Characterization of the Ecological Integrity of Commercially Grazed Rangelands Using Remote Sensing-based Ecological Indicators

Updated 02/27/2009

Robert A. Washington-Allen^1,2 , Neil E. West¹, R. Douglas Ramsey³, and Carolyn T. Hunsaker⁴

Summary of Research:

We are using 27 years of wet and dry season Landsat satellite imagery from (1972 to 1998); a Geographic Information System (GIS) digital database of site biological, physical, and administrative characteristics and analysis tools; historical and current ranch management records; and a multiple-time by nested multiple-scale experimental design to establish causal links between possible threshold response of ecological indicators diagnostic of land degradation and human management interventions in order to assess the ecological integrity of ecosystems within a semi-arid landscape subject to commercial livestock grazing (Fig. 1).

Range scientists define rangeland degradation as: (1) a change in plant species composition, e.g., an increase in low quality forage species compared with high quality forage; (2) a decrease in plant productivity; (3) a reduction in soil quality; (4) accelerated soil erosion (Pickup 1989, Behnke and Scoones 1993); and a change in landscape composition and pattern that affect ecosystem functions. We view ecological terminology such as landscape and ecosystem as spatio-temporal scale-independent criteria (sensu Allen and Hoeskstra 1992). Ecological integrity is defined as an equilibrial or non-equilibrial ecosystem or landscape which possesses a full set of natural components and processes in "good working order" (West et al. 1994). Composition and pattern refer to the spatial organization of ecosystem characteristics when viewed as patches within a mosaic, e.g., a shrub vegetation patch within a bare soil and grassland mosaic (Noy-Meir 1973, Ludwig and Tongway 1995). Measurement indicators derived from these five assessment endpoints (an explicit statement of the actual environmental value that is to be protected, Suter 1993) will provide a diagnostic basis for an estimate of the degree of land degradation. We are using indices of vegetation cover and soil integrity, the Soil Adjusted Vegetation Index (SAVI, Huete 1988) and the Soil Stability Index (SSI, Pickup and Nelson 1984), and metrics of landscape structure (Turner 1989) on categorical maps of plant lifeforms and bare soil. Figure 2 is an overview of how the analyses of the measurement idicators are related to their derivation from the same landscape and its changes in space and time (Figure 2A). The analyses concentrate on possible thresholds associated with state changes of ecological sites (Figure 2B) and landscape pattern (Figure 2C), changes at concentration areas, e.g., waterpoints, and the use of change vector analyses to measure ecological resilience (Figure 2D). Discrete or rapid change in an indicator is indicative of a threshold. A transition threshold is defined as the boundary in space and time between two seral states, e.g., the observed changes of grassland to predominantly woody vegetation (Archer 1989, Figure 3). Land managers have a different perception of thresholds that is primarily informed by ecological and economic considerations. From their perspective, a transition threshold is the point where the initial shift across the boundary is not reversible on a practical time scale without substantial intervention of cultural inputs, e.g., heavy machinery, or fire (Archer 1989, Friedel 1991, Figure 3 (the right-hand Y-axis)).

The study site selected for this research is the 88,000 ha Deseret Land and Livestock Ranch (McMurrin 1989) located in the northeast corner of the Utah panhandle. The first stage of our project has been GIS and remote sensing data acquisition and development (Table 1). Extension activities with Deseret's land managers has yielded eighteen years of livestock (1980 to 1997) and sixteen years of wild ungulate (1982 to 1997) management data (Figures 4 and 5). Twenty-six years of wet and dry season imagery (49 scenes) and two years of annual imagery (1985 and 1986, 9 scenes/yr) are being image-to-image rectified. Image normalization of the data set is occurring via transformation to exo-atmospheric reflectance values and use of an empirical radiometric normalization technique to relatively atmospherically-correct imagery to a reference image. Landsat MSS data collection ceased after 1992, thus wet and dry season Landsat Thematic Mapper (TM) imagery from 1986 to 1997 (Table 1.) were acquired and multiple regression techniques are being used to allow TM sensor response to simulate MSS sensors. The image processing procedures used in this study are labor intensive, but we are automating this process in collaboration with scientists from a DoD Strategic Environmental Research and Development Program (SERDP) conservation project that is concerned with developing remote sensing-based environmental monitoring protocols for military bases.

Results

Seasonal images from 1972 to 1992 were converted to the soil-adjusted vegetation index (SAVI) images and analyzed (Fig. 6). SAVIseasonal state space mean/variance plots for Deseret at the landscape scale reveal that vegetation cover dynamics clearly discriminate into two domains of high mean/low variance in the 1970s and low mean/high-low variance in the 1980's and 90s. The two domains corresponded to wet and drought periods indicated by the Palmer Drought Severity Index (PDSI). The mean/variance plots temporal trajectories were aperiodic with somewhat periodic orbit pattern which identify the two domains as "strange attractors". This pattern is reminiscent of Lorenz's Butterfly and thus indicative of a chaotic system. In preliminary studies, seasonal images from 1972 to 1992 were converted to SAVI images and there trend analyzed Three hypotheses were tested at the landscape and administrative scales: which were 1) at the landscape scale, the El Niño wet year of 1983-84 would be manifest when compared to mean conditions; 2) The period from 1972 to 1983 would have lower vegetational cover than the period 1984 to 1992 because of better ranch management; and at the administrative scale, 3) vegetation cover within Bureau of Land Management (BLM) islands would not differ from that of the privately-owned paddocks they were within because Deseret Land and Livestock does not separately manage on the basis of land ownership. We found that 1) the El Niño wet year of 1983-84 allowed most of the landscape to recover from an apparent drought in 1982, except for riparian areas which had lower than mean vegetational cover 2) Contrary to our hypothesis, the period from 1972 to 1983 had higher vegetation cover than the period 1984 to 1992; and 3) as expected, vegetation cover between BLM islands and the paddocks which contained them did not differ with correlation coefficients ranging from 0.86 to 0.99 (Figure 7). The overall SAVI trend suggests a decline in vegetational cover and examination of the trend at the scale of individual vegetation types revealed a similar response compared to the landscape scale. This response suggests a climatic constraint. However, comparison at the landscape-scale of the lagged water year PDSI and the SAVI had a low correlation of 0.17. There are anomalous years: 1978 (high SAVI and dry year), 1981 (high SAVI and dry year) , and 1982 (low SAVI and wet year) which are not consistent with the wet and dry precipitation periods. However, the relationship was fairly consistent for the dry period 1987 to 1992. A possible explanation for the behavior in 1982 may be related to the livestock and wildlife grazing record and other land management practices which have occurred, e.g., prescribed burning. Later analyses will consider these factors.

