CAUSES AND CONSEQUENCES OF LAND COVER CHANGE IN A GREATER ECOSYSTEM: TREND AND RISK ASSESSMENT, MONITORING, AND OUTREACH

ANDREW HANSEN, ALISA GALLANT, JAY ROTELLA Biology Department, Montana State University Bozeman, MT WARREN COHEN USDA Forest Service Forestry Sciences Laboratory Corvallis, OR JERRY JOHNSON Political Science Department, Montana State University Bozeman, MT BRUCE MAXWELL Plant, Soil and Environmental Sciences Department, Montana State University Bozeman, MT RAY RASKER Earth Sciences Department, Montana State University Bozeman, MT

"Greater ecosystems" are often areas where nature reserves are surrounded by gradients in land use. Thought of as refugia for biodiversity, greater ecosystems around the world now have rapid human population growth. Thus, the ecological characteristics that attract human immigrants to greater ecosystems may be at risk from increased human land use. Therefore, we are studying the interactions between the ecological and human community of the Greater Yellowstone Ecosystem.

Objectives

1. Quantify changes in natural and human forcing functions, ecological processes, land cover and use, and ecological responses across the GYE from 1973-1996.
2. Test hypotheses on forcing functions, ecological processes, land cover, and ecological responses:
1. High elevation and patchy distributions of topography, climate, and soil cause net primary productivity (NPP) to be low and variable over most of the GYE.
2. Species abundance and richness are correlated with abiotic gradients and NPP, and are high only in localized hot spots across the landscape.
3. Human population growth is most rapid in counties with high opportunities for economic diversification, due in part to ecosystem-based businesses.
4. Human land use is correlated with environmental gradients such that land use is most intense at hot spots for ecological productivity and biodiversity.
5. Intense land use reduces NPP and native bird abundance, diversity and nest success.
3. Assess current and future risk to ecological hot spots and potential for restoration.
4. Develop and implement an approach to monitor ecological and human interactions.
5. Communicate results to stakeholders via workshops, publications, and decision-support tools.

Land cover/use and other key human and ecological variables will be quantified for 1995 using satellite, field, and in situ data. An innovative change-detection approach will be tested and used to create land cover/use maps at four previous time periods back to 1973 (year of earliest available satellite data). Accuracy assessments will be done on all resulting maps using aerial photos and field data.

Hypotheses will be tested primarily by multivariate correlation analyses across space and time. Some hypotheses are also being tested with field experiments under other funding. The results will be used to determine the places in the landscape most important to biodiversity and under the greatest risk from changing land use. Places with high potential for restoration will be identified. Feedbacks from the ecosystem to the human community will also be examined. Continued monitoring is essential for ecosystem management, and we will develop and demonstrate an approach for taking the ecological and human "pulse" of the GYE in the future. Finally, outreach projects will be designed to quickly communicate results of the study to stakeholders and policy makers.

This study is unique in testing important new ecological theory concerning abiotic controls on biodiversity. It also will directly examine the notion that human communities are closely tied to the surrounding ecosystem. The apparent mismatch between ecological and administrative boundaries may underlie many of the resource conflicts in the GYE. The results will provide a basis for managing the GYE to sustain both ecological and human communities. Our methods will be exportable to other greater ecosystems and, hopefully, will facilitate sustainable management strategies before these refugia for biodiversity are jeopardized by unplanned human-population growth.

Originally describing the range of the Yellowstone grizzly bear, "greater ecosystem" has come to denote an area defined by especially strong linkages between ecological and human communities. Greater ecosystems often consist of nature reserves surrounded by gradients in human density and land use. Typically thought of as refugia for natural processes and biodiversity, these ecosystems are now undergoing dramatic increases in human population density. We have suggested that the Greater Yellowstone Ecosystem (GYE) (depicted at left), and other greater ecosystems around the world share some biophysical properties that cause them to be very sensitive to human density. Knowledge of the linkages among biophysical factors, biodiversity, and human land use is needed to design strategies to sustain both biodiversity and human communities.

Abiotic factors

With funding from USDA National Research Initiative, we have been studying bird diversity and demography across gradients in abiotic factors, cover type, and vegetation structure. Preliminary results suggest that abiotic factors associated with elevation strongly influence bird abundance and diversity. Within a forest type (mature and old-growth lodgepole pine), we found that bird abundance and richness were more than twice as high at lower elevations than at higher elevations.

