2000 Progress Report: Development And Testing Of A Multi-Resource Landscape-Scale Ecological Indicator: Forest Fragmentation, Structure, and Distribution Relative to TopographyEPA Grant Number: R826598
Title: Development And Testing Of A Multi-Resource Landscape-Scale Ecological Indicator: Forest Fragmentation, Structure, and Distribution Relative to Topography
Investigators: Seagle, Steven W. , Townsend, Philip A.
Institution: University of Maryland Center for Environmental Science
EPA Project Officer: Packard, Benjamin H
Project Period: October 1, 1998 through September 30, 2003
Project Period Covered by this Report: October 1, 2000 through September 30, 2001
Project Amount: $683,374
RFA: Ecological Indicators (1998) RFA Text | Recipients Lists
Research Category: Ecological Indicators/Assessment/Restoration , Ecosystems
The objective of this research project is to understand how water quality and avian habitat quality vary across landscapes as a function of forest fragmentation and forest topographic position. Our primary hypothesis is that these two resources co-vary in a predictable manner, and that indicators, which simultaneously reflect both resources, can be developed from databases of topography (digital elevation models), land use, and remotely sensed forest structure. Field sites being employed are located within the Mid-Atlantic Highlands within the Appalachian Plateau and the Ridge-and-Valley physiographic provinces.
During this funding period, efforts focused on: (1) a second field season of forest-interior bird censuses on our field sites; (2) field and laboratory studies to quantify forest productivity for our field sites; (3) measurement of forest structure on our field sites; (4) preparation of synthetic aperture radar scenes for analysis to predict forest structure; (5) classification of land use for the Mid-Atlantic Highlands from Landsat images; and (6) quantification of forest invertebrate abundance.
Avian Surveys. The second and final season of forest-interior bird surveys were conducted on 22 field sites in western Maryland from May through mid-August of 2000. Each site was surveyed twice weekly in the early morning using "behavior mapping" to estimate diversity, density, and reproduction by species. This technique uses mapping of behaviors to document territories, pairing, nesting attempts, and fledging. For the year 2000 data collection, we introduced the use of pen-sensitive field computers to enter the data directly into digital form. Spatial data were downloaded daily to desktop computers in the lab, thus removing any errors associated with data transcription. Programs and protocols to accept the downloaded information, archive the data, and convert it into spreadsheet form identical to the 1999 database also were developed. An automated mapping system developed in 1999, imports the database and separates the observations by species and field site. Final determinations of territories, population sizes, and reproductive success for both years are now near completion for statistical analyses.
Forest Productivity. Forest productivity is being used as a correlate of avian productivity and as an interesting indicator to be predicted from topographic data. Productivity was estimated for each study site by measuring the age and height of canopy trees, and calculating Schnurr's site index value based on an equation for upland oaks. Five healthy canopy trees were selected at five spatially distributed plots within each site for a total of 25 trees measured per field site. Because site index relationships vary among tree species, we focused on canopy-dominant northern red oak (Quercus rubra) and supplemented these trees with other dominant canopy species as needed. Age was sampled using an increment borer and height was measured using a laser rangefinder. Diameter at breast height (DBH) was also recorded. Sample points were georeferenced with a global positioning system (GPS). A preliminary regression of site index values on various topographic indices generated from a 30 m resolution digital elevation model (DEM) indicates three topographic moisture indices useful for extrapolating forest productivity across landscapes: (1) the topographic convergence index; (2) relative slope position; and (3) a southwest/northeast transformation of slope aspect.
Characterization of Forest Structure. Field sampling of forest structure and composition was completed during the summer of 2000. A total of 180 60m x 60m plots have been sampled, with 110 of these plots located on the field sites where bird censuses were conducted and corresponded to those plots for which forest productivity quantified. The remaining 70 plots are distributed throughout the general study area, and are arrayed to serve three primary purposes: (1) to capture the range of environmental/vegetation gradients found throughout the region; (2) to facilitate spatial extrapolation of the analyses across the study area; and (3) to provide data for validation of our landscape-scale analyses. Notably, our sampling design for each plot consisted of five subplots, thus allowing computation of a standard deviation for each variable and quantification of variability within each plot. Frequency distributions of these structural variables indicate that our sampling captured the range of forest properties in the region. Forest structural properties are comparable between the Appalachian Plateau and Ridge-and-Valley.
