Document 12786854

advertisement
Gap Analysis of Conserved Genetic Resources
for Forest Trees
SARA R. LIPOW,∗ †† KENNETH VANCE-BORLAND,∗ J. BRADLEY ST. CLAIR,† JAN HENDERSON,‡
AND CINDY MCCAIN§
∗
Department of Forest Science, Oregon State University, Corvallis, OR 97331, U.S.A.
†U.S. Forest Service, Forestry Sciences Laboratory, Corvallis, OR 97331, U.S.A.
‡U.S. Forest Service, Supervisor’s Office, Mt. Baker–Snoqualmie National Forest, Mountlake Terrace, WA 98043, U.S.A.
§U.S. Forest Service, Siuslaw National Forest, Corvallis, OR 97339, U.S.A.
Abstract: We developed a gap analysis approach to evaluate whether the genetic resources conserved in
situ in protected areas are adequate for conifers in western Oregon and Washington (U.S.A.). We developed
geographic information system layers that detail the location of protected areas and the distribution and
abundance of each tree species (noble fir [Abies procera Rehd.] and Douglas-fir [Pseudotsuga menzeisii Mirb.]).
Distribution and abundance were inferred from available spatial data showing modeled potential and actual
vegetation. We stratified the distribution of each species into units for genetic analysis using seed and breeding
zones and ecoregions. Most strata contained at least 5000 reproductive-age individuals in protected areas,
indicating that genetic resources were well protected in situ throughout most of the study region. Strict in
situ protection was limited, however, for noble fir in the Willapa Hills of southwestern Washington. An in situ
genetic resource gap arguably occurred for Douglas-fir in the southern Puget lowlands, but this gap was filled
by extensive ex situ genetic resources from the same region. The gap analysis method was an effective tool for
evaluating the genetic resources of forest trees across a large region.
´
Análisis de Claros de Recursos Gen´eticos Conservados para Arboles
de Bosque
Resumen: Desarrollamos un m´etodo de an´alisis de claros para evaluar si los recursos gen´eticos conservados
in situ en áreas protegidas son adecuados para conı́feras en el oeste de Oregon y Washington (E. U. A.).
Desarrollamos capas de sistema de informaci´
on geogr´
afica que detallan la localizaci´
on de areas
protegidas y
´
la distribuci´
on y abundancia de cada especie de arbol
(Abies procera Rehd. y Pseudotsuga menzeisii Mirb). La
´
distribución y abundancia fueron inferidas a partir de datos espaciales disponibles que muestran la vegetación
potencial modelada y la vegetaci´
on existente. Estratificamos la distribuci´
on de cada especie en unidades para
el an´
alisis gen´etico utilizando zonas de semillas y reproducci´
on y ecoregiones. La mayor´ıa de los estratos
contenı́an por los menos 5000 individuos en edad reproductiva en areas
protegidas, lo que indica que los
´
recursos genéticos estaban bien protegidos in situ en casi toda la regi´
on de estudio. Sin embargo, la protecci´
on
in situ estricta estaba limitada para A. procera en las Colinas Willapa del suroeste de Washington. Se podrı́a
decir que ocurrió un claro de recursos genéticos in situ para P. menzeisii en las tierras bajas del sur de Puget,
pero este claro fue llenado por recursos genéticos extensivos ex situ de la misma región. Encontramos que el
método de an´
alisis de claros es una herramienta valiosa para evaluar los recursos gen´eticos de arboles
de
´
bosque en una región amplia.
††Current address: 2600 State Street, Oregon Department of Forestry, Salem, OR 97310, U.S.A., email lipow@odf.state.or.us
Paper submitted February 19, 2002; revised manuscript accepted July 29, 2003.
412
Conservation Biology, Pages 412–423
Volume 18, No. 2, April 2004
Lipow et al.
Introduction
Biological diversity refers to the variety and abundance of
species and the communities in which they occur and to
the genetic composition of individual species. Land-use
changes, disease conditions, and climatic change directly
threaten forest species and communities. They also jeop­
ardize the genetic variation that enables tree species to
evolve and thrive under changing environmental condi­
tions. Threats to tree species can affect the long-term sur­
vival of the associated flora and fauna. Genetic variation
of forest trees is also essential for sustainable production
of forest products and therefore has important social and
economic implications.
Concerns for biological diversity and the genetic as­
pects of sustainable forest management prompted a group
of public and private organizations in western Oregon and
Washington to form the Pacific Northwest Forest Tree
Gene Conservation Group. One objective of this group is
to identify whether areas in the region exist where addi­
tional conservation measures are necessary to ensure that
the adaptation and evolutionary potential of important
tree species are maintained. To identify possible areas,
we have compiled data on genetic resources conserved
both at their original location (in situ) and at some other
location (ex situ).
We developed a gap analysis approach to investigate ge­
netic resources conserved in situ in protected areas. We
present results for Douglas-fir (Pseudotsuga menziesii
[Mirb.] Franco var. menziesii) and noble fir (Abies pro­
cera Rehd.). Results for six other tree species are reported
separately: Tsuga heterophylla (Raf.) Sarg., Thuja plicata
Donn ex D. Don, Pinus ponderosa Dougl. ex Laws, Picea
sitchensis (Bong.) Carr., Pinus monticola Dougl. ex D.
Don, and Pinus lambertiana Dougl. (S.R.L., K.V.-B., J.B.S.,
J.A.H., & C.M., unpublished data). These species are com­
mon in western Oregon and Washington (U.S.A.) and are
commercially important.