We hypothesize that either NPP or climate is limiting to birds at higher elevations, and are now collecting data to test these possibilities. Across all cover types we found that cottonwood, aspen, and willow communities had much higher bird abundance and richness than other stand types. We suspect this is because these are high both in NPP and in structural complexity, providing both high levels of energy availability and habitat diversity. Spatial patterns in abiotic factors and vegetation result in the abundance of several species and species richness being high only in localized hot spots. These hot spots may be population source areas for some species and be critical to maintaining population viability over the GYE.

The upper image depicts shaded relief for the Greater Yellowstone Ecosystem. The lower left image portrays the abundance of yellow warbler (Dendroica petechia) in the northwest portion of the Greater Yellowstone Ecosystem. The abundance values were produced from field data based on cover type, seral stage, and elevation. This species specializes on aspen, cottonwood, and willow habitats, which are patchily distributed in the study area. Notice the patchy nature of suitable habitats (red patches) for this species. The lower right image displays predicted bird species richness based on cover type, seral stage, and elevation. Darker shades of red indicate greater predicted richness. Hot spots for bird richness are small in area and found mostly at lower elevations.

Human Population Growth

Like birds, people prefer to occupy certain places in the landscape. Homesteaders chose the more productive sites, consequently private lands occupy lower elevation, more productive settings, while nature reserves are in harsher portions of the landscape. We speculate this trend continues today with the immigrants who come to the GYE because of the quality of the ecosystem, settling in and near hot spots for biodiversity. Riparian zones and aspen groves on mountain toe slopes are among the most popular settings for home construction. Beyond conversion of these hot spot habitats, human settlement may also have more subtle effects on biodiversity. We found that bird reproductive success was extremely low in hot spot habitats near human development relative to hot spots in more natural habitats, possibly because predators and nest parasites are very dense in human landscapes. So, it appears that WHERE humans choose to settle in the landscape may strongly influence native species in subtle (trickle-down effects like changing nest success) and not so subtle (converting native habitats into subdivisions) ways.

The figure displayed above illustrates the Normalized Difference Vegetation Index (NDVI) derived from AVHRR data for June 9-22, 1995, for the GYE. NDVI is correlated with primary productivity. Notice that NDVI is typically lower (lighter color) on federal lands (outlined in black) than in the surrounding private lands of the lower elevations.

Goal of study

The GYE is probably typical of many greater ecosystems surrounding nature reserves. Abiotic factors result in biodiversity and intense human land use overlapping on private lands outside of nature reserves. Thus, these reserves, typically thought of as refugia for biodiversity, may be insufficient for maintaining native species. The goal of this study is to better understand these linkages between biodiversity and land use. Hopefully, with this knowledge, decision makers can find ways to better sustain both native species and the growing human community in the GYE.

A conceptual model for the study is presented in the following figure.

Land cover is strongly influenced by interactions among abiotic factors, natural disturbance, and succession. The spatial patterning of these abiotic factors causes land cover to be heterogeneous with only localized hot spots where net primary productivity (NPP) and species diversity are high. Human socioeconomic factors interact with abiotic factors such that land use is most intense within these ecological hot spots. Consequently, land use is more strongly influencing biodiversity and productivity than would be expected based on human density alone. The state of the ecosystem, in turn, feeds back to influence land use and the human community.

Objectives

1. Quantify change in natural (climate, terrain, soils) and human (socioeconomic) forcing functions, ecological processes (disturbance, succession), land cover and use, and ecological responses (NPP and biodiversity) across the GYE from 1973-1995.
2. Better understand interactions among forcing functions, ecological processes, land cover, and ecological responses by testing hypotheses:
1. H1: High elevation and patchy distributions of topography, climate, and soil cause NPP to be low and variable over most of the GYE.
2. H2: Species abundance and richness are correlated with abiotic gradients and NPP and are high only in localized hot spots across the landscape.
3. H3: Areas with the greatest extent of ecological hot spots have the highest socioeconomic performance.
4. H4: Human land use is most intense at hot spots for ecological productivity and biodiversity.
1. H4.1: Counties with the greatest area of ecological hot spots will have the highest human growth rates.
2. H4.2: Rural residential development has preferentially occurred in sites high in NPP and biodiversity.
5. H5: Intense land use reduces NPP and native bird abundance, diversity and nest success.
3. Perform a "risk assessment" to determine the places now most important for biodiversity and/or ecological productivity or have the greatest potential for restoration.
4. Develop and implement an approach to monitor natural and human forcing functions, ecological processes, land use and cover, and ecological responses in the future.
5. Rapidly communicate results to stakeholders and environmental decision-makers.

