442
Mapping Patterns of Human Use and Potential Resource Conflicts on Public Lands
tional Forest, including 4 amphibians, 11 mammals,
and 26 birds. Examples include the Coeur d'Alene
salamander (Plethodon idahoensis), Townsend's
big-eared bat (Corynorhinus townsendii), gray wolf
(Canis lupus), lynx (Lynx canadensis), common
loon (Cavia immer), northern goshawk (Accipiter
gentilis), and flammulated owl (Otus flammeolus).
When predicted habitat is overlaid for all 41
species, richness values range from 0 to 18 species
per 90-m2 cell (Figure 29.3b). Although here we
focus on overall richness, each species merits individual consideration as well. In this landscape,
species of special concern tend to be associated
with forested areas, whereas human populations are
concentrated in lower, open areas. High species
richness counts and concentrated human use do,
however, coincide in places (shown in dark blue in
Figure 29.3c), especially where density of roads
and trails is high. And even in areas where human
use is highest, there is almost always at least one
species of special concern to be considered. By intersecting layers of predicted species richness and
human use, we are better able to identify areas
where conflicts are more likely to occur.
29.4.3 Impacts to Bull Trout
This type of analysis allows public land managers
to identify specific areas (stream reaches) where
continued or increased levels of human use could
adversely affect bull trout habitat and then to factor this information into management plans. In this
example, the most important areas to examine are
those with high levels of predicted human use
(shown in dark gray in Figure 29.4). If predicted
use is high at key points in the stream network, this
could affect the accessibility of upstream habitat
and thereby the future security of bull trout
throughout larger watersheds. Results also show
the very limited proportion of spawning and rearing streams (4.9% in the entire National Forest
boundary) that fall outside a l-km zone of predicted
human impact and that these relatively undisturbed
streams tend to occur in the uppermost portions of
watersheds (shown as double-line segments in Figure 29.4).
29.4.4 Limitations
We are not attempting with this chapter to present
a method complete with ideal parameters or specific results for a given area. Rather, our goal is to
illustrate a flexible way to incorporate census data
with other GIS layers, creating a far more accurate
picture of human presence in ecosystems. We present the use of modifiable weighting ratios as an
aid in this process. For example, the 80-10-5-5 ratio used to allocate population counts to regions is
arbitrary and could be improved if revised based
on actual field study or other empirical evidence.
One option might be to examine correlations between census data and land cover types through
multivariate regression, as Yuan et al. (1997) have
done for four counties in Arkansas. The largest differences, however, between the dasymetric maps
shown here and traditional choropleth maps arise
from the exclusion of uninhabited lands, not the
80-10-5-5 weighting. Again, the method we present here offers a general approach; many of its
steps require parameters to be defined, yet insufficient information is available to support these
definitions.
Another limitation of the method is its reliance
on fairly coarse scale data (e.g., census data) to predict fine-scale effects. To some extent, the use of
a search radius (in this case, 1 km) in generating
model output helps to compensate for this limitation by projecting results across broader areas.
Nonetheless, as is always the case, users should
bear in mind the nature of the data inputs when interpreting outputs.
29.5 Conclusions
In any ecosystem, humans are an integral part of
the equation, but traditionally they have been neglected from an analytical standpoint. The approach described herein offers a management tool
for predicting, evaluating, and visualizing human
impacts in a spatial sense. But analyses need not
be limited to where people live and where they
might travel. Human use could be evaluated in
other ways as well. For example, point data on
campgrounds, mines, summer cabins, and other anthropogenic features could be incorporated or distance to urban areas measured and weighted. Although this approach is well suited to assessment
of human impacts on other terrestrial and aquatic
species, it could be applied to problems in all disciplines, particularly risk assessments. One potential application is mapping fire risk; both the likelihood of human-caused ignitions and potential
damage to property could be assessed. Mapping
risk of exposure to contaminants is another possibility; for example, human use could be analyzed
with respect to pollution levels around Superfund
sites. In fact, a simple protocol like this could assist land managers and planners with future decisions in any area where suitable input data are available.
