Robert V. O’Neill

CONSIDERATIONS FOR THE DEVELOPMENT OF A TERRESTRIAL INDEX OF ECOLOGICAL INTEGRITY

Abstract Ecological systems are composed of complex biological and physical components that are difficult to understand and to measure. However, effective management actions and policy decisions require information on the status, condition, and trends of ecosystems. Multiple levels of information are needed to make effective decisions and the ideal indicators for measuring ecosystem integrity will incorporate information from multiple dimensions of the ecosystem. A terrestrial index of ecological integrity would be a useful tool for ecosystem managers and decision makers. The ideal requirements of the terrestrial index of ecosystem integrity (TIEI) are that it be comprehensive and multi-scale, grounded in natural history, relevant and helpful, able to integrate concerns from aquatic and terrestrial ecology, and that it be flexible and measurable. The objective of this research is to investigate if an index, or indices, could be developed that would summarize the condition of ecosystems so that changes can be tracked over time and this information utilized as a tool to support environmental decision making.

The value of ecosystem services: putting the issues in perspective

a Department of Zoology, Center for En6ironmental Science, Uni6ersity of Maryland, Box 38, Solomons, MD 20688, USA b Institute for Ecological Economics, Uni6ersity of Maryland, PO Box 38, Solomons, MD 20688, USA c Department of Economics (emeritus), Uni6ersity of Wyoming, Laramie, WY, 82070, USA d Center for En6ironment and Climate Studies, Wageningen Agricultural Uni6ersity, PO Box 9101, 6700 HB Wageningen, The Netherlands e Graduate School of Public and International Affairs, Uni6ersity of Pittsburgh, Pittsburgh, PA 15260, USA f Institute for Ecological Economics, Uni6ersity of Maryland, PO Box 38, Solomons, MD 20688, USA g Department of Geography, Uni6ersity of Illinois, Urbana, IL 61801, USA h NCSA, Uni6ersity of Illinois, Urbana, IL 61801, USA i Institute of Ecosystem Studies, Millbrook, NY, USA j Department of Ecology, E6olution and Beha6ior, Uni6ersity of Minnesota, St. Paul, MN 55108, USA k En6ironmental Sciences Di6ision, Oak Ridge National Laboratory, Oak Ridge, TN 37831, USA l Department of Ecology, Faculty of Agronomy, Uni6ersity of Buenos Aires, A6. San Martin 4453, 1417 Buenos Aires, Argentina m Jet Propulsion Laboratory, Pasadena, CA 91109, USA n Department of Geography, National Center for Geographic Information and Analysis, Uni6ersity of California at Santa Barbara, Santa Barbara, CA 93106, USA o Ecological Economics Research and Applications, PO Box 1589, Solomons, MD 20688, USA

A hierarchical neutral model for landscape analysis

Empirical studies have revealed scaled structure on a variety of landscapes. Understanding processes that produce these structures requires neutral models with hierarchical structure. The present study presents a method for generating random maps possessing a variety of hierarchical structures. The properties of these scaled landscapes are analyzed and compared to patterns on totally random, unstructured landscapes. Hierar-chical structure permits percolation (i.e., continous habitat spanning the landscape) under a greater variety of conditions than found on totally random landscapes. Habitat clusters on structured maps tend to have smaller perimeters. The clusters tend to be less clumped on sparsely occupied landscapes and more clumped in densely occupied conditions. Hierarchical structure changes the expected spatial properties of the landscape, indicating a strong need for this new generation of neutral models.

Heterogeneity and Spatial Hierarchies

To apply the traditional scientific method, ecologists ordinarily focus on the mean or central tendency of a data set. For example, a typical hypothesis test would involve demonstrating that the mean is significantly different from a control measurement. However, ecological systems are heterogeneous, and much information may be lost if the variance of a data set is ignored. This chapter shows that a specific prediction of hierarchy theory can be tested by examining how variance changes as measurements are taken across a range of scales.

