National Gap Analysis Program, Moscow, Idaho
As human populations continue to increase, we see a corresponding increase in the demand for
space and for consumptive uses of our natural resources (for example, timber, water, and
minerals). This in turn hastens loss of biodiversity by fragmenting habitat and disrupting
ecological processes. Our traditional approach to development and use of natural resources has
been haphazard and conducted more by accessibility and opportunity than by careful
comprehensive planning to integrate and protect other resources. During the first decade, the
GAP program successfully explored technical methodologies to generate spatially explicit
information for key elements of biodiversity (Jennings et al. 1999). However, information alone
is not enough. It is crucial to focus future activities for the Gap Analysis Program on synthesis
and interpretation of the data and to continue to apply GAP data to test new hypotheses for
reserve identification, selection, and design.
Last year the U.S. Geological Surveys (USGS) Geologic Division and Biological Resources
Division collaborated on a research proposal, which was subsequently funded through the
Integrated Science Program as an initiative of the Director of the USGS. The Geologic Division
is in the process of completing a three-year feasibility study to conduct nonfuel mineral resource
assessments. Recognizing that exploration and development can have disruptive effects on
biological resources, either from the direct consequences of mining or from simply opening up
an area to rapid growth by other sectors, the geologists wish to integrate other sources of data on
biodiversity and water resources. This practical approach would augment the
comprehensiveness of their assessment and focus the exploration planning. The Spokane Field
Office of the Geologic Division and the GAP Operations office in Moscow, Idaho, are working
together to develop fundamental data layers from each division. Once generated, these layers
can be synthesized to create a three-dimensional output of rankings for the landscape ranging
from high mineral potential, strong juxtaposition of biodiversity elements and conservation status
to low biological value, low mineral potential, and no conservation status.
The two divisions chose to test the process with the state of Montana because both GAP data and
minerals data are available. I used an ARC/INFO Arc Macro Language (AML) script written by
Jason Karl, formerly of the Landscape Dynamics Lab at the University of Idaho. The algorithm
starts with areas that capture the most number of species and then iteratively selects additional
areas on the landscape to add new species. This AML (called "r select") is available on our Web
site at http://www.gap.uidaho.edu/Scripts/default.htm and was originally used in the Gap
Analysis project for the Lewis and Clark Trail (Crist and Jennings 1999).
I resampled Montanas vertebrate grids from 90-
meter to 1000-meter resolution and created
hyperdistribution grids for the taxa (nonpasserine and passerine birds, mammals, amphibians, and reptiles.) I ran the reserve AML on each hypergrid and created another grid of
selected reserve sites for each group of animals. I used the entire extent of the state for the
processing. I merged the grids to make one final grid of all reserve sites (Figure 1). The points
in the map show the location of potential reserve areas but are not scaled to size.
Figure 1. Map of Montana with current conservation lands (Status 1 and 2) and the results of the
predicted reserve sites.
The tabular results of merging the reserve grids are displayed in Table 1. Each site shows the
number of cells and the number of species per taxon selected. For example, Site 4 has 147
contiguous cells and captures 25 mammal species. The zeroes for the remaining species do not
mean that the animals were not predicted to occur there, simply that the site was not selected by
the algorithm for those groups. The size of the selected sites range from 1 to 280 cells. The
richness columns are an output of the reserve AML. The columns show the sites selected to
acquire all of the species for one group. For example, 11 sites represent all mammals across the
state of Montana. While approximately 2.5% of all the predicted reserve sites overlap with the
conservation lands already established, there was no overlap in the sites selected for each taxon.
Table 1. Summary of sites for all groups.
Limitations to the Algorithm - The distribution of the selected reserve sites appears to be fairly well
dispersed and would suggest a diverse range of environments for all species have been selected.
However, this is only a first look at a very complex issue. The algorithm was designed to create a representative sample of sites to include all species. There is no decision rule for maximizing area. This is evident when comparing the output area of the sample reserve sites (1191 km 2 ) to the area of Status 1 and 2 land (29,078 km 2 ). Figure 2 illustrates one selected site. The algorithm may emphasize edge habitat and, if so, would favor common species over rare species. These would be adverse solutions for species that require continuous habitat and especially for those species of special concern that have lost critical habitat. Additionally, the algorithm is not sensitive to spatially constrained species, i.e., species that only occur in some areas and not other areas of similar habitat. It requires a closer look for sensitive, threatened, orendangered species.