Remote Sensing. One of our primary objectives is to use forest measurements in conjunction with remote sensing imagery to map the distribution of forest structure. We are using synthetic aperture radar (SAR) in conjunction with Landsat Thematic Mapper data to accomplish this, because Landsat (optical) imagery tends to be most sensitive to vegetation greenness; SAR is more sensitive to canopy structure. Because our study area is located in the steep terrain of the Mid-Atlantic Highlands, one of the critical factors that influence our ability to interpret the imagery is the effect of the terrain on surface reflectance (Landsat) and radar backscattering (SAR). For the Landsat analyses, we have developed a method to rapidly correct multiple images for topographic shading. This method greatly reduces the effects of shading on the analysis of Landsat imagery, and permits the comparison of vegetation on slopes with differing orientations, but otherwise similar vegetation. In addition, these terrain corrections improve the accuracy of the land cover classification being developed for analyzing the effects of forest topographic distribution and forest fragmentation on water quality. The correction of SAR data for terrain effects has proven to be more problematic. Currently, we are pursuing a post-transformation method that empirically corrects for differences in SAR illumination caused by topography. In anticipation of potential problems along these lines, we acquired SAR data imaged in both ascending and descending modes; meaning that our study sites have been imaged with the SAR sensor facing both shaded and unshaded directions. At a minimum, we can stratify our analysis by slope orientation to account for terrain effects.
Quantification of Forest Invertebrates. Forest invertebrates are an important source of food for neotropical migratory forest-interior birds. During the summer of 2000, eight field sites were chosen as representative of the range of topographic and moisture conditions found among our 22 field sites. Within each of these eight sites, forest invertebrates were sampled three times during the avian breeding season. Sampling focused on litter invertebrate biomass (food for ground foraging birds) and frass fall as an index of canopy arthropod larvae (food for foliage gleaning birds). Within each site, sampling was conducted on those five plots for which forest productivity and forest structure were intensively measured, with five replicates per plot. This sampling scheme resulted in 200 litter and 200 frass samples for each of the three sampling dates, and allowed calculation of variability within and between sampling dates at the plot, site, and topographic position levels. Data on most invertebrate taxa were collected as both counts and dry weight biomass. Additional microhabitat variables were measured for each litter invertebrate sampling site. Although data analysis is still in progress, it is clear that significant spatial variation exists in invertebrate biomass. In addition, Collembola populations (a primary decomposer) appear to be notably lower than most values reported in the literature. Invertebrate biomass will be used as a predictor of avian reproductive success, and also will be examined for correlation with forest structure and topographic variation. Selected taxa that are readily recognizable and opportunely sampled will be examined as potential indicators of both forest productivity and avian productivity.
During the next year, we will initially focus on the analysis of field data and the completion of remotely sensed image (both SAR and Landsat) classification. Data analyses will concentrate on final quantification of avian reproductive success, invertebrate biomass estimates, forest productivity, and forest structure. In completing these analyses, we expect to address several significant statistical issues. Paramount among these are: reconciliation of spatial scale differences between avian reproductive measures and topographic measures, development of robust statistical characterizations of forest productivity, and defining the best geometric combination of forest variables to index 3-dimensional forest structural heterogeneity. Our SAR and Landsat satellite data analysis techniques are now established, and analyses should be accomplished readily. For example, we have recently extracted the SAR/Landsat data from the satellite images for each of our field plots and analyses of relationships with forest structural data is underway. In conclusion, our data and satellite image analyses will provide several unique data sets to address mechanisms behind landscape-scale patterns in avian reproduction, forest structure, and invertebrate biomass. Based on these mechanisms, our ultimate objective is the development of landscape-scale indicators appropriate for both water quality and avian reproductive success. Indicator development will hinge on statistically valid relationships of: (1) avian reproductive success to topographic variables and forest structure; and (2) water quality to forest topographic distribution and fragmentation. Water quality data for establishing this second relationship has been obtained from U.S. Environmental Protection Agency Environmental Monitoring and Assessment Program (EMAP), thus completing the data sets necessary for our analyses. With success in defining these relationships, we expect to produce spatially explicit indictors at landscape to regional scales, and to define the spatial resolution at which these indicators are valid for both avian reproduction and water quality.