Future Activities:

We will complete the database, including production of soil attribute layers and digital elevation models, acquisition and scanning of aerial photos for accuracy assessment, and then we will proceed to concentrate on analyses. We will analyze the temporal and spatial behaviour of remote sensing-based ecological indicators derived from various definitions of rangeland degradation. The assessment of these indicators will occur at multiple scales including: landscape, watershed, administrative (i.e., public versus private land), individual paddock, ecological site, and piosphere (waterpoints). Continued extension activities will occur with Deseret's natural resource managers, including transfer of Deseret's GIS database to CD-ROM for use by Deseret.

Literature Cited

Allen, T. F. H. and T. W. Hoekstra. 1992. Toward a Unified Ecology, Columbia University Press, New York, 384 p.

Archer, S. 1989. Have southern Texas savannas been converted to woodlands in recent history? American Naturalist 134:545-561.

Behnke, R.H. and I. Scoones. 1993. Rethinking Range Ecology: Implications for rangeland management in Africa. pp. 1-30, In Behnke, R.H., I. Scoones and, C. Kerven (Eds.) Range Ecology at Disequilibrium. Overseas Development Institute, International Institute for Environment and Development, and Commonwealth Secretariat, London.

Friedel, M.H. 1991. Range condition assessment and the concept of thresholds: A Viewpoint. Journal of Range Management 44:422-426.

Huete, A.R. 1988. A soil-adjusted vegetation index (SAVI). Remote Sensing of Environment 25:295-309.

Ludwig, J.A. and D.J. Tongway. 1995. Spatial organization of landscapes and its function in semi-arid woodlands, Australia. Landscape Ecology 10:51-63.

McMurrin, J. 1989. The Deseret Live Stock Company : the first fifty years, 1890-1940. MS Thesis, University of Utah, Salt Lake City, Utah.

Noy-Meir, I. 1973. Desert ecosystems: environment and producers. Annual Review of Ecology and Systematics 4:25-51.

Pickup, G. 1989. New land degradation survey techniques for arid Australia: problems and prospects. Australian Rangeland Journal 11:74-82.

Pickup, G. and D.J. Nelson. 1984. Use of Landsat radiance parameters to distinguish soil erosion, stability and deposition in arid central Australia. Remote Sensing of Environment 16:195-209.

Schlesinger, W. H., J.F. Reynolds, G. L. Cunningham, L.F. Huenneke, W. M. Jarrell, R.A. Virginia, and W.G. Whitford. 1990. Biological feedbacks in Global Desertification. Science 247:1043-1048

Suter II, G.W. 1993. Ecological Risk Assessment. Lewis Publishers, Boca Raton. 538 p.

TGUCT (Task Group on Unity in Concepts and Terminology). 1995. New Concepts for assessment of rangeland condition. Journal of Range Management 48:271-282.

Turner, M.G. 1989. Landscape Ecology: The effect of pattern on process. Annual Review of Ecology and Systematics 20:171-197.

West, N.E., K. McDaniel, E.L. Smith, P.T. Tueller, and S. Leonard. 1994. Monitoring and interpreting ecological integrity on arid and semi-arid lands of the western United States. Western Regional Research Coordinating Committee-40. New Mexico Range Improvement Task Force, Report No.37, Las Cruces, New Mexico.

Figure 1. An overview of the proposed GIS/Remote Sensing-based framework for assessing the degree of land degradation on a landscape managed for commercial livestock grazing.

Figure 2. Conceptual basis for the development of a multi-temporal Remote Sensing/GIS-based assessment of rangeland integrity.

Figure 3. State Changes and Transition Thresholds

Figure 4. The Stocking Rate (Animal Unit (AU) per hectare for a year) on Deseret Land & Livestock Ranch from 1980 to 1997.

Figure 5. Population of elk (A), pronghorn (B), and deer (C) on Deseret Land & Livestock Ranch from 1982 to 1997. The acronym AD means animal days, but in this case is equal to animal numbers.

Figure 6. Mean /Variance plots of the Wet and Dry Season Soil Adjusted Vegetation Index (SAVI). SAVI indicates the amount of vegetation cover on the ranch where high variance and low mean years are indicative of drought and increased susceptibility to accelerated soil erosion.

Figure 7. Comparison from 1975 to 1992 of the soil adjusted vegetation index (SAVI), an index of vegetation cover, between Bureau of Land Management land islands and the private land which contains them.

< Back to Agenda