The in situ genetic resources we evaluated are only
one component of an overall gene conservation strat­
egy (Yanchuk & Lester 1996; Lipow et al. 2001). The
tree species also have extensive genetic resources in ex
situ collections, including progeny tests, seed orchards,
and seed stores (Lipow et al. 2001; S.R.L., K.V.-B., J.B.S.,
J.A.H., & C.M., unpublished data). In Oregon and Wash­
ington, progeny from >1679 noble fir selections from
natural populations are maintained in genetic tests or
in 1 of 14 seed orchards. The tested selections span
the species’ range, excluding the Willapa Hills of south­
western Washington. Hundreds of additional selections
are maintained in Europe. For Douglas-fir, over 1 million
progeny from >29,000 selections are maintained in re­
gional first-generation progeny tests. Second-generation
tests will contain >2000 of the selections evaluated in
the first-generation tests. Regional seed stores include
>1460 and >20,000 seed lots stored by family for noble
Genetic Gap Analysis for Forest Trees
413
fir and Douglas-fir, respectively. These ex situ collections
are a valuable component of the total available genetic
resources for the species.
The goal of our in situ analysis was to identify places
where large populations of each species are protected
in reserves and places where few or no trees are pro­
tected (gaps). We did this by performing a gap analysis
in a geographic information system (GIS). Gap analysis
typically refers to a scientific process that identifies the
degree to which native species and natural communities
are represented in present-day conservation lands (Scott
& Jennings 1997). Those species and communities not ad­
equately represented in the existing network of conserva­
tion lands constitute conservation “gaps.” The methods
of gap analysis were originally developed for application
to vertebrate species and land-cover types (Scott & Jen­
nings 1998), but they are relevant to a wide range of taxa
and hierarchies of biodiversity. We applied them to de­
termine whether the genetic variation within species is
adequately represented. Gap analysis involves intersect­
ing digital maps displaying protected areas with those
showing species occurrences and, in this case, patterns
of genetic variation.
Methods
Study Area
The study area included a wide region extending from
the coast of Oregon and Washington through the eastern
slopes and foothills of the Cascades (Fig. 1). Douglas-fir
occurs throughout much of the study area. Noble fir is
found from the Cascades of northern Washington to the
McKenzie River Valley in Oregon and at high peaks in
the Coast Range and Willapa Hills (Fig. 2a). South of the
McKenzie River, noble fir overlaps the range and intro­
gresses with Shasta red fir (Abies magnifica shastensis
Lemm.) (Sorenson et al. 1990).
Protected Areas
We projected all GIS coverages and grids (Table 1) in Uni­
versal Transverse Mercator Projection, Zone 10. Analyses
were done with ARC/INFO and Arcview software (Envi­
ronmental Systems Research Institute 1999, 2000).
A protected-areas coverage was developed following
conventions employed by the National Gap Analysis Pro­
gram (GAP). This program assigns land to four status levels
(Scott et al. 1993). We considered all status 1 and 2 lands
protected. Management plans for status 1 lands call for
maintaining a natural state and allowing natural distur­
bance events to proceed; examples include wilderness
areas, national parks, and U.S. Forest Service (USFS) re­
search natural areas. Status 2 lands are generally managed
for natural values but may be used in ways that degrade
the quality of existing natural communities; examples
Conservation Biology
Volume 18, No. 2, April 2004
414
Genetic Gap Analysis for Forest Trees
Lipow et al.
The protected-areas coverage combined data from sev­
eral available coverages (Table 1; Fig. 3). In Oregon most
status 1 and 2 lands were identified on the land manage­
ment and stewardship coverage (Oregon GAP). Most pro­
tected areas in Washington were identified on the major
public lands coverage ( Washington Department of Nat­
ural Resources) or the natural areas coverage (Interior
Columbia Basin Ecosystem Management Project). We fol­
lowed the Oregon GAP land designations when assign­
ing reserve status to these lands. For example, because
the Oregon GAP designated wilderness areas as status 1,
we assigned them to status 1 in Washington. Preserves
of The Nature Conservancy and natural-area preserves
and natural-resource conservation areas of the Wash­
ington Department of Natural Resources were also in­
cluded as status 1 lands. Federal Late Succession Reserves
(FEMAT 1993) were included as status 2 lands.
Species Distributions
Figure 1. The study area boundary and underlying
data layers used to generate tree-distribution maps.
Codes are defined in Table 1. Data types are described
in the text.
include national wildlife refuges and most state parks. Sta­
tus 3 lands are subject to extractive uses such as timber
harvest but have legal mandates that generally prevent
permanent land-cover change from natural or seminat­
ural communities to anthropogenic cover types; exam­
ples include most “matrix” lands of the Bureau of Land
Management (BLM) and the USFS. Although we did not
consider status 3 lands protected, they often contain con­
siderable in situ genetic resources. Finally, status 4 lands
are managed for extensive human uses.
Conservation Biology
Volume 18, No. 2, April 2004
Developing high-resolution species-distribution maps was
essential to the gap analysis. Existing range maps (e.g., Lit­
tle 1971) were inadequate for our purposes because they
indicated only species boundaries, not whether a species
was present at a specific location within that range or the
frequency of a species at a location. We derived distribu­
tion maps from spatial data showing vegetation type or
plant association group. The quality and type of data var­
ied across the region. For each area, we chose the data
set of highest resolution (Fig. 1).
The tree-distribution maps in Washington and north­
western Oregon were based on the plant-association
group (PAG) submodel of the potential natural vegeta­
tion (PNV) model (Henderson 1998). Plant-association
groups have similar floristics and environment and are
suitable units for mapping. They are defined by the cli­
max composition of overstory and understory vegetation
and are recognized in field plots by the theoretical climax
stage, although they include all seral stages (Hall 1998).
Accordingly, the tree distributions we derived from the
PAG submodel represented those expected under climax
or subclimax conditions. This was advantageous for our
gap analysis because the maps predicted where species
could potentially exist, including areas where they were
no longer present because of human disturbance. Al­
though we would liked to have known the actual distri­
bution of species in protected areas, the necessary data
were not available. We assumed that potential and actual
distributions were equivalent in protected areas. This as­
sumption was probably valid for many protected areas
because Douglas-fir and noble fir are found in a variety of
seral stages, and protected areas presumably suffer from
minimal human disturbance.