Ecological Implications of Landscape Fragmentation

The effects of scale-dependent changes in habitat patterns were investigated by simulating two consumer populations on random and hierarchically random landscapes. The first consumer depleted resources as it randomly walked across the landscape, while the second consumer spread outwards to suitable sites from an initial point. Simple random landscapes were generated by randomly selecting the habitat suitability of each site, while hierarchically random maps were generated by a nested series of rules which changed the probabilities with spatial scale. These two map types were compared to digitized land-use data. Three effects were observed: (1) the reduction of available habitat always reduced population abundance and spread, (2) small changes in landscape pattern produced sudden changes in abundance which could be predicted from the interaction between landscape characteristics and population specific life history parameters, and (3) the hierarchically structured landscapes affected abundances most when habitat fragmentation coincided with the scales at which the consumer populations utilized spatial resources.

Recovery in complex ecosystems

Current ecosystem theory has a deceptively simple representation of recovery. In actual practice,recovery is affected by the frequency and extent of disturbances and by the spatial heterogeneity of the ecological system. Environmental changes may pass through thresholds causing recovery to a different plant and animal community. The sheer complexity of the system combined with unanticipated synergistic effects can make recovery trajectories difficult or impossible to predict. New theoretical constructs,based on stochastic nonlinear theory, will be needed to guide research and applications.

A Percolation Model of Ecological Flows

The boundary zone between adjacent communities has long been recognized as a functionally important component of ecosystems (Odum 1959). The diversity and abundance of species (Noss 1983), the flow and accumulation of material and energy (Ranney et al. 1981), and the propagation of disturbances (Picket and White 1985, Turner et al. 1989) may all be affected by landscape boundaries. However, the spatial arrangement of different habitats and their boundaries has received little direct study (Wiens et al. 1985, Krummel et al. 1987). It is not surprising, therefore, that Hansen et al. (1988) have noted, “the extent to which landscape boundaries influence ecological flows is not well known and recent treatments of the topic remain speculative.”

Simulation of the Scale-Dependent Effects of Landscape Boundaries on Species Persistence and Dispersal

The relationships between life history characteristics and broad-scale patterns of species abundance were investigated with a model that simulates the dispersal of populations through heterogeneous landscapes. Movement was simulated on randomly generated landscapes and on forested landscapes digitized from aerial photographs. Simulation results indicated that population abundances will change suddenly near the critical threshold in habitat connectivity as predicted from percolation theory. The existence of critical thresholds is important for many management and conservation issues, but these thresholds suggest that data are required at spatial scales that are specific to the dispersal characteristics of the simulated population.

Epidemiology theory and disturbance spread on landscapes

AbtractEpidemiology models, modified to include landscape pattern, are used to examine the relative importance of landscape geometry and disturbance dynamics in determining the spatial extent of a disturbance, such as a fire. The models indicate that, except for very small values for the probability of spread, a disturbance tends to propagate to all susceptible sites that can be reached. Therefore, spatial pattern, rather than disturbance dynamic, will ordinarily determine the total extent of a single disturbance event. The models also indicate that a single disturbance will seldom become endemic,i.e., always present on the landscape. However, increasing disturbance frequency can lead to a landscape in which the proportion of susceptible, disturbed, and recovering sites are relatively constant.

Exploring Aggregation in Space and Time

Population and ecosystem processes are heterogeneous in both time and space, and every ecological study requires some level of aggregation or abstraction. Aggregating organism or environmental dynamics is challenging because the processes occur at a variety of spatial and temporal scales, and the scale-dependent effects of aggregation are not well understood. We used a spatially explicit individual-based simulation model of winter foraging and survival of free-ranging ungulates in northern Yellowstone National Park to explore effects of aggregation in space, in time, and of individual animals on model predictions. Aggregation in space was examined by (1) varying the heterogeneity represented in forage abundance across the landscape and (2) eliminating spatial heterogeneity in the accumulation of snow. Results suggest that any aggregation that averages the broad-scale patterns of forage biomass availability underestimates ungulate survival. Aggregation in time was examined by varying the temporal grain used to simulate snow accumulation through the winter. Ungulate survival was not sensitive to this temporal grain, probably because the response remained linear within the range explored. Aggregation of individuals was done by varying the number of individuals contained within ungulate groups assumed to contain identical individuals. Aggregating across individuals was reasonable for small group sizes but led to substantial underestimates of survival for large group sizes. The effect of aggregation on an ecosystem or population parameter is a function of the question asked and a specified spatial and temporal scale. Even successful aggregation of processes will be reliable only if dynamic thresholds are not crossed, if keystone species are not eliminated, and if feedback loops remain intact.