permits development of a hierarchial evolutionarily related classification well suited to mapping extensive areas for assessment of animal-plant distributions. Biomes are natural communities of plants and animals characterized by a distinctive vegetation physiognomy within a formation (forest, scrubland, grassland, etc.). Building a classification based on the biome (IUCN, 1974) addresses some of vegetation-classification issues raised by Gleason (1926) and many others. We emphasize that the BLP system was intended by its authors to be a means or pattern for organizing a classification, and was not intended to be a comprehensive "type" classification index by its authors.
Funded through the National Biological Service's Cooperative Park Studies Unit (NBS-CPSU) at The University of Arizona, Warren et al. (1980, 1982, 1992) developed, tested, and applied the prominence value concept for elaboration of the BLP examples into a classification suitable for mapping large areas. The prominence value is a rating that combines estimated dominance, biomass, and commonness. Prominence values are the most important part of vegetation description for classification and mapping (Colorado Plateau Vegetation Advisory Committee 1992). Prominence values are a concept distinct from, but related to, importance values and other qualitative measures described by Braun-Blanquet (1968) and others. Appropriate uses of prominence values are intuitive and become apparent upon examination of our vegetation field forms. Automated mapping at Warren's sophisticated level of classification, or detection of successional vegetation change, if feasible, will most likely require substantial software "training" based on known polygonal information, GIS analysis, and perhaps computer-based neural-net analysis.
The problems with vegetation classification are formidable. Discrimination between the physiognomically similar oval-crowned Apache, Arizona, and Western Yellow Pines may not be possible with remote sensing techniques. Similarly, discrimination between the ecologically dissimilar species of Ceanothus (buckthorn) or between buckthorn and Cercocarpus (mountain mahogany) is difficult with low-altitude stereo photographs and may be impossible using Thematic Mapper type data. However, such detailed, species-specific discriminations are often requested and are thought necessary by today's natural resource manager.
The vegetation mapping effort involves numerous steps: (1) the creation and interpretation of 1:12,000 natural-color aerial photography, (2) interpretation of the Roseberry and Dole 1939 map polygons, and associated transect data to a standardized but preliminary BLP vegetation classification, (3) classification and interpretation of satellite imagery, (4) creation of an assortment of biotic and abiotic ArcInfo polygon coverages and corresponding relational datasets, to assist in vegetation-community classification, (5) determination of an appropriate stratified field sampling design, and (6) establishment of an effective, rapid ground truthing program using GPS/GeoLink in field mapping techniques, with ArcInfo-generated background coverages, grids, and images. GPS/Geolink mapping technologies are extremely useful in the field to: (1) relocate historic vegetation plots and map polygons, (2) check the accuracy of vegetation classes and associated polygons, (3) increase the number of vegetation plots to better classify vegetation polygons, (4) check the results of supervised and unsupervised satellite-based spectral classifications, and (5) create new geo-relational datasets. Fieldwork is used to relate actual vegetation associations with image features, as delineated on imagery during preliminary interpretation, or with various vegetation coverages in the GIS. In general, fieldwork is required to refine identifiable vegetation associations on the basis of more subtle discriminations than were used to perform the preliminary image classifications.
Our task was to use Roseberry and Dole's map as the "historic truth" for measuring vegetation change and for testing various remote-sensing classification techniques. Specifically, can multi-band Thematic Mapper data be manipulated to yield sufficient information useful to produce a detailed, community-based vegetation map for the resource manager? What combination of techniques is best suited for each level of classification detail? Because the 1939 classification is prominently physiognomic; i.e., chaparral, woodland, forest, barren, etc., there are indications that perhaps TM data could be filtered to discriminate between different vegetation textures at the biome and perhaps the series levels. The Roseberry and Dole map represented an easier test model than community-based maps generated from sophisticated, species-based classification systems such as those of Warren et al. (1980, 1982, 1992).
Multispectral data from Landsat satellites have been used for land-cover mapping for more than two decades. Six of the seven bands of Landsat Thematic Mapper (TM) data were used in this study, covering reflective wavelengths from visible blue, green, and red, to near and mid infrared; the thermal infrared band, which has lower spatial resolution, was not used. Chiricahua National Monument encompassed 298 by 232 pixels within the TM scene (acquired on 13 June 1993), an area approximately 8.5 km by 6.6 km. Although the scene had been commercially "terrain-corrected," or geometrically rectified and registered to USGS 3 arc-second DTM data, a small additional linear offset was necessary to register it with other thematic layers.
