Volume No. 11, 2002
Aquatic GAP
We are developing Aquatic GAP applications for two centers of aquatic biodiversity, the Alabama-Coosa-Tallapoosa (ACT) and Apalachicola-Chattahoochee-Flint (ACF) river basins. The ACT and ACF basins span broad ranges of physiographic settings and harbor exceptionally high levels of species richness and endemism, providing ideal opportunities for testing and refining approaches to predict species occurrences and community attributes in relation to physical variables. The ACT basin (58,708 km2) originates in the Blue Ridge province of the Southern Appalachian Mountains in Georgia and Tennessee, drains extensive portions of the Valley and Ridge and Piedmont provinces in west Georgia and east Alabama, and of the Coastal Plan in lower Alabama and southwestern Georgia (note: most of the lower Flint River is in the Coastal plain). Physiographic and climatic diversity, combined with a geologic history of isolation punctuated by interbasin dispersal and protection from Pleistocene glaciation, have fostered development in the ACT of some of the highest levels of aquatic faunal diversity and endemism recorded in temperate freshwaters. At least 184 native freshwater fishes occur in the ACT (Warren et al. 2000). The Coosa River system alone contains at least 15 endemic fishes as well as remnants of an exceptionally diverse molluscan fauna (Bogan et al. 1995, Burkhead et al. 1997, Neves et al. 1997). The Chattahoochee and Flint Rivers (our focus in the ACF) together drain 44,607 km2 of Georgia and east Alabama, including the Blue Ridge, Piedmont and Coastal Plain provinces. Fish and molluscan faunas are distinct from those in both Atlantic Slope drainages to the east and the ACT to the west and include at least 97 native fishes and 32 native mussel species (Couch et al. 1996, Brim Box and Williams 2000). One fourth of the native ACF mussel fauna is endemic to the basin (Brim Box and Williams 2000), along with at least six fish species (Warren et al. 2000).
The need for an Aquatic GAP application in these river systems is no less than urgent. At least 114 aquatic species in the ACT, Chattahoochee, and Flint rivers are considered imperiled as a result of habitat degradation and loss (Ziewitz et al. 1995). Federally listed animals include 6 mussels and 1 fish native to the Chattahoochee and Flint systems, 14 ACT mussel and snail species, and 10 ACT fishes. Levels of species imperilment likely underestimate the actual extent of loss for unique stream types with high water quality and faunal integrity. Conversion from forest to agriculture, urban growth, and river impoundment for hydropower and navigation have altered stream and river habitat throughout much of the basins. For example, dams and reservoirs impound approximately 44% of the ACT mainstem rivers and 64% of the Chattahoochee mainstem. Presently, parts of the region are experiencing some of the highest population growth rates in the nation, resulting in urban sprawl, impervious surface proliferation, and increasing pressures on streams for water supply. At least 16 water supply reservoirs are planned for construction on streams in the Coosa, Tallapoosa, Chattahoochee, and Flint systems in Georgia. Georgia, Florida, and Alabama are locked in an interstate controversy over water use and water allocation in these systems. Georgia has recently enacted legislation to facilitate removing some irrigated lands from production during extreme droughts in order to protect stream flows in the Flint River system. The intense and growing competition for water in these systems―to support population growth, expanding agriculture, industry, and hydropower and to provide for healthy stream communities―reflects the urgency with which scientifically sound tools are needed to facilitate landscape-level planning and biodiversity conservation.
Our goal for this project is to develop methods and define
appropriate scales for application of Aquatic GAP to all states by
integrating terrestrial GAP with Aquatic GAP. Our specific
objectives are to:
This regional project will be conducted in the subbasins of the ACT/ACF basins, located primarily in eastern Alabama and western Georgia. As described above, these basins are diverse in both habitat (on multiple scales) and aquatic biodiversity. We will specifically conduct this research in the Upper Coosa River (above Weiss Reservoir), the Tallapoosa River, and the Flint River basins.
Full integration with the terrestrial GAP projects will be maintained through interaction with project steering committees and oversight committees for each of the activities.
We will build data layers at various scales for use in developing faunal distribution models. In addition to using existing layers produced from AL- and GA-GAP projects, we will build on existing hydrography layers and create other layers that will be used in predictive models of aquatic species distributions.
