banner literature about projects tools meetings search links BBS GAP home GAP home
home

 


<-
Previous | First | Next ->

Assessing Data Quality and Information
Distribution Prospects for County Conservation
Planning Using New Mexico Gap Analysis Data
A Report on GAP Research Project #14-45-0009-1572

ROBERT A. DEITNER
New Mexico Cooperative Fish and Wildlife Research Unit, and 
Department of Fishery and Wildlife Sciences, New Mexico State
University, Las Cruces

BRUCE C. THOMPSON
New Mexico Cooperative Fish and Wildlife Research Unit, and 
U. S. Geological Survey

AND JULIE S. PRIOR-MAGEE
New Mexico Cooperative Fish and Wildlife Research Unit, and 
U. S. Geological Survey

The New Mexico Gap Analysis Project (NM-GAP) was completed
in December 1996. As one of the early projects to complete such
data sets and analysis, we sought to evaluate the quality and utility
of the resulting products, have staff available for consultation with
other GAP projects and potential users, and investigate other pos-
sible uses of Gap Analysis products for conservation planning. Our
objectives were to 1) assess data quality and accuracy patterns for
42 mapped land cover classes in existing digital map files, 2) pro-
vide digital files and cooperative assistance to edge-match 42 New
Mexico land cover class assignments and predicted distributions
for 584 New Mexico animal species relative to comparable infor-
mation for Arizona, Colorado, Oklahoma, Texas, and Utah, and 3)
analyze predicted distributions of 584 animal species relative to 36
classes of land stewardship, four classes of management status, and
33 county boundaries for public conservation planning applications.

Data Sharing and Edge-Matching
Early on, we contacted personnel from Gap Analysis projects in all
states surrounding New Mexico. We provided technical contact
information, made NM-GAP data sets available, and offered assis-
tance in transfer and comparative analysis of mutual data sets. There
was minimal response from the other projects. It appeared that
those projects generally were not yet prepared to examine appre-
ciable edge-matching questions during the time frame of our re-
search. This remains an important communication challenge as
GAP projects progress on staggered completion schedules. [Editor’s
note: The national GAP office continues to undertake regionalization
of data. Products are available at http://www.gap.uidaho.edu/
Projects.
]

Data Quality Assessment
An initial assessment of the NM-GAP data set was prepared in out-
line form and sent to the people involved with the initial project.
We also asked national GAP and the University of New Mexico
Earth Data Analysis Center (EDAC) to comment on data quality.
We did not receive substantive critical comments in response to
these inquiries; so that document was the starting point for our da-
tabase improvement. During our project, national GAP was pro-
ducing NM-GAP data on CDs, but that process was independent of
NM-GAP, and we were not directly involved in any digital modifi-
cations made during production. National GAP corrected some
topological problems in the original vector coverages. Use of the
data by others and us predictably identified some errors in the data.
The most substantial was that our original coverages actually used
the NAD27 datum rather than NAD83 as was reported to us by the
university’s Geography Department, our original partner in spatial
analyses and data product development. Metadata were changed to
reflect this error.

Accuracy assessment for the NM-GAP land cover map was ad-
equate for directly describing overall quality of the map but was
not adequate to identify sources of errors. We ignored spatial error
during map assessment, but this probably caused problems with
observers attempting to view polygons in the field because some
boundaries were not distinct. We also discovered that there is sub-
stantial subjectivity to land cover designation. Different people
assign different land cover types to the same parcel of land. This
could be for several reasons, all unknown in magnitude and impli-
cations. Future use of the National Vegetation Classification Sys-
tem and mapping at finer resolution likely will ameliorate many of
these problems.

The theory of development of the animal algorithm (expressing
species distribution as a logical function of landscape attributes and
being able to develop these relations with the advice of experts)
seems to work. The execution of algorithms needs improvement.
In particular, vector modeling should be abandoned. It should be
executed on one platform only, and ways to make it faster and more
user-friendly should be explored.

Our review of data quality indicated a high degree of confidence in
the herpetozoan and mammal distribution predictions but a lesser
degree of confidence in the birds. We extensively reviewed the bird
distributions (324 species year-round and 257 breeding species state-
wide) with two statewide experts and two graduate research assis-
tants familiar with New Mexico birds. We focused on species for
which there was consistency in suggested revisions among expert
reviewers. Ultimately, we altered only 21 species’ predictive algo-
rithms (20 birds, one mammal). Most of the 21 species changes
were in watersheds, not in habitat relations. The swift fox (
Vulpes
velox
) algorithm was altered because of a misinterpretation of ini-
tial expert opinion that was not discovered until after the original
NM-GAP report was filed.