Each pixel of a TM image represents an integrated signal from all ground surfaces within a 30 by 30 meter area, including soil, litter, and rock background as well as vegetation. Radiance measurements recorded by the satellite sensor are influenced not only by proportions and structure of each surface-cover type, but also by other factors unrelated to surface cover such as illumination angle. Contributions by all factors must be taken into account when relating reflectance data to surface cover (Smith et al., 1990, and others). The magnitudes and relative importance of variables which affect radiance measurements vary spatially within and between scenes, and temporally with season and time of day. This implies that calibration and corrections performed on spectral data are scene-dependent. There are no automated procedures to extract only the information related to cover type.
Vegetation distribution in Chiricahua National Monument was examined using a number of image-processing techniques, summarized in Table 1. These methods have been developed primarily for use in agricultural and forest areas, where cover types may be more homogeneous than in natural semi-arid settings, and thus few pixels would represent complex mixtures of vegetation and background (Smith et al., 1990, and others). Vegetation index images ( NDVI = Normalized Difference Vegetation Index; Tucker, 1979; SAVI = Soil-Adjusted Vegetation Index; Huete, 1988) showed similar patterns related to percentage of vegetation cover, as did spectral-band ratios (bands 4/3, 2/1, 5/2). Principal-component analysis produced mutually uncorrelated images which were linear combinations of original spectral data; their physical significance, however, was not clear. Since no training site data were available, it was not possible to perform supervised image classification for cover type mapping.
| Data type | Advantages | Disadvantages |
| Landsat Thematic Mapper image | Complete coverage for US; 18 day repeat; 34,000 km per scene; 7 spectral bands | May be expensive; atmospheric correction difficult; resolution too coarse for detailed vegetation studies |
| Image processing method | Advantages | Disadvantages |
| Vegetation indices (NDVI, SAVI) | Based on only two spectral bands; simple to compute; widely used and understood | May require data preprocessing including atmospheric correction; sensitive to topographic shadowing; related to percent vegetation cover, not vegetation type |
| Spectral band ratios | May remove topographic shadows; enhance spectral features of interest | Sensitive to inadequate atmospheric correction; detector striping noise may be enhanced |
| Principal components analysis | Included in image processing software; reduces data volume to classify | May be difficult to interpret physical meaning of results |
| Supervised classification | Spectral classes correspond to known training sites; can use ancillary information | Must have ground data for training sites; training sites may not be representative, or may be too heterogeneous to define unique spectral classes; vegetation classes may be too spectrally heterogeneous to classify successfully |
| Unsupervised classification | No information about scene needed; only variable is number of classes | Spectral classes primarily related to topographic illumination and percent vegetation cover |
Initial results from unsupervised classification of spectral band data indicated that variations in illumination, due to interaction of sun angle and topography, dominated the classification, so a first-order topographic normalization was performed using a DEM generated from digitized topography (Civco, 1989). Unsupervised classification of normalized spectral data was then carried out for 6, 12, and 24 classes; however, the identity of resulting thematic classes could be determined only by field verification. In a preliminary field check of the six-class map, four classes corresponded to structural vegetation groups, such as semi-desert grassland or conifer forest. The remaining two classes appeared to be complex mixtures of vegetation, shadows, soils, and/or the scenic rock spires for which the Monument is noted. The same spectral class represented different cover combinations at different locations. Although using more classes may separate these heterogeneous groups into consistent subclasses, extensive field checking would still be required to verify that each spectral class represented the same vegetation in each polygon.