The PNV model was an environmental gradient model
that depended on quantifiable aspects of temperature and
moisture regimes. Predictor variables included elevation,
Lipow et al.
Genetic Gap Analysis for Forest Trees
415
Figure 2. Noble fir (a) density and distribution derived from modeled potential natural vegetation, ( b) gap
analysis results based on the seed-zone stratification, and (c) gap analysis results based on the ecoregion
stratification. In (c), only ecoregion polygons containing noble fir are displayed, and all protection class 2 and 3
strata are labeled. Protection classes are defined in Table 2.
precipitation, mean annual temperature, temperature
lapse rate, fog effect, aspect, site moisture, cold-air
drainage effects, and a logistical stratification (PNV eco­
zones). We used data from ecology plots on federal lands
and various other plots (Henderson 1997) to develop and
calibrate the model. At a minimum, the plot data indicated
geographic location, PAG, and a set of environmental fac­
tors. The nine PAG grids we used contained data from
>6600 plots in Oregon and >10,000 plots in Washing­
ton.
The PAG submodel generated a GIS grid in which each
pixel represented 30 × 30 m2 . We resampled to 90 ×
90 m2 to reduce processing time and because our other
data were coarser than 30 × 30 m2 . We derived tree dis­
tributions from the PAG output through a set of tables
to predict the density of each species in each PAG by
PNV-ecozone combination. The tables showed the ex­
pected density of mature individuals of each species for
each combination. We modeled four density levels: high
(>100 stems/ha), medium 10–100 stems/ha), low (<10
stems/ha), and absent.
Data for the plant-association group submodel were not
available for the entire study area. For the Deschutes and
Winema National Forests, we derived tree distributions
from GIS coverages of climax plant associations based
on aerial photography and field reconnaissance (1:24,000
scale). Again, we generated tables showing four densities
of trees and joined them to the plant-association cover­
ages to create distribution maps.
For southern Oregon and the remaining areas with
no PAG data, we used a GIS layer produced by Oregon
GAP using Landsat Thematic Mapper digital data and con­
ventional remote-sensing analysis and classification tech­
niques (Kilsgaard 1999). This layer showed generalized
land-cover types representing actual vegetation and had
lower accuracy and precision than the spatial data for
Conservation Biology
Volume 18, No. 2, April 2004
Genetic Gap Analysis for Forest Trees
416
Lipow et al.
Table 1. Coverages and grids used in the gap analysis of the genetic resources conserved in situ in protected areas.
Source
Protected areas
Interior Columbia Basin Ecosystem Management
Project
Oregon GAP Analysis Program
The Nature Conservancy of Washington
The Wilderness Society
Washington Department of Natural Resources
Washington Natural Heritage Program
Tree distributions
J. Henderson, Mt. Baker–Snoqualmie National
Forest
J. Henderson, Mt. Baker–Snoqualmie National
Forest
J. Henderson, Mt. Baker–Snoqualmie National
Forest
J. Henderson, Mt. Baker–Snoqualmie National
Forest
J. Henderson, Mt. Baker–Snoqualmie National
Forest
C. McCain, Siuslaw National Forest
C. McCain, Siuslaw National Forest
C. McCain, Siuslaw National Forest
C. McCain, Siuslaw National Forest
Oregon Natural Heritage Program
Deschutes National Forest
Fremont National Forest
Winema National Forest
Corvallis Forestry Sciences Laboratory,
Laboratory for Applications of Remote Sensing
in Ecology
Genetic and ecological stratifications
Washington Department of Natural Resources
Oregon Department of Forestry
Environmental Protection Agency
Deschutes National Forest
Winema National Forest
∗ NM,
Coverage or grid (code)
Scale or pixel size∗
natural areas
1:24,0001:500,000
land management and land stewardship
Nature Conservancy preserves
Option 9 lands (late-successional reserves)
major public lands (1997 version)
Washington Department of Natural Resources
protected areas
1:100,000
NM
NM
1:100,000
1:24,000
Olympic National Forest plant association groups
( January 2000 version) (OLY)
Okanogan National Forest plant association groups
( January 2000 version) (OKA)
Mt. Baker–Snoqualmie National Forest plant
association groups ( January 2000 version) (MBS)
Gifford Pinchot National Forest plant association
groups ( January 2000 version) (GIP)
Wenatchee National Forest plant association groups
( January 2000 version) (WEN)
Siuslaw National Forest plant association groups
(August 2000 version) (SIU)
Willamette National Forest plant association groups
(August 2000 version) (WIL)
Mt. Hood National Forest plant association groups
(August 2000 version) (MTH)
Willamette Valley plant association groups (August
2000 version) (VAL)
land cover for Oregon: OR-GAP Version 2 (SOR)
plant association of the Deschutes National Forest
(DES)
ecoclass mapping of the Fremont National Forest
(FRE)
vegetation plant community of the Winema
National Forest (WIN)
Pacific Northwest Ecosystem Research Consortium:
Willamette River Basin land use/land cover map
v.3, Oetter et al. 2001
30 × 30 m2
Washington seed zones, Randall & Berrang 2002
Oregon seed zones, Randall 1996
ecoregions (level IV), Pater et al. 1998
Deschutes National Forest breeding zones
Winema National Forest breeding zones
NM
NM
1:250,000
NM
NM
30 × 30 m2
30 × 30 m2
30 × 30 m2
30 × 30 m2
30 × 30 m2
30 × 30 m2
30 × 30 m2
30 × 30 m2
1:100,000
1:24,000
1:24,000
1:24,000
25 × 25 m2
no metadata provided by the source agency.
other areas. For analysis, we converted the original poly­
gon coverage to a grid of 25 × 25 m2 resolution before
resampling to 90 × 90 m2 .