Hydrographic Data - Within the study basins, subwatersheds (U.S.Geological Survey 6th code, 12-digit, hydrologic units; mean size 7,800 ha) comprise one basic unit for our landscape-level analysis. In addition, subwatershed boundaries have been delineated by hand based on digital raster graph (DRG) images or digital elevation models (DEM) of 1:24,000 and 1:100,000 USGS topographic quadrangles. In some cases, subbasins have been delineated by agencies (e.g., NRCS, AL). Definition of subwatershed boundaries at fine scales has been determined based on position in the drainage network, especially relative to features such as dams. Digital hydrography will be obtained either from USGS or from other agencies (e.g., 1:24,000 Georgia DOT linear hydrography files). Again, to determine appropriate scale of assessment, both 1:24,000 and 1:100,000 data will be used. Nested within each subwatershed, perennial streams will be divided into segments (sensu Frissell et al. 1986).
Landscape Data - We are in the process of determining landscape variables for each 12-digit HUC, delineated subwatershed, and stream segment. Various data layers have been obtained from the terrestrial components of GAP and include land use/land cover and stewardship coverages. Because the Alabama GAP LU/LC is not complete, we are using 1992 MRLC data as a basis for most models that include these layers. We are calculating landscape indices relative to these coverages, using either traditional methods or software such as FRAGSTATS (Cunha 2000). In addition, various features of terrestrial and aquatic landscapes have been related to distribution of aquatic fauna. Therefore, we are assembling other data layers from various sources. As examples, these include road density, dam density, point source discharges, and terrestrial nutrient sources located within 100 m of a stream (e.g., poultry farms). Landscape features identified on the stream segment scale include position in the drainage network (e.g., link magnitude and stream order), mean elevation, valley confinement, gradient, and flow regime (e.g., regulated versus unregulated).
Faunal Data - Faunal data will include fishes, aquatic reptiles and amphibians, and key groups of aquatic invertebrates depending on data coverage and availability. We have identified data regarding faunal distribution through a review of gray and published literature and through direct contact with biologists working throughout the region. We have made a particular effort to locate data that provide a broad geographic representation within the basins. Data from faunal collections in basins have been obtained from numerous sources. Observations were included in the database when they met several criteria: (1) sampling methods were documented either through written work or direct correspondence with the principal source, (2) the sampling site was spatially located either with direct geographic coordinates or by designation on a typical topographic quad, (3) the sampling date was recorded, and (4) aquatic species were principally targeted and recorded during sampling. We are matching historical data with contemporary records. Where data gaps exist (e.g., sites with historical but not current data), we will conduct surveys.
Because data quality can significantly affect analyses, we have examined various aspects of data quality. Rule sets have been developed for including data collected for previous studies and for weighting observations for analyses. For example, incorporating misleading evidence (i.e., false negatives) into distribution models can bias the models. Therefore, zero catch data will be weighted by an estimate of the probability of detecting a species or groups of species (e.g., Bayley and Peterson 2001).
Development and Testing of Predictive Models - We will compare and contrast several approaches for evaluating species status distributions and identifying unique or important areas, and estimate the relative accuracy of these approaches via 5-fold cross-validation (Breiman and Spector 1992). Responses of fauna to landscape variables will be assessed via hierarchical (aka multilevel) models. Fully hierarchical Bayes models and conditional models can use landscape data and segment-level characteristics simultaneously. Thus, the landscape data are used to provide "context" for the observed smaller scale (stream reach) phenomenon (e.g., aquatic community integrity/species status). For example, Dunham and Rieman (1999) found that some trout species did not occur in small landscape catchments although there was appropriate habitat. Explicitly incorporating this context has also been shown to increase the accuracy of empirical fish species detection estimates. Hierarchical models will allow for assessment of appropriate scales for application of Aquatic GAP.
To identify unique or important areas in terms of fish species diversity and the integrity and composition of fish communities, we will fit models relating landscape and stream segment characteristics (henceforth, predictors) to three basic community responses. First, we will estimate the degree of change in native community composition using the historical and current number of native taxa as dichotomous dependent variables and relate these to the predictors via logit (Agresti 1990) and hierarchical logit models (Bryk and Raudenbush 1992) in which segments are nested within subwatersheds. Second, we will estimate an index of biotic integrity (IBI) of each site and relate these to the predictors via simple linear and hierarchical linear models. The third approach will be to relate the distribution or status (sensu Thurow et al. 1997) of sensitive target taxa to the predictors via logit or similar categorical modeling techniques (e.g., Haas et al. in press). Comparisons will be made among the approaches by identifying critical areas based on the estimates from each technique (i.e., low degree of community change, high IBI score, strong populations of target taxa) and examining the degree concordance among predictions.