To analyze vegetation change over the past 60 years we are examining the complex relationships between the 1939 Roseberry and Dole map; new BLP classified maps; airborne color photography and the derived ortho-rectified, digital, color mosaic; and remote-sensing imagery. To describe landscape patterns between thematic GIS datasets and imagery, numerous ArcInfo GIS analyses and remote-sensing procedures are used. The results of these analyses are converted into field-compatible background maps (coverages) which can easily be validated. By using GPS and a computer laptop in the field as a means of survey control, we can visually determine the spatial accuracy and classification efficiency of vegetation thematic datasets and image classifications. To examine and better delineate polygon edge issues, which reflect the inherent nature of "fuzzy" (Zadeh 1965) vegetation datasets, additional geo-relational datasets are created in the field as necessary. The longer-term goal is to use newly acquired field data to update GIS coverages and attributes, then through a reiterative process, refine classification rules and algorithms to improve overall accuracy of our vegetation classification-prediction models. During initial field trials, it was determined that the Roseberry and Dole map was a relatively good model to use, considering the vegetation detail and overall classification accuracy. This--coupled with more representative, species-specific field data on a wider variety of Roseberry and Dole vegetation types and a significant concomitant increase in the number of representative polygons for each vegetation class and cover type--allows us to refine earlier unsupervised and develop supervised remote sensing vegetation classifications that should better predict historic vegetation classes.
It has been our experience that it is very difficult to utilize "off-the-shelf" data or imagery to create detailed, spatially accurate, BLP series-level vegetation- community classifications in complex southwest arid terrains without extensive field calibration, intensive ground-truthing, and advanced computer technologies such as GPS and GIS. Even with integrated technologies such as GPS and GIS, the degree may be severely limited to which locally based sensor calibrations, vegetation community rules, processing algorithms, and classification schema may be transferable to other geographic localities. The primary benefit of these technologies and techniques is that it is much easier to capture, ground-truth, and process information to produce a highly detailed series-level vegetation map, complete with substantial tabular data to support classification decisions and schema. After all, maps with only polygon labels are of limited value and use. Without recent advances in GIS and GPS technologies, increases in remote-sensing resolution, and integrated computer mapping techniques, the creation of vegetation maps would be more difficult and less efficient.
The authors would like to express their appreciation to: The National Park Service, for funding; The Monument staff, for their encouragement; and our corporate sponsors. Motorola of Scottsdale, Arizona, generously donated the Global Positioning System receivers used on this project (and others), and GeoResearch Inc. of Billings, Montana, for the donation of GeoLink Automated Mapping Software. With state-of-the-art tools, this project was done more efficiently.
Bahre, C. 1991. Legacy of Change. Tucson, AZ: The University of Arizona Press,
Bennett, P.S., and M.R. Kunzmann. 1992. The applicability of generalized fire prescriptions to burning Madrean evergreen forest and woodland. J. Ariz.-Nev. Acad. Sci. 24-25:79-84.
Bennett, P.S., and M.R. Kunzmann. 1992. Factors affecting plant species richness in the Madrean Archipelago north of Mexico. Chiricahua Mountains Research Symposium Proceedings, Eds. A.M. Barton and S. A. Sloan, Methodist Camp, Chiricahua Mountains, AZ, 16 March 1992. Tucson, AZ: Southwestern Monuments Assn.
Braun-Blanquet, J. 1965. Plant Sociology: The Study of Plant Communities. (Translated, revised, and edited by C.D. Fuller and H.S. Conrad) London: Hafner, 1965.
B Brown, D.E., C.H. Lowe, and C.P. Pase. 1979. A digitized classification system for the biotic communities of North America, with community (series) and association examples for the Southwest. J. Ariz. Nev. Acad. Sci. 14 (suppl. 1): 1-16.
Civco, D.L. 1989. Topographic normalization of Landsat Thematic Mapper digital imagery. Photogrammetric Engineering and Remote Sensing 55(9):1303-1309.
Colorado Plateau Vegetation Advisory Committee. 1992. Colorado Plateau Vegetation Advisory Committee Delineation Manual. Tucson, AZ: NBS Cooperative Park Studies Unit, The University of Arizona,
Gleason, H.A. 1926. The individualistic concept of the plant association. Bull. Torrey Bot. Club 53:7-26.
Hastings, J.R., and R.M. Turner. 1965. The Changing Mile: An Ecological Study of Vegetation Change with Time in the Lower Mile of an Arid and Semiarid Region. Tucson, AZ: The University of Arizona Press,
Huete, A.R. 1988. A soil-adjusted vegetation index (SAVI). Remote Sensing of Environment 25:295-309.
I.U.C.N. (International Union for Conservation of Nature and Natural Resources). 1974. Biotic Provinces of the World--Further Development of a System For Defining and Classifying Natural Regions for Purposes of Conservation. IUCN: Brussels. I.U.C.N. Occasional Paper 9.