Finally, we identified low-elevation (<120 m), forested
portions of the Willamette Valley on a land-cover grid pro­
duced by the Pacific Northwest Ecosystem Research Con­
sortium (Oetter et al. 2001). Most low-elevation land in
the Willamette Valley has been permanently converted to
agriculture or urban development. The land-cover grid,
which showed actual vegetation, was therefore more
likely to reflect true tree distributions in Willamette Val­
ley protected areas than the respective PAG grid, which
showed potential distributions. The land-cover grid was
Conservation Biology
Volume 18, No. 2, April 2004
based on Landsat thematic mapper data and showed
conifer cover and age class. Again, we generated tables
showing four densities of trees and joined them to the
plant-association coverages to create distributions maps.
Again, we joined a table of tree abundances within cover
types to the grid to produce species distributions.
Accuracy Assessment of Species Distributions
To estimate the accuracy of the resulting species-distri­
bution maps, we analyzed data on species occurrences
from an independent data set produced by the USFS’s Pa­
cific Northwest Region Current Vegetation Survey (CVS)
Lipow et al.
Genetic Gap Analysis for Forest Trees
417
tribution map was based on the Oregon GAP land-cover
coverage because it showed actual vegetation. Only plots
in national forests or BLM districts where a given species
was known to occur were evaluated. For instance, the
analysis for noble fir included plots in the Gifford-Pinchot
National Forest but not the Olympic National Forest be­
cause the latter is outside the species’ range (Little 1971).
To estimate the accuracy of the species-distribution
maps, we compared pixels on the distribution maps to
the corresponding CVS plots. We compared the number
of pixels where a species presence was predicted at ei­
ther high or medium density with the number of CVS
plots in which a species was detected. The number of
pixels where a species absence was predicted was also
compared to the number of CVS plots in which a species
was not detected. We could not adequately assess the ac­
curacy of the low-density level with the CVS plot data be­
cause of limitations of the survey-plot sampling design.
Patterns of Genetic Variation
Figure 3. Location of protected areas in western
Oregon and Washington. Status 1 and 2 codes are as
defined by the National Gap Analysis Program (Scott
et al. 1993).
( Johnson 1998). Survey plots were distributed in a grid
and were spaced 2.7 km apart on most federal lands, ex­
cept in wilderness areas, where they were spaced 5.5 km
apart. Species was recorded for all trees >1.219 m di­
ameter at breast height (dbh) in a 1-ha plot and for trees
0.330–1.218 m dbh in five 0.076-ha subplots, 0.076–0.329
m dbh in five 0.017-ha subplots, and 0.025–0.124 m dbh
in five 0.004-ha subplots. The data set we used showed
species presence in all CVS plots but did not indicate
subplot.
In areas where the distribution maps were based on the
PNV model, the analysis was limited to plots in protected
areas, where we assumed that actual and potential dis­
tributions were equivalent. Because of limited protected
areas in the Deschutes and Winema national forests, we
examined all CVS plots there regardless of protected sta­
tus. All CVS plots were analyzed for regions where the dis-
We applied two systems to stratify each species into pop­
ulations for analysis: (1) seed and breeding zones and
(2) ecoregions. The species-specific seed zones were de­
signed to inform land managers about risks associated
with moving seed from a source environment to another
location during reforestation (Randall 1996; Randall &
Berrang 2002). Their sizes and boundaries reflect infor­
mation obtained from common garden, genecological,
isozyme, and seed-movement studies indicating the ex­
tent and geographic patterns of genetic variation in the
studied populations. Species like Douglas-fir with a high
proportion of among-population genetic diversity have
small seed zones subdivided into relatively narrow eleva­
tional bands (Fig. 4b). Noble fir has less population-level
genetic variation and, consequently, larger seed zones
with wider elevational bands (Fig. 2b).
Seed-zone polygons were divided into the defined el­
evation bands (Randall 1996; Randall & Berrang’s 2002)
based on a 90-m digital elevation model. Randall’s (1996)
seed zones for Douglas-fir in Oregon do not extend east
of the Cascade crest. For this area, we substituted breed­
ing zones and elevation bands delineated by the USFS
(Fig. 4b).
The ecoregion stratification was developed by a group
of government agencies, including the Environmental
Protection Agency and the U.S. Geological Survey, with
the intent of providing a spatial framework for environ­
mental resource management (Omernik 1995; Pater et
al. 1998). Ecoregions denote areas of general similarity
in ecosystems and were based on environmental factors
and vegetation. We used the most detailed level IV ecore­
gions (see Gallant et al. 1989). The study region contained
55 level IV ecoregions represented by 335 polygons (i.e.,
they were not continuous). Application of this stratifica­
tion to genetic analysis required assuming correlation of
Conservation Biology
Volume 18, No. 2, April 2004
418
Genetic Gap Analysis for Forest Trees
Lipow et al.
Figure 4. Douglas-fir (a) density and distribution derived from data layers showing potential natural vegetation
or actual vegetation (location of a disjunct population on the Fremont National Forest is indicated with a star).
(b) Gap analysis results based on seed- and breeding-zone stratifications. Each stratum is numbered and assigned
a three-digit code in which the first value indicates the number of elevation bands assigned to protection class 1,
the second the number assigned to protection class 2, and the third the number assigned to protection class 3. Seed
and breeding-zone-by-elevation-band polygons assigned to protection classes 2 and 3 are shaded. In some seed
zones, protection class 2 and 3 elevation bands include only a small fraction of the total area. (c) Gap analysis
results based on the ecoregion stratification. All class 2 and 3 ecoregions are shaded and labeled.
adaptive and environmental variation across a species’
range. We expect the extent of this correlation, and the
accuracy of the assumption, to vary among species.