Mapping Patterns of Human Use and Potential Resource Conflicts on Public Lands
tional Forest, including 4 amphibians, 11 mammals,
and 26 birds. Examples include the Coeur d'Alene
salamander (Plethodon idahoensis), Townsend's
big-eared bat (Corynorhinus townsendii), gray wolf
(Canis lupus), lynx (Lynx canadensis), common
loon (Cavia immer), northern goshawk (Accipiter
gentilis), and flammulated owl (Otus flammeolus).
When predicted habitat is overlaid for all 41
species, richness values range from 0 to 18 species
per 90-m2 cell (Figure 29.3b). Although here we
focus on overall richness, each species merits individual consideration as well. In this landscape,
species of special concern tend to be associated
with forested areas, whereas human populations are
concentrated in lower, open areas. High species
richness counts and concentrated human use do,
however, coincide in places (shown in dark blue in
Figure 29.3c), especially where density of roads
and trails is high. And even in areas where human
use is highest, there is almost always at least one
species of special concern to be considered. By intersecting layers of predicted species richness and
human use, we are better able to identify areas
where conflicts are more likely to occur.
29.4.3 Impacts to Bull Trout
This type of analysis allows public land managers
to identify specific areas (stream reaches) where
continued or increased levels of human use could
adversely affect bull trout habitat and then to factor this information into management plans. In this
example, the most important areas to examine are
those with high levels of predicted human use
(shown in dark gray in Figure 29.4). If predicted
use is high at key points in the stream network, this
could affect the accessibility of upstream habitat
and thereby the future security of bull trout
throughout larger watersheds. Results also show
the very limited proportion of spawning and rearing streams (4.9% in the entire National Forest
boundary) that fall outside a l-km zone of predicted
human impact and that these relatively undisturbed
streams tend to occur in the uppermost portions of
watersheds (shown as double-line segments in Figure 29.4).
29.4.4 Limitations
We are not attempting with this chapter to present
a method complete with ideal parameters or specific results for a given area. Rather, our goal is to
illustrate a flexible way to incorporate census data
with other GIS layers, creating a far more accurate
picture of human presence in ecosystems. We present the use of modifiable weighting ratios as an
aid in this process. For example, the 80-10-5-5 ratio used to allocate population counts to regions is
arbitrary and could be improved if revised based
on actual field study or other empirical evidence.
One option might be to examine correlations between census data and land cover types through
multivariate regression, as Yuan et al. (1997) have
done for four counties in Arkansas. The largest differences, however, between the dasymetric maps
shown here and traditional choropleth maps arise
from the exclusion of uninhabited lands, not the
80-10-5-5 weighting. Again, the method we present here offers a general approach; many of its
steps require parameters to be defined, yet insufficient information is available to support these
definitions.
Another limitation of the method is its reliance
on fairly coarse scale data (e.g., census data) to predict fine-scale effects. To some extent, the use of
a search radius (in this case, 1 km) in generating
model output helps to compensate for this limitation by projecting results across broader areas.
Nonetheless, as is always the case, users should
bear in mind the nature of the data inputs when interpreting outputs.
29.5 Conclusions
In any ecosystem, humans are an integral part of
the equation, but traditionally they have been neglected from an analytical standpoint. The approach described herein offers a management tool
for predicting, evaluating, and visualizing human
impacts in a spatial sense. But analyses need not
be limited to where people live and where they
might travel. Human use could be evaluated in
other ways as well. For example, point data on
campgrounds, mines, summer cabins, and other anthropogenic features could be incorporated or distance to urban areas measured and weighted. Although this approach is well suited to assessment
of human impacts on other terrestrial and aquatic
species, it could be applied to problems in all disciplines, particularly risk assessments. One potential application is mapping fire risk; both the likelihood of human-caused ignitions and potential
damage to property could be assessed. Mapping
risk of exposure to contaminants is another possibility; for example, human use could be analyzed
with respect to pollution levels around Superfund
sites. In fact, a simple protocol like this could assist land managers and planners with future decisions in any area where suitable input data are available.