Moir, W.H. 1974. Resources monitoring system (Chiricahua National Monument). Chiricahua National Monument. Unpublished report.
Mouat, D.A., K.L. Reichhardt, and P.L. Warren. 1981. The Vegetation of Grand Canyon National Park. Applied Remote Sensing Program, Office of Arid Lands Studies, The University of Arizona, Tucson, AZ.
Nichol, A.A. 1937. The natural vegetation of Arizona. Tucson, AZ: The University of Arizona College of Agriculture Tech. Bull. 68.
Reeves, T. 1976. Vegetation and flora of Chiricahua National Monument, Cochise County, Arizona. Master's thesis, Arizona State University, Tempe, AZ. 180 pages.
Reynolds, H.G., and J.W. Bohning. 1956. Effects of burning on a desert grass-shrub range in southern Arizona. Ecology 37:769-777.
Roseberry, R.D., and N.E. Dole. 1939. Vegetation type survey of Chiricahua National Monument--National Park Service. Branch of Forestry, San Francisco, CA. Unpublished report.
Swetnam, T.W., C.H. Baisan, P.M. Brown, and A.C. Caprio. 1989. Fire history of Rhyolite Canyon, Chiricahua National Monument. Tucson, AZ: NBS Cooperative Park Studies Unit, The University of Arizona. Technical Report No. 32, 47 pages.
Smith, M.O., S.L. Ustin, J.B. Adams, and A.R. Gillespie. 1990. Vegetation in deserts: Part I. A regional measure of abundance from multispectral images. Remote Sensing of Environment 31:1-26.
Tucker, C.J. 1979. Red and photographic infrared linear combinations for monitoring vegetation. Remote Sensing of Environment 8:127-150.
Turner, M.G., F. Sklar, and R. Costanza. 1989. New methods to compare patterns in landscape modeling and analysis. Ecological Modeling 48 (1989):1-18.
Warren, P.L., J.E. Bowers, B.D. Treadwell, and K.L. Reichhardt. 1980. Vegetation of Organ Pipe Cactus National Monument. Tucson, AZ: NBS Cooperative Park Studies Unit. Technical Report No. 8, 79 pages.
Warren, P.L., K.L. Reichardt, D.A. Mouat, B.T. Brown, and R. Roy Johnson. 1982. Vegetation of Grand Canyon National Park. Tucson, AZ: NBS Cooperative Park Studies Unit. Technical Report No. 9, 140 pages.
Warren, P.L., M.S. Hoy, and W.E. Hoy. 1992. Vegetation and flora of Fort Bowie National Historic Site, Arizona. Tucson, AZ: NBS Cooperative Park Studies Unit. Technical Report No. 43, 78 pages.
Zadeh, L.A. 1965. Fuzzy Sets. Information and Control, 8:338-353.
Michael R. Kunzmann is an ecologist for the U.S. National Biological Service Cooperative Park Studies Unit (NBS-CPSU) and is a participant in the Advanced Resources Technology Program at the University of Arizona. Correspondence may be sent to: NBS-CPSU, 125 Biological Sciences East, The University of Arizona, Tucson, AZ. 85721, phone (520) 621-7282, Internet email: MRSK@npscpsu.srnr.arizona.edu
Peter S. Bennett is a Research Scientist for the U.S. National Biological Service Cooperative Park Studies Unit locate at the Western Archeological and Conservation Center, 615 North 6th Avenue, Tucson, AZ. 85719. Correspondence may be sent to: NBS-CPSU, 125 Biological Sciences East, The University of Arizona, Tucson, AZ. 85721 or by phone (520) 670-6896.
Susan M. Skirvin, a graduate student in The University of Arizona Arid Lands Resource Science program, works as a Remote Sensing Specialist in the Advanced Resources Technology (ART) Program in the School of Renewable Natural Resources (SRNR). Correspondence may be sent to: ART Program (SRNR), 203 Biological Sciences East, The University of Arizona, Tucson, AZ, 85721.
Craig A. Wissler is an ART GIS Coordinator in the Advanced Resources Technology Program in the School of Renewable Natural Resources (SRNR). Correspondence may be sent to: ART Program (SRNR), 203 Biological Sciences East, The University of Arizona, Tucson, AZ, 85721, phone (520) 621-9588, or by Internet email: Craig@nexus.srnr.arizona.edu.