Gap Analyses
For the seed- and breeding-zone gap analyses, we over­
layed the protected-areas coverage on each seed or
breeding-zone-by-elevation coverage. We then tabulated
the area of each density class in each elevation band of
each seed or breeding zone and the area of each density
class in protected areas in each elevation band of each
seed or breeding zone. For the ecoregion analyses, we
overlayed the protected-areas coverage on the ecoregion
Conservation Biology
Volume 18, No. 2, April 2004
coverage and tabulated the area in each density class in
each ecoregion and in protected areas in each ecoregion.
Next, we assigned species in strata to one of three
classes of protection defined by the minimum expected
population sizes of species in whole strata and in status 1
and 2 protected areas (Table 2).We estimated minimum
expected
population sizes as: (ha at high
density ×
100) + (ha at medium density × 10) + (ha at low
density).
Protection class 1 was defined as well protected in sta­
tus 1 protected areas. For strata where species were com­
mon (minimum expected population size >25,000), class
1 required at least 5000 individuals to be in status 1 pro­
tected areas. For strata where species were uncommon
(minimum expected population size of 5000–25,000),
Lipow et al.
Genetic Gap Analysis for Forest Trees
419
Table 2. Definitions of the classes of protection assigned to strata in the gap analysis of conserved genetic resources for Douglas-fir and noble fir.
Minimum expected population size∗
Class of
protection
Species
occurrence
Description of class
entire stratum
1
1
common
not common
well protected in status 1 areas alone
well protected in status 1 areas alone
>25,000
5,000–25,000
2
common
>25,000
2
not common
3
3
common
not common
well protected in status 1and 2 areas
combined
well protected in status 1and 2 areas
combined
not well protected
not well protected
5,000–25,000
>25,000
5,000–25,000
status 1
protected areas
status 1 and 2
protected areas
>5,000
>10% of entire
stratum
<5,000
>5,000
<10% of entire
stratum
>10% of entire
stratum
<5,000
<10% of entire
stratum
∗ Minimum
expected population size calculated as Σ(ha at high density × 100) + Σ(ha at mid density × 10) + Σ(ha at low density). Status 1
lands have management plans that call for maintaining a natural state and allowing natural disturbance events to proceed (Scott et al. 1993).
Status 2 lands are generally managed for natural values but may be used in a way that degrades the quality of existing natural values (Scott et
al. 1993).
class 1 required the number of individuals in protected
areas to exceed 10% of the number in the whole stratum.
Protection class 2 was defined as well protected, consider­
ing both status 1 and 2 protected areas. A stratum ranked
in this class if, when the species was common, at least
5000 individuals were in status 1 and 2 protected areas
combined, with less than 5000 individuals in status 1 pro­
tected areas alone. For strata where a species was uncom­
mon, at least 10% of all individuals in a stratum must have
been in status 1 and 2 protected areas. The cutoff popu­
lation size of 5000 is in line with recommendations from
research addressing the desirable population size for the
purposes of gene conservation (Lande 1995; Lynch 1996;
Yanchuk & Lester 1996; Yanchuk 2001). Protection class
3 was defined as not well protected in either status 1 or 2
protected areas. We did not classify species in strata with
a minimum expected population size of <5000.
Results
Accuracy of Species-Distribution Maps
Considering only those national forests within the range
of noble fir, the distribution map (Fig. 2a) predicted
species presence on 309 CVS plots, and, of these, de­
tections occurred on 160 (51.8%) (Table 3). Of the 1130
plots for which the PNV model predicted no noble fir,
none were detected on 1055 (93.4%). Overall, Douglasfir was detected on 88.6% of plots for which it was pre­
dicted, with successful detection percentages equaling
83.4% for areas based on the PNV model, 71.3% on the De­
schutes and Winema National Forests, and 96.3% for areas
based on the Oregon GAP land-cover coverage (Table 3).
Of the plots where Douglas-fir absence was predicted,
validation percentages varied from 79.0% for PNV model
areas to 96.6% on the Deschutes and Winema National
Forests to 60.0% for Oregon GAP land-cover areas.
Gap Analyses
For noble fir, the gap analysis indicated adequate con­
servation of genetic resources in all three seed zone by
elevation band combinations. The analysis categorized
the species as well protected in status 1 protected ar­
eas in both the high- and low-elevation Cascades zones
(Table 4; Fig. 2b). The Coast Range zone was categorized
as well protected in status 1 and 2 protected areas, al­
though our field observations indicated that the actual
number of trees in status 1 protected areas exceeded
5000.
Of the 10 ecoregions containing noble fir, only the
Willapa Hills ecoregion was classified as not well pro­
tected (class 3). This ecoregion occurs in the Coast Range
of northern Oregon and southwestern Washington. The
Table 3. Number of current vegetation survey plots on which a species
was predicted to be present or absent at the corresponding location
on the distribution map and, of these, the number of plots validated.
Prediction
Number
predicted
Number
validated
Ratio
Noble fir: potential natural vegetation model
present
309
160
0.518
absent
1130
1055
0.934
Douglas-fir: potential natural vegetation model
present
1564
1305
0.834
absent
433
343
0.790
Douglas-fir: Deschutes and Winema national forests
present
108
77
0.713
absent
645
623
0.966
Douglas-fir: Oregon gap land cover
present
1272
1225
0.963
absent
25
15
0.600
Conservation Biology
Volume 18, No. 2, April 2004
Genetic Gap Analysis for Forest Trees
420
Lipow et al.
Table 4. Protection class and minimum expected population size in whole strata and in status 1 and 2 protected areas for the gap analysis of noble
fir based on seed zone and ecoregion stratifications.