We devoted particular attention to labeling of attributes in the stew-
ardship map. Our original stewardship map was based on the Bu-
reau of Land Management’s Public Land Survey System (PLSS)
map (1995) with additional polygons that we generated to repre-
sent special management tracts not contained in the original map.
The attribute table was given to us without metadata, so it took a bit
of detective work to discern that the fields indicated the original
PLSS identification of landowner and management status, a de-
scription of the tract (polygon) if it was derived by our lab, and its
final ownership and management status. National GAP examined
the coverage for topological problems, and during this process we
discovered a few corrupt polygons, some spurious quadrant bound-
aries, and a few polygons placed incorrectly outside the state bound-
ary. We made the underlying vector map into two grids with the
same topology as the land cover maps. In the future, the grid ver-
sion should consist of one coverage that identifies the polygons as
well as owner and management status.

Our stewardship coverage is useful for identifying broad categories
of ownership, but specific tracts or complex land management agree-
ments are difficult or impossible to identify. Utility of this cover-
age could be greatly enhanced if data structures allowed easy ex-
traction of specific land tracts. For example, individual wilderness
areas and wildlife refuges are not indicated by name or identifier in
either the PLSS map or our modifications. Other difficulties exist
as some tracts are subject to complicated agreements between agen-
cies that cloud classifications to single land stewards. An example
is a military installation variously composed of military land, other
federal withdrawn land, and agreement or contract lands. Serious
early planning is necessary to ensure such tracts are thoroughly
understood and attributed properly in the GIS to extract desired
land summaries later. Similar considerations apply to a variety of
federal and state land holdings.

Land Cover and Animal Conservation Assessment
For the 18 most sensitive land cover classes on the New Mexico
landscape, most have substantial percentages of their distribution

on lands under U.S. Forest Service (USFS), Bureau of Land Man-
agement (BLM) , private, state trust, or tribal stewardship. Combi-
nations of U.S. Forest Service, BLM, and private land stewards ac-
count for >60% of the occurrence of 16 of the 18 sensitive classes.
For the sensitive wooded, higher-elevation land cover classes, USFS
and private land stewards likely have the most opportunity to affect
conservation of these types, whereas BLM, tribal, state trust, and
private land stewards are more implicated in conservation opportu-
nities for sensitive shrubland and grassland types. Sensitive ripar-
ian land cover classes overwhelmingly occur on private lands, with
some types having appreciable occurrence on BLM and Bureau of
Reclamation lands.

The 35 most sensitive vertebrate animal species, as defined by pre-
liminary gap analysis practices, occur primarily on a combination
of private, BLM, and state trust lands. The greatest percentage of
occurrence is estimated to be under private land stewardship for 25
of the species and BLM for 10 species. The second greatest per-
centage of occurrence for 18 of the species is on state trust lands.
Thus, a combination of private, BLM, and state trust lands appear
implicated in appreciable conservation opportunities for these spe-
cies. The frequency and extent of occurrence of the 35 most sensi-
tive vertebrate species on refuges operated by the U.S. Fish and
Wildlife Service currently are insufficient to meet conservation in-
terests for any of the species. Of the 35 species, 15 are estimated to
occur on refuges in New Mexico. Just seven species occur on more
than one refuge, and none of those species has >1% of its distribu-
tion on a national refuge.

Extension to County Planners
Information on the ecological and land steward context for each
county is needed to judge the merits of how conservation planning
in individual counties may aid in providing benefits to the at-risk
natural land cover classes and sensitive species. We provided two
documents to county planners. The first, “A Proactive Approach to
Conservation Planning: How New Mexico Counties Can Use Gap
Analysis Project Data,” provided the counties with: 1) basic infor-
mation on the Gap Analysis Program; 2) factors to consider when
using NM-GAP data; 3) examples of uses of GAP data at the county
level; and 4) information on how to obtain NM-GAP data via the
Internet and CD-ROM. The second document, “NM-GAP Data
Needs Inquiry for County Planners,” contained questions regard-
ing the county’s potential use of NM-GAP data. These questions
related to: 1) the specific type of data the county would find most
useful in its planning activities, 2) the county’s interest in data for
specific species or land cover classes, 3) the current mapping sys-
tem used by the county (e.g., GIS program or hand-drawn maps),
and 4) how NM-GAP data could be made easily available and use-
ful to the county planning office given its technical capabilities and
data needs.

We contacted 33 county planners or county officials responsible
for planning. Eight counties scattered across the state responded to
the data needs inquiry (Table 1). The response of county planners,
while not showing enthusiastic interest in using NM-GAP data state

Table 1.Summary of responses from eight New Mexico counties regarding prospective use of NM-GAP data 
products, summer, 1998.

Needs Questiions County Ressponse
What type of NM-GAP data would your county find most useful? Land cover, land use, hydrology, and man-made features.
Location of vertebrate species and land cover classes, and associations of
animals with land cover classes. Particularly as these relate to land use or
development applications and as they relate to sensitive, endangered, or at-risk species.
Would your county be interested in species-at-risk data? Yes, to analyze how development might impact these species and other natural resources.
Yes, to use for solid waste projects.
Would not use this information now unless required by federal or state law.
What specific species or land cover classes would be of interest to your county? Antelope and their native food habitat.
Grasses.
To be determined later.
In the future, will be interested in endangered species or species perceived as threatening to
humans, livestock, or domestic animals.
Mule deer, prairie chicken, blue and bobwhite quail.
What mapping system is used by your county and who is responsible? ArcView and AutoCAD used in-house by GIS staff.
ArcInfo used by Utilities Department.
Fastmap 7000 used in-house. In the process of implementing GIS E911.
Hand-drawn maps used at present, but in 2-3 years will implement a GIS E911 system.
No computer mapping capability presently, but will work toward a county GIS system.
AutoCAD and hand-drawn maps. In the near future will have a GIS system.
How would your county prefer to access data? Internet and CD-ROM.
Paper maps now with the potential for future use of digital data.
Other comments. I believe that natural resource assessment is the first step in good planning. I support the
concept of this NM-GAP project