OBJECTIVE 1: CHANGE DETECTION (1973-1996)

Key variables will be quantified at various time intervals from 1973 to 1996 using satellite data, field data, census reports and other sources (Tables 1 and 2).

Table 1. Ecological variables to be quantified over the period 1973-96 in this study.

Table 2. Socioeconomic variables to be quantified over the period 1973-96 in this study.

Land Cover

Land cover will be classified hierarchically. An accuracy assessment will be done at each level (using aerial photos and a random sampling design). This way, we can select the classification level that has the best mix of cover type specificity and accuracy for each analysis. Land cover for 1995 will be mapped using Landsat TM data for two to three dates. A combination of unsupervised and supervised classification techniques will be used to delineate and label cover classes. The classification algorithms will be calibrated with training data from aerial photos and terrain data from DEMs.

Our goal is to produce land cover maps at 5 or 6-year intervals from 1973-1995. To do this as efficiently and accurately as possible, we will test a method that starts with the 1995 cover map and works backward in time, evaluating spectral change at each time interval, relative to 1995 cover. We will identify multispectral-difference thresholds that represent specific types of change. For areas identified as "unchanged," we will assign the cover class label from the 1995 map. For areas identified as "changed," we will derive a cover class label based on classifying pixels from the earlier image. Because this new approach is untested, we will apply it over sample areas and compare results with those obtained from post-classification change assessment. Error rates will be ascertained using reference air-photo data.

Change detection will be carried out using TM sensor data for the 1984, 1990, and 1995 sample periods. MSS data will be used for the periods prior to the availability of TM data (1978 and 1973), as well as for the 1984 sample period. Therefore, we will have data from both sensors for 1984 to aid in calibrating the cover type signatures between the two data types. We will use random sampling procedures to validate change classes.

Land Use

Land use will be mapped based on land cover, ownership, and other available sources.

Climate

Simulation models, such as MT-CLIM or PRISM, will be used to extrapolate climate variables over the landscape based on data from meteorological stations.

Net Primary Productivity

Three methods of quantifying NPP are being compared under NASA EPSCoR funding: simulation with model FOREST-BGC, GIS-based extrapolation from field data, and classification based on spectral reflectance. We will use the method that is most accurate for our study area.

Species diversity

We will focus on species of birds, trees, and shrubs because we have field data on these taxonomics groups from a USDA study. These were selected because many species of many life histories can be sampled efficiently. Ungulates and vascular plants may also be considered. Key response variables will be species abundance, species richness, and for birds, population source/sink dynamics for some species. These will be extrapolated over the GYE using statistical functions derived from field study in the original study area.

Socioeconomics

Key data are in Table 2. These will be derived primarily from census and economic reports.

Analyses

ANOVA will be used to determine the statistical significance of change across space and over the 6 time periods. Transition probabilities of land cover and use will be calculated.

OBJECTIVE 2: HYPOTHESIS TESTING

Detailed methods have been formulated for evaluating each hypothesis. Generally, these methods involve statistical comparisons among time periods, spatial locations, and variables.

OBJECTIVE 3: RISK ASSESSMENT

We define risk as the likelihood that the most important places in the landscape for ecological productivity and biodiversity will be negatively influenced by human activity. First, an index of habitat quality will be generated for each location based on abiotic factors, vegetation structure, and species abundance and diversity. Then, current land use will be compared to habitat quality to identify places of high habitat value that now have intense human land use. This will indicate the places that would most benefit from immediate conservation and management attention. Likelihood of future threat will be predicted using land-cover transition probabilities from 1973-1995 to predict land-use trends for 1995-2015. The projected land use patterns for 2015 will be compared with current habitat quality to predict the ecological hot spots most likely to come under threat over the next two decades. Finally, potential for ecological restoration will be mapped.

OBJECTIVE 4: MONITORING

The methods developed in the study will be used as the basis of a monitoring approach that can be applied periodically to "take the ecological and socioeconomic pulse" of the GYE. This will be implemented using data from 1998. New hyperspectral imagery will hopefully be available for this effort.

