A need for sound biodiversity conservation planning
Recent declines in biodiversity (Ehrlich and Wilson 1991) driven
by anthropogenic alteration, fragmentation and loss of habitat
(Wilson 1988) have prompted efforts aimed at development of laws
and policies for sound biodiversity conservation planning (Naveh
and Lieberman 1994, Bojorquez-Tapia et al. 1995). It is critical
that species rich areas be protected since such areas tend to
support a high number of rare species (Wright and Reeves 1992)
and proper management of these regions optimizes resources for
conservation (Scott et al. 1993). In addition, high diversity
ecosystems tend to promote higher productivity (Tilman et al.
1996), resource utilization (Tilman 1982, 1997), and resistance
to disturbance (Elton 1958).
Recent international and national conventions have been dedicated to developing methods to monitor and manage ecosystems for conservation of biodiversity (Kitching 1994, Smythe et al. 1996). It is clear that basic biological data are needed to make informed conservation decisions, however, comprehensive field sampling over large study areas can be time consuming and labor intensive (Tucker et al. 1997). Thus, geographic information systems (GIS) are increasingly used in biodiversity conservation planning (Norton and Williams 1992, Tucker et al. 1997). GIS models that could reliably predict biotic assemblages from landscape attributes would be particularly valuable in regions where biological surveys have not been completed or would be difficult to accomplish (e.g., Kirkpatrick and Brown 1994, Bojorquez-Tapia et al. 1995).
The gap analysis approach
The use of GIS models to predict biotic communities has primarily
involved terrestrial environments (e.g., Scott et al. 1993, The
Nature Conservancy 1994, Bojorquez-Tapia et al. 1995). Gap Analysis
was developed in 1988 in an effort to ensure that regions rich
in species diversity are conserved with the hope that this will
eliminate the need to list species as threatened or endangered
in the future (Scott et al. 1993). Gap Analysis quickly became
the largest effort ever mounted to map the biological resources
of the United States. The gap analysis approach (Scott et al.
1993) uses maps of vegetation and predicted animal distributions
to locate centers of species richness outside areas currently
managed for biodiversity protection. These are considered the
"gaps" of Gap Analysis. They are assumed to be critical
for the protection of biological resources. Far less attention
has been paid to the development of models for the prediction
of aquatic communities (Angermeier and Winston 1999) due primarily
to the difficulty in developing an analog for the Gap Analysis
vegetation map that is used to classify habitat types. Most of
the existing aquatic habitat classification efforts are hierarchical
(e.g., Frissell et al. 1986, Moyle and Ellison 1991, Rosgen 1994;
Angermeier and Schlosser 1995, Maxwell et al. 1995, Higgins et
al. 1998), have extensive data requirements (e.g., Ellison 1984,
Bazata 1991, Meador et al. 1993, Seelbach et al. 1997), vary in
methods among flowing and standing waters, or are based on only
a single landscape attribute (e.g., Lotspeich 1980, Aadland 1993).
Thus, although many elements of an aquatic GIS for conservation
planning are available, no complete method is currently in use.
Extension of landscape gap analysis methods to aquatic systems
A pilot gap analysis project for aquatic systems began in 1995
for western New York in an effort to define the methodology and
evaluate the feasibility of predicting biodiversity distribution
at the watershed scale. Similar to gap analysis performed in terrestrial
environments, gap analysis for aquatic systems seeks to identify
the locations of high biodiversity in watersheds, use remotely-sensed
data to map habitats, use habitats to infer aquatic biodiversity
distributions, and provide watershed-scale information useful
for targeting conservation measures. The pilot project was a low-level
effort (~0.75 staff yr/yr) for four years. It is now completed;
this report is the final task of the pilot effort. Since mid-1996,
a similar gap analysis demonstration project for aquatic systems
was implemented in the State of Missouri.
The original purposes of gap analysis have remained unchanged. However, the connected nature of aquatic habitats in a watershed and the mobility of aquatic species will likely complicate and diminish the value of static displays of species distributions used in terrestrial gap analysis. Therefore, both species-specific and community-based displays of biological diversity will be employed.
Emphasis has been placed on streams and rivers since these waterbodies harbor a large majority of the freshwater biodiversity in the United States and are the focus of water quality assessments by management agencies. Although some notable fish extinctions occurred in lakes prior to the mid-1900s, a large majority of endangered species are stream inhabitants. Furthermore, the quality of most lakes is dependent on their watersheds via inputs from tributary streams and are typically managed and studied as individual habitat units often with extensive data by waterbody. In addition, some wetlands harbor unique faunas, but their importance is primarily associated with hydrologic and water quality functions in a watershed context.
Impacts to a waterbody are cumulative due to the mobile nature of water in streams, thus methods were developed to reflect the effect of degradation over a connected framework of stream segments. In addition, aquatic biodiversity conservation will need to focus on land management, not land acquisitions, since aquatic biodiversity will generally be highest in large flowing waters making land acquisitions impractical.
Lastly, biodiversity prediction must be performed at a scale relevant for management and planning and appropriate for the size of the project (Seelbach et al. 1997). Studies performed at a coarse, broad scale may filter out spatial variation occurring at finer scales. On the other hand, fine-scaled studies are more time and labor intensive, and possibly too detailed to accurately perform predictions helpful to conservation planners. This issue was addressed by performing predictions at both the finer 1:24,000-scale and the coarser 1:100,000-scale for comparison of appropriateness of the models. Hand delineation of parameters from maps requires a great deal of time and labor and thus, a shift was performed from manual attribution used earlier in aquatic gap analysis to automated attribution in an attempt to decrease the input of time and labor and increase the flexibility and accuracy of the overall models.
Project purpose
Previous final reports for aquatic gap analysis described the
method of habitat characterization involving static, manually
intensive classifications from topographic and Mylar land use
overlay maps. At that time only the Allegheny River watershed
was used as a study area and no information had been collected
for use in accuracy analysis. This report will present the methodology
and accuracy analysis of two non-hierarchical GIS model approaches
(automated and calibrated) to predicting diversity in three study
areas (Allegheny River, Fall Creek and French Creek watersheds)
from multiple landscape attributes using commonly available data.
Data collection for accuracy analysis will be described and the
results of the analysis enumerated and discussed. The results
of predictions from these models should be of interest to resource
planners for management and restoration of aquatic habitat and
conservation of biodiversity.
The goal of our project was to demonstrate the feasibility and utility of the gap analysis methodology for predicting biodiversity distribution at the watershed scale. The specific objectives of the project were to:
The expected gains from the project fall into three categories: GIS data layers, databases and protocols. GIS data layers were created or obtained for stream segments, drainage basins, cultural features, water quality, stream size, habitat quality, stream gradient, riparian forest cover, precipitation, soils, landcover, surficial geology, depth to bedrock, bedrock geology, point source pollution, priority waters status, and managed and protected areas (including state parks, wildlife management areas, regulated wetlands, public fishing rights, managed water quality, stocked waters, and regulated fishing areas). The databases include: classifications of fish species by habitat type (habitat association, pollution tolerance, stream size association), classifications of macroinvertebrate families by habitat type (pollution tolerance, life habits and substrate association, feeding guild), and collections of fish and macroinvertebrate field data. Lastly, automated protocols were developed for characterization of stream size, gradient, habitat quality, and riparian forest cover, and water quality through a nonpoint source pollution model (modification of current method).