Considering the 33 New Mexico counties as conservation planning
districts, just seven have >10% of their area in Status Class 1 and 2;
10 counties have greater than 90% of their area in Status Class 4.
Of the other 16 counties, at least 11 have appreciable percentages
of land in Status Class 3 that could be evaluated for opportunities to
enhance conservation provisions sufficient to warrant reclassifica-
tion to Status 2. Clearly such action would need to take the inter-
ests of individual land stewards into account.

We constructed an extensive set of tabulations to help county plan-
ning interests judge how their jurisdiction relates to land cover state-
wide, sensitive animal species, and land cover within Status Class
1 and 2. We identified the distribution of 42 land cover classes
among counties that can be used to judge the relative degree to
which a county can affect conservation of a specific class among
counties as well as in relation to other land uses in the county. A
tabulation of the percentage of habitat among counties for the 35
sensitive animal species allows county planners and others inter-
ested in county-level conservation planning to judge a county’s
possible contribution to conserving species habitat (e.g., Colfax
County contains all of the habitat for the prairie vole
Microtus
ochrogaster
, Grant County has nearly 45% of habitat for the Sonoran
spotted whiptail
Cnemidophorus sonorae ). Another tabulation can
extend planning further by examining the degree that various land
cover classes (especially the most restricted ones) are already dis-
tributed on Status Class 1 and 2 lands in each county. All of the
handily read tabulations allow for detailed understanding of how
land cover classes, animal species, and stewardship can be accounted
for in specific county-level conservation planning questions.

Implications
From our data review, we believe the following actions warrant
attention in further work with the NM-GAP data and in the next
generation of gap analysis in the Southwest.

• Specific attention is needed to derive an efficient process for
managing animal algorithms, especially in attempts to consoli-
date predictive work among past project databases. We recom-
mend against using vector-based processing.

• Greater specificity is needed in assigning and tracking steward-
ship categories. Simply compiling previous maps evolved for
other purposes does not allow desired capability to extract the
accurate boundaries of tracts or stewardship categories of inter-
est.

• Critically examine how preclustering operations relate to as-
signment of land cover boundaries for recognizable natural fea-
tures.

• Recognize how far in time and computing capability a county’s
planning functions are from effective use of GAP products.
Similarly, recognize the opportunity that exists, at least in New
Mexico, to work with counties in the formative stages of de-
signing and delivering data products that they can immediately
adapt to county-level planning operations.

Suggested Readings
Deitner, R.A., B.C. Thompson, and J.S. Prior-Magee. 1999. As-
sessing inter-project data compatibility and information distri-
bution for conservation planning using New Mexico Gap Analy-
sis data. Research Completion Report to USGS National Gap

Analysis Program, Research Work Order No. 29, New Mexico
Cooperative Fish and Wildlife Research Unit, New Mexico State
University, Las Cruces.

Edwards, T.C., Jr., E.T. Deshler, D. Foster, and G.G. Moisen. 1996.

Adequacy of wildlife habitat relation models for estimating spa-
tial distributions of terrestrial vertebrates.
Conservation Biol-
ogy
10:263-270.

Forester, D.J., G.E. Machlis, and J.E. McKendry. 1996. Extend-
ing gap analysis to include socioeconomic factors. Pages 39-
53 in J.M. Scott, T.H. Tear, and F.W. Davis, editors. Gap analy-
sis: A landscape approach to biodiversity planning. American
Society for Photogrammetry and Remote Sensing, Bethesda,
Maryland.

Johnson, N.C. 1995. Biodiversity in the balance: Approaches to
setting geographic conservation priorities. Biodiversity Sup-
port Program, consortium of World Wildlife Fund, The Nature
Conservancy, and World Resources Institute, Washington, D.C.

Scott, J.M., T.H. Tear, and F.W. Davis, editors. 1996. Gap analy-
sis: A landscape approach to biodiversity planning. American
Society for Photogrammetry and Remote Sensing, Bethesda,
Maryland.

Thompson, B.C., P.J. Crist, J.S. Prior-Magee, R.A. Deitner, D.L.

Garber, and M.A. Hughes. 1996. Gap analysis of biological
diversity conservation in New Mexico using geographic infor-
mation systems. Research Completion Report to USGS Na-
tional Gap Analysis Program, Research Work Order No. 13, New
Mexico Cooperative Fish and Wildlife Research Unit, New
Mexico State University, Las Cruces.

<-Previous | First | Next ->