OBJECTIVE 5: OUTREACH

Data and knowledge from this study will be communicated to stakeholders in four ways. First, all data generated will be managed, housed, and distributed at the Data Clearinghouse of the Mountain Research Center at Montana State University. Second, a GYE atlas will be produced both in hard copy and CD ROM that includes key maps, charts, and figures resulting from the study. Third, we will conduct workshops annually at the Yellowstone Seminar for Local Officials. Finally, the LUCCPS decision-support system will be made available to local officials to help them visualize future land cover/use if current trends continue. Hopefully, these efforts will allow the study results to quickly be considered relative to public policy.

ANDREW HANSEN - Abiotic factors, biodiversity, risk assessment, project leader
Biology Department, Montana State University, Bozeman, MT 59717
Voice: (406) 994-6046 FAX: (406) 994-3190
email:hansen@montana.edu

WARREN COHEN - Remote Sensing and vegetation classification
USDA Forest Service, Forestry Sciences Laboratory, Corvallis, OR 97331
Voice: (541) 750-7322
email:cohenw@ccmail.orst.edu

ALISA GALLANT - Remote Sensing and vegetation classification
Biology Department, Montana State University, Bozeman, MT 59717

JERRY JOHNSON – Socioeconomic analyses
Political Science Department, Montana State University, Bozeman, MT 59717
Voice: (406) 994-5164
email:upojj@msu.oscs.montana.edu

BRUCE MAXWELL – Land-use decision support modeling
Plant, Soil, and Environmental Science Department, Montana State University, Bozeman, MT 59717
Voice: (406) 994-5717 FAX: (406) 994-3933
email:ussbm@msu.oscs.montana.edu

RAY RASKER – Socioeconomic analyses
Earth Sciences Department, Montana State University, Bozeman, MT 59717
Voice: (406) 587-7331
email:ray@sonoran.org

JAY ROTELLA - Population viability modeling
Biology Department, Montana State University, Bozeman, MT 59717
Voice: (406) 994-5676 FAX: (406) 994-3190
email:rotella@montana.edu

Egbert, S.L. 1997. Modeling land cover on the high plains using multi-seasonal thematic mapper imagery.

Egbert, S.L., R. Lee, K. P. Price, M.D. Nellis, R. Boyce. In Review. Mapping conservation reserve program lands using multi-seasonal thematic mapper imagery.

Hansen, A.J., and J. Rotella. In Prep. Environmental heterogeneity and biodiversity. In M. Hunter, Jr., ed., Maintaining Biodiversity in Forest Ecosystems. Island Press, New York, USA.

Johnson, J. and B. Maxwell. 1996. Community sustainability through ecosystem management and planning. Montana Policy Review 6:22-31.

Johnson, J. D. and R. Rasker. 1995. The Role of Economic and Quality of Life Values in Rural Business Location. Journal of Rural Studies. Vol.11(4): 405-416.

Myer, W.B., and B.L. Turner II, eds. 1994. Changes in Land Use and Land Cover: A Global Perspective. Press Syndicate of the University of Cambridge, Melbourne, Australia.

Rasker, R. and G.J. Roush. 1995. Economics and Environmental Quality: New Roles for Environmentalists. In Brick P. and G. Cawley (eds.) A Wolf in the Garden: The Land Rights Movement and the Renewal of the American Environmental Movement. Rowman and Littlefield, N.Y.

Rasker, R. N. Tirrell, D. Kloepfer. 1992. The Wealth of Nature: New Economic Realities in the Yellowstone Region. The Wilderness Society, Washington, D.C.

Rasker, R.N., and D. Glick. 1994. The footloose entrepreneurs: Pioneers in the new west. Illahee 10(1):34-43.

Sader, S.A. 1995. Spatial characteristics of forest clearing and vegetation regrowth as detected by Landsat Thematic Mapper imagery. Photogr. Eng. and Remote Sensing 61(9):1145-1151.

Sader, S.A., T. Sever, J.C. Smoot, M. Richards. 1994. Forest change estimates for the northern Peten region of Guatemala: 1986-1990. Human Ecology 22(3):317-332.

Dr. Hansen's Landscape Biodiversity Lab

NASA's Mission to Planet Earth

EPA's Multiple Resolution Land Cover Mapping

Global Land Cover Mapping