Minimum expected population size∗
Stratum
Seed zone-elevation (code)
Coast Range, all (1)
Cascade Range, low (2)
Cascade Range, high (2)
Ecoregions (code)
Coastal volcanics (1d)
Willapa hills (1f )
Western Cascades lowlands (4a)
Western Cascades montane highlands (4b)
Cascade crest montane forest (4c)
Cascades subalpine/alpine (4d)
North Cascades highland forests (77b)
Grand fir mixed forest (9b)
∗ See
stratum
status 1
status 1
and 2
Protection
class
32,922
2,995,788
5,506,527
3,483
1,399,964
783,264
22,397
2,394,652
2,263,462
2
1
1
69,317
27,811
1,080,050
3,774,190
2,651,963
409,477
30,214
22,617
3,605
0
62,239
669,968
1,094,981
246,758
5,263
6,336
24,875
2,859
295,400
1,975,290
1,771,308
383,244
7,768
13,628
2
3
1
1
1
1
1
1
Table 2 for explanation.
only protected area in it with noble fir is Oregon’s Sad­
dle Mountain State Park (status 2), where several hundred
scattered individuals grow throughout the park in stands
dominated by Douglas-fir and Sitka spruce (S.R.L., per­
sonal observation).
Douglas-fir genetic resources were adequately con­
served throughout most of the species’ regional range
(Fig. 4b). Of the 204 seed-zone-by-elevation bands in
Washington and west of the Cascade Crest in Oregon,
198 (97.1%) were well protected in status 1 or 2 pro­
tected areas (classes 1 and 2). Of the 18 breeding-zone­
by-elevation bands defined for the Deschutes and Winema
national forests east of the Cascade Crest in Oregon, 14
(77.8%) were well protected in status 1 or 2 protected ar­
eas. In the analysis using ecoregions, 52 of the 54 (96.3%)
ecoregions containing Douglas-fir were well protected in
status 1 or 2 protected areas. Thus, a total of 12 Douglas-fir
strata were identified as putative gene-conservation gaps
(class 3) (Table 5).
Three putative gene-conservation gaps (two seed­
zone–by-elevation bands and a partially overlapping
ecoregion) occurred in the southern Puget lowlands.
These elevation bands were comprised of scattered
higher-elevation sites in the otherwise low-elevation area.
The southern Puget lowland consists primarily of pri­
vately owned and agricultural or developed land and
has few protected areas. Four putative gene-resource
gaps (three breeding-zone–by-elevation bands, all below
1829 m, and a corresponding ecoregion) were found on
the Chiloquin Ranger District of the Winema National For­
est, where Douglas-fir occupies one north-south running
ridge system with little designated protected area. The
lowest elevation band (<1524 m) in the Fort Rock breed­
ing zone of the Deschutes National Forest was identified
as a putative gap. In the north Oregon Coast Range, the
highest elevation band received class 3 designation. In
Conservation Biology
Volume 18, No. 2, April 2004
the next-lower elevation band, large Douglas-fir stands are
present in late-successional reserves and in four widely
spaced status 1 protected areas. The categorization of
the remaining class 3 seed-zone-by-elevations bands for
Douglas-fir was based largely on the arbitrary delineation
Table 5. For strata belonging to protection class 3 in the Douglas-fir
gap analyses,a the minimum expected population size in each stratum
as a whole and in status 1 and 2 protected areas.
Minimum expected
population sizeb
Location of class-3 stratum
stratum
Seed-zone-by-elevation-band analysis
OR5 by 915–1066 m
86,297
OR11 by 0–305 m
122,075
OR15 by 610–762 m
123,412
OR16 by 306–457 m
738,317
WA6 by 306–610 m
7,790,915
WA6 by 611–914 m
643,470
Breeding-zone-by-elevation-band analysis
Deschutes-Fort Rock
36,933
by <1524 m
Winema-Chiloquin
43,460
by <1524 m
Winema-Chiloquin
51,473
by 1525–1676 m
Winema-Chiloquin
32,040
by 1677–1829 m
Ecoregion analysis
2h-Cowlitz/Chehalis
5,540,481
foothills
9h-Fremont Pine/
41,755
Fir forest
a Gap
status 1
status 1
and 2
0
1458
0
510
0
0
243
4,212
0
585
0
0
820
820
43
2,104
0
3,979
0
4,101
0
2,568
40
1,384
analyses were done using seed-zone-by-elevation bands for
Washington and west of the Cascade crest in Oregon,
breeding-zone-by-elevation bands for the Deschutes and Winema
National Forest, and ecoregions for the entire study region.
b See Table 2.
Lipow et al.
of seed-zone boundaries. These elevation bands included
limited areas toward the margin of a seed zone, with many
individuals at the same elevation effectively conserved in
an adjacent seed zone.
Thirty-three strata were identified as having Douglas-fir
well protected in status 1 and 2 protected areas com­
bined, but not in status 1 protected areas alone (class
2). Fourteen of these represented either the highest or
lowest elevation sites in a seed zone, included limited
area, or lay along a zone border where the same eleva­
tional band in the adjacent zone was assigned to class
1. Two class 2 seed-zone-by-elevation bands contained a
protected area that was bisected by the seed-zone bor­
der. Had the zone boundaries included the entire pro­
tected area, these strata would have ranked in class 1.
Land designated as late-successional reserve comprised
the majority of the protected area in seven other class
2 combinations of seed-zone-by-elevation bands. For five
class 2 ecoregions, the seed zone with the most similar
geography ranked in class 1. Four other class 2 ecoregions
overlapped with seed zones that were variously assigned
to classes 1–3.
Discussion
The gap analysis proved an effective method for evalu­
ating genetic resource conservation for the forest trees
we studied. Genetic resources for Douglas-fir and noble
fir were well protected in situ throughout much of the
study area. This conclusion holds regardless of whether
seed zones or ecoregions were used to stratify the distri­
bution of species into units of genetic conservation. By
well protected we mean that at least 5000 reproductiveage individuals were growing on sites in each stratum
that were unlikely to be harvested during the next sev­
eral decades. In most strata, the number of protected trees
was much greater. Additionally, the majority of strata have
large populations in status 1 protected areas so that, in
the unlikely event of future unsustainable logging in latesuccessional reserves or other status 2 protected areas,
the genetic resources would remain protected.