Development of a Decision Support System - Resource agencies require decision support tools for managing aquatic resources. The landscape-scale models developed in our project will provide a basis for developing tools for making decisions regarding future land management and sampling/monitoring decisions. For instance, we have recently (2002) initiated a concurrent project, funded by the Georgia Department of Natural Resources (GADNR), to develop quantitative decision models to assist the assessment and planning of river regulation and water resource development activities in the Flint River Basin. These spatially explicit tools will combine predictive models of current species distribution (from GAP) with flow, habitat, and aquatic community response models. To explicitly incorporate uncertainty, relationships between current species distributions, streamflow, habitat, and the aquatic community response will be modeled as conditional dependencies and recast as probabilistic networks (see Peterson and Evans 2003 for an example). The probabilistic network format then will allow us to integrate the model into user-friendly software that will be used by the GADNR to evaluate various flow scenarios. We also plan to develop similar Web-based decision support systems for other resource agencies managing aquatic resources in the ACT/ACF basins.
Agresti, A. 1990. Categorical data analysis. Wiley and Sons, New York.
Bayley, P.B., and J.T. Peterson. 2001. Species presence for zero observations: An approach and an application to estimate probability of occurrence of fish species and species richness. Transactions of the American Fisheries Society 130:620-633.
Bogan, A.E., J.M. Pierson, and P. Hartfield. 1995. Decline in the freshwater gastropod fauna in the Mobile Bay basin. Pages 249-252 in E.T. LaRoe, G.S. Farris, C.E. Puckett, P.D. Doran and M.J. Mac, editors. Our living resources: A report to the nation on the distribution, abundance, and health of U.S. plants, animals, and ecosystems. U.S. Department of the Interior, National Biological Service, Washington D.C. 530 pp.
Brim Box, J., and J.D. Williams. 2000. Unionis mollusks of the Apalachicola Basin in Alabama, Florida, and Georgia. Bulletin of the Alabama Museum of Natural History 21:1-143.
Breiman, L., and P. Spector. 1992. Submodel selection and evaluation in regression: The X-random case. International Statistical Review 60:291-319.
Bryk, A. S., and S.W. Raudenbush. 1992. Hierarchical linear models: Applications and data analysis methods. Sage, Newbury Park, California.
Burkhead, N.M., S.J. Walsh, B.J. Freeman, and J.D. Williams. 1997. Status and restoration of the Etowah River, an imperiled southern Appalachian ecosystem. Pages 375-444 in G.W. Benz and D.E. Collins, editors. Aquatic fauna in peril: The southeastern perspective. Special Publication 1, Southeast Aquatic Research Institute, Lenz Design and Communications, Decatur, Georgia. 554 pp.
Cunha, A. 2000. Influence of landscape patterns on spatial dynamics of larval fish in two southeastern rivers. Ph.D. dissertation. Auburn University, Alabama.
Couch, C.A., E.H. Hopkins, and P.S. Hardy. 1996. Influences of environmental settings on aquatic ecosystems in the Apalachicola-Chattahoochee-Flint River Basin. U.S. Geological Survey, Water Resources Investigations Report 95-4278.
Dunham, J.B., and B.E. Rieman. 1999. Metapopulation structure of bull trout: Influences of habitat size, isolation, and human disturbance. Ecological Applications 9:642-655.
Frissell, C.A., W.J. Liss, C.E. Warren, and M.D. Hurley. 1986. A hierarchical framework for stream habitat classification: Viewing streams in a watershed context. Environmental Management 10:199-214.
Haas, T.C., J.T. Peterson, and D.C. Lee. In press. An evaluation of parametric and nonparametric models of fish population response. Ecological Modelling.
Neves, R.J., A.E. Bogan, J.D. Williams, S.A. Ahlstedt, and P.W. Hartfield. 1997. Status of aquatic mollusks in the southeastern United States: A downward spiral of diversity. Pages 43-85 in G.W. Benz and D.E. Collins, editors. Aquatic fauna in peril: The southeastern perspective. Special Publication 1, Southeast Aquatic Research Institute, Lenz Design and Communications, Decatur, Georgia. 554 pp.
Peterson, J.T., and J.W. Evans. 2003. Decision analysis for sport fisheries management. Fisheries 28(1):10-20.
Thurow, R.F., D.C. Lee, and B.E. Rieman. 1997. Distribution and status of seven native salmonids in the interior Columbia River basin and portions of the Klamath River and Great Basins. North American Journal of Fisheries Management 17:1094-1110.
Warren, M.L., and 11 co-authors. 2000. Diversity, distribution, and conservation status of the native freshwater fishes of the southern United States. Fisheries 25(10):7-29.
Ziewitz, J.W., B.K. Lupek, and G.A. Carmody. 1995. Protected species inventory and identification in the Alabama-Coosa-Tallapoosa and Apalachicola-Chattahoochee-Flint River basins. Report by the U.S. Fish and Wildlife Service, Panama City, Florida, to the Technical Coordinating Group of the ACT-ACF Comprehensive Study.
Return to Table of Contents