home
<- Previous | First | Next ->
National Gap Analysis Program, Moscow, Idaho
The eleven western states have completed their first round of mapping for GAP. The Gap
Analysis Program has always had the objective of moving beyond state boundaries to create
regional maps and making regional, and ultimately national, assessments. Beyond reporting,
regional and rangewide maps provide a frame of reference for more in-depth analyses of
individual species or vegetation types across their range.
I selected several species for the taxa (mammals, birds, reptiles, and amphibians) from the eleven
western states. The GIS coverages were converted to a common projection and cell resolution
and then merged. Some patterns emerged and are particularly relevant to the work done by Karl
et al. (1999). The authors conducted a review of primary literature (peer-reviewed journals) to
assess how many articles had been written on selected species in three categories and how many
of those articles described habitat. I used their three groups (game species, species of special
interest, and species that have general habitat requirements) and added one more “species that
have special (e.g., microhabitat) requirements”. I compared the merged model outputs from the
eleven states with availability of habitat information as per Karl et al. (1999).
Karl et al. (1999) found numerous articles on elk (ervus elaphus), 30 of which had descriptions
of habitat. Figure 1 shows the merged state models with consistent mapping across state
boundaries. This suggests that, for the most part, there were fairly similar ideas about elk
habitat. In contrast, the models for chukar (Alectoris chukar), another game species, do not
match well and, interestingly, there was only one article found in the primary literature
referencing habitat (Figure 2). Elk habitat is generally described as coniferous forest, alpine
meadows, marshy meadows, and shrub steppe. These are all components that are fairly
commonly mapped in land cover. Components of chukar habitat range from mountain slopes of
grassy vegetation to deserts with sparse grass, barren plateaus, and rocky hillsides. These are not
categories that are commonly included in land cover mapping from one state to another and may
contribute to some discrepancies across borders. Therefore, two possible explanations for the
discrepancies seen here for chukar are: 1) because of a lack of habitat information for a species,
there are varying ideas of habitat preferences from one state to the next, or 2) even if we have a
common understanding of habitat, it may not be mapped consistently in the land cover from one
state to the next.
The species in this category are experiencing declining populations, most commonly as a result
of declining habitat. Two examples are southern red-backed vole
(Clethrionomys gapperi) and the fringed myotis (Myotis thysanodes). The southern red-backed vole had 20 references to
habitat in the primary literature, whereas the fringed myotis had zero. Figures 3 and 4 illustrate
the model outputs. The vole’s habitat preferences are cool, moist deciduous, mixed, or
coniferous forests whereas the fringed myotis prefers desert, grassland, and woodland habitats.
The agreement for the vole suggests an underlying agreement in habitat and land cover elements,
but the myotis map shows more discontinuities. Typically, grasslands and woodland have been
very difficult to map using satellite imagery. Moreover, without descriptions available in the
primary literature there may be inconsistencies in the understanding of habitat use from state to
state as seen for the fringed
myotis.
I consider microhabitat features to be small features such as wetlands, snags, rocks, and down
woody debris, or linear features such as streams. These features are not generally mappable
from satellite imagery. Figure 5 shows the American beaver
(Castor canadensis) and Figure 6 the bald eagle (Haliaeetus leucocephalus), two species not surveyed by Karl et al. (1999). Both species require a component of water, either adjacent to woods for the beaver or snags for the
eagle. Clearly, these are components that have been mapped quite differently from one state to
the next. Table 1 summarizes the mapped water categories by state. The states varied
dramatically in their thematic resolution for this component. Some indicate floristic composition
(cottonwoods or willows) while others show structure (tree vs. riparian) and elements such as
wetlands and vernal pools.
Other habitat requirements difficult to delineate at this scale are orchards, city parks, shade trees,
and agricultural features such as irrigated crops. The thematic resolution varied from state to
state as seen in Figures 7 and 8 and Table 2 for species that use these features.
|
|
Figure 10. Predicted distribution of Black-capped Chickadee habitat (Parus atricapillus) |
Figure 9. Predicted Distribution of Mule Deer habitat (Odocoileus hemionus) |
This category of species had some surprising results. Intuitively, one would expect ubiquitous
species to show the most continuity in mapping. Figures 9 and 10 contradict this assumption.
Karl et al. (1999) found 51 articles addressing habitat for mule deer (Odocoileus hemionus) and
14 for the black-capped chickadee (
Parus atricapillus). Despite the wealth of information, there
is still quite a bit of discrepancy between state model outputs for each. Both species have
habitats that have been noted earlier for inconsistency in mapping (woodlands, parks,
brushlands).
A cursory examination of merged vertebrate models for the western states reveals some
interesting results. Continuity, or the lack of it, across state boundaries is influenced by several
factors:
5. available information about the habitat preferences for the species,
6. mappability of habitat elements as land cover or ancillary layers, and
7. seamless minimum mapping unit, thematic resolution and mapping categories of base data
for modeling vertebrate distributions.
The GAP projects continue to test methods to increase mappability of features, and we have
moved to more consistent minimum mapping units (two hectare to single, 30 meter pixel).
Because description of habitats for species is largely unavailable in primary literature, it would
behoove us to access, and to contribute to, secondary sources such as biological Web sites (e.g.,
the Fish and Wildlife Information Exchange), the National Biological Information Infrastructure
(metadata serving of information), NatureServe (Association for Biodiversity Information or
ABI) and Ecological Archives (appendix and supplemental materials to Ecological Society of
America publications).
Thanks to Derek McNamara for his assistance in processing the state coverages and for
summarizing the tabular data.
Karl, J., N.M. Wright, P.J. Heglund, and J.M. Scott. 1999. Obtaining environmental measures
to facilitate vertebrate habitat modeling.
Wildlife Society Bulletin 27(2):357-365.
<- Previous | First | Next ->