Our gap analysis includes as conserved only those trees
growing “naturally” in status 1 and 2 protected areas.
Many additional genetic resources are conserved in situ
through application of current forest-management prac­
tices, including natural regeneration and the placement
of streamside and landscape buffers throughout the man­
aged landscape. Reforestation with wild stand seedlots
and seed-orchard seed of appropriate population size and
from the local area also ensures the maintenance of ge­
netic diversity. Additionally, there are extensive regional
genetic resources in ex situ forms, including progeny
tests, seed orchards, and seed stores.
The Willapa Hills of southwestern Washington is the
one area where we advise further consideration of the
Genetic Gap Analysis for Forest Trees
421
adequacy of conservation of genetic resources for noble
fir. This area has the northernmost, coastal populations of
the species. Here, noble fir is found as scattered individ­
uals and in small stands. The closest protected noble fir
occurs at high elevations in Saddle Mountain State Park in
northern Oregon. Willapa Hills populations have the po­
tential to have diverged genetically due to the effects of
isolation, genetic drift, and natural selection. The only ex
situ genetic resources for these populations are found in
operational seed stores held by the Weyerhaeuser Com­
pany, the Washington Department of Natural Resources,
and The Campbell Group. These are not presently man­
aged as gene archives.
Increased conservation of Willapa Hills noble fir
through either ex situ or in situ measures offers several po­
tential advantages. In several European countries where
noble fir is grown commercially for boughs, the Willapa
Hills is a preferred provenance (U. Nielson, personal com­
munication). Tree breeders in Germany have established
a clone bank in response to their concerns over in situ
conservation of Willapa Hills populations, but it has only a
few selections (Ruetz et al. 1990; W. Ruetz, personal com­
munication). As the bough industry grows in Oregon and
Washington, additional Willapa Hills selections may yield
economic benefits. Moreover, geographically peripheral
and isolated populations deserve high conservation prior­
ity because they may be important to species migration,
especially in response to global climate change (Lesica &
Allendorf 1995).
Our gap analysis revealed a possible in situ genetic re­
source gap for Douglas-fir in the southern Puget lowlands,
especially at the highest elevations. This area is heavily
forested, and Douglas-fir is ubiquitous throughout. Refor­
estation with local seed is especially important to gene
conservation here, given the paucity of protected land.
Hundreds of selections from the southern Puget lowlands
are located in genetic tests and comprise a valuable ex
situ genetic resource (S.R.L., K.V.-B., J.B.S., J.A.H., & C.M.,
unpublished data).
Other apparent genetic-resource gaps for Douglas-fir re­
vealed by our analysis occur in the Fort Rock (Deschutes
National Forest) and Chiloquin ( Winema National For­
est) breeding zones. In the Fort Rock breeding zone,
most Douglas-fir occupies sites on or near lava flows
that are unlikely to be harvested (R. Evans, personal
communication). In the Chiloquin breeding zone, an es­
timated 60% or more of the natural Douglas-fir stands
have not been logged previously, and future harvesting
is unlikely because mistletoe has reduced their economic
value (S. Puddy, personal communication). Wildfire poses
the biggest potential threat to the Chiloquin populations,
because several large fires could eliminate them. The USFS
holds seed stores for the Chiloquin zone, however, that
could be used for reforestation.
A disjunct population of Douglas-fir occurring on the
Fremont National Forest in south-central Oregon was not
Conservation Biology
Volume 18, No. 2, April 2004
422
Genetic Gap Analysis for Forest Trees
included in the gap analysis. It is in an area known as
the Punchbowl, approximately 110 km east of the next
closest known stand. This population consists of approx­
imately 2000 individuals of all ages scattered across an es­
timated 50 ha. The Fremont National Forest is committed
to protecting this presumably unique genetic resource
in situ (D. Stubbs, personal communication) and stores
seed from 29 parent trees. The population is potentially
threatened by wildfires and competition from true firs.
The accuracy and resolution of the tree-distribution
maps largely determine the robustness of our gap anal­
ysis. The data necessary to generate high-resolution treedistribution maps is not yet available for many forested
regions. Our study area, however, contains a high pro­
portion of public land and has been subject to extensive
mapping of vegetation and landscape features. We expect
the distribution maps to be most accurate in protected ar­
eas, at least those on federal lands, because this is where
the ecology plots used to calibrate the PNV-model are
concentrated. We suspect that many of the map errors
that do occur stem from problems with the underlying
plant-association group and vegetation layers rather than
from miscoding in the associated tree-distribution tables,
because Douglas-fir and noble fir are indicator species
for many plant-association groups and vegetation types
(Hall 1998; Kilsgaard 1999). The accuracy of the treedistribution maps was lowest in southwestern Oregon,
where they were based on the relatively poor-resolution
Oregon GAP coverage.
For most of the study area, the distribution maps re­
flected expected tree distributions under theoretical cli­
max conditions. We assumed that the actual distributions
were equivalent to these theoretical climax distributions
in protected areas. Because of natural disturbance and
human activities, this assumption was surely violated in
some places, adding a source of error to our analysis. In
particular, late-successional reserves, designated status 2
and a large component of the total protected area, varied
considerably in the amount of late-successional and oldgrowth forest they contained. In some, past management
practices involved planting of Douglas-fir or noble fir.
The usefulness of the genetic gap analysis is also in­
fluenced by the stratification used to subdivide species
distributions into populations for conservation. An ideal
stratification would be based on a thorough assessment of
a species’ genetic variability and structure, with emphasis
given to variation in adaptive traits (Libby & Critchfield
1987; Eriksson 1995). Because genetic knowledge was
incomplete for the study species, we conservatively em­
ployed two independent stratifications, seed and breed­
ing zones and ecoregions, to increase the likelihood of
detecting all genetic-resource gaps.
When developing seed zones, Randall (1996) and
Randall and Berrang (2002) incorporated much of the
accumulated genetic information about each species. For
noble fir, however, this was limited to one 3-year nursery
Conservation Biology
Volume 18, No. 2, April 2004
Lipow et al.
study (Sorenson et al. 1990) plus a few provenance tests
done in other countries (Randall & Berrang 2002). Hence,
the noble fir seed zones might best be considered first ap­
proximations. More information, including genecological
and allozyme studies, was available to direct delineation
of Douglas-fir seed zones.
The gap analysis done with ecoregions is intended to
complement and add to the one done with seed zones.
Ecoregions are widely used in Oregon and Washington
for demarcating areas of similar environmental and eco­
logical characteristics (Omernik 1995). They can serve as
a surrogate for data on genetic structure if genetic struc­
ture is well correlated with environmental and ecological
characteristics. Another reason for running the gap anal­
ysis with ecoregions is that various U.S. federal and state
agencies use them in conservation assessments. Their
use should therefore facilitate the integration of these re­
sults with those from analyses of other ecosystem com­
ponents.
Acknowledgments
Organizations that contributed to the Pacific North­
west Forest Tree Gene Conservation Group are Boise
Cascade, Olympic Resource Management, the Oregon
Department of Forestry, Oregon State University, The Tim­
ber Company, the U.S. Forest Service (Region 6, Pacific
Northwest Research Station, and State and Private), the
U.S. Bureau of Land Management, the Washington De­
partment of Natural Resources, Weyerhaeuser Company,
and Willamette Industries. We thank C. Chappell of the
Washington Natural Heritage Program for reviewing the
tree-distribution maps and J. Ohmann of the Pacific North­
west Research Station for providing compiled plot data
from the Northwest Region Current Vegetation Survey.
L. Riggs, W. Libby, and R. Burdon initially developed the
gap analysis concept for conserved genetic resources in
forest trees. We also thank P. Berrang, C. Dean, R. Evans,
J. Hamlin, and N. Mandel for helpful comments on the
manuscript.
Literature Cited
Environmental Systems Research Institute (ESRI). 1999. ARC/INFO
user’s manual. ESRI, Redlands, California.
Environmental Systems Research Institute (ESRI). 2000. ArcView Spa­
tial Analyst white paper. ESRI, Redlands, California. Also available
from http://www.esri.com/library/whitepapers/pdfs/avspanal.pdf
(accessed July 2003).
Eriksson, G. 1995. Which traits should be used to guide sampling for
gene resources? Pages 349–358 in P. Baradat, W. T. Adams, and G.
Muller-Starck, editors. Population genetics and genetic conservation
of forest trees. SPB Academic Publishing, Amsterdam.
Forest Ecosystem Management Assessment Team (FEMAT). 1993. For­
est ecosystem management: an ecological, economic, and social as­
sessment. U.S. Forest Service and U.S. Bureau of Land Management,
Washington, D.C.
Lipow et al.
Gallant, A. L., T. R. Whittier, D. P. Larson, J. M. Omernik, and R. M.
Hughes. 1989. Regionalization as a tool for managing environmental
resources. EPA/600/3089/060. U.S. Environmental Protection
Agency, Environmental Research Laboratory, Corvallis, Oregon.
Hall, F. C. 1998. Pacific northwest ecoclass codes for seral and poten­
tial natural communities. General technical report PNW-GTR-418.
U.S. Forest Service, Pacific Northwest Research Station, Portland,
Oregon.
Henderson, J. A. 1997. The PNV model. Mt. Baker/Snoqualmie National
Forests, Mountlake Terrace, Washington.
Henderson, J. A. 1998. The USFS Potential Natural Vegetation Mapping
Project. Mt. Baker/Snoqualmie National Forests, Mountlake Terrace,
Washington.
Johnson, M. D. 1998. Region 6 inventory and monitoring system:
field procedures for the current vegetation survey, version 2.03.
Natural resource inventory. U.S. Forest Service, Pacific Northwest
Region, Portland, Oregon. Available from http://www.fs.fed.us/r6/
survey/document.htm (accessed October 2002).
Kilsgaard, C. 1999. Land cover type descriptions: Oregon gap analy­
sis 1998 land cover for Oregon. Oregon Natural Heritage Program,
Portland.
Lande, R. 1995. Mutation and conservation. Conservation Biology
9:782–791.
Lesica, P., and F. W. Allendorf. 1995. When are peripheral populations
valuable for conservation? Conservation Biology. 9:753–760.
Libby, W. J., and W. B. Critchfield. 1987. Patterns of genetic architecture.
Annales Forestales (Zagreb) 13:77–92.
Lipow, S. R., J. B. St. Clair, and G. R. Johnson. 2001. Ex situ gene conser­
vation for conifers in the Pacific Northwest. General technical report
PNW-GTR-528. U.S. Forest Service, Pacific Northwest Research Sta­
tion, Portland, Oregon.
Little, E. L. Jr. 1971. Atlas of United States trees. 1. Conifers and impor­
tant hardwoods. Miscellaneous publication 1146. U.S. Department
of Agriculture, Washington, D.C.
Lynch, M. 1996. A quantitative-genetic perspective on conservation is­
sues. Pages 471–505 in J. C. Avise and J. L. Hamrick, editors. Conser­
vation genetics: case histories from nature. Chapman and Hall, New
York.
Oetter, D. R., W. B. Cohen, M. Berterretche, T. K. Maiersperger, and
Genetic Gap Analysis for Forest Trees
423
R. E. Kennedy. 2001. Land cover mapping in an agricultural setting
using multi-seasonal thematic mapper data. Remote Sensing of En­
vironment 76:139–155.
Omernik, J. M. 1995. Ecoregions: a spatial framework for environmental
management. Pages 49–62 in W. S. Davis and T. P. Simon, editors.
Biological assessment and criteria: tools for water resource planning
and decision making. Lewis Publishers, Boca Raton, Florida.
Pater, D. E., S. E. Bryce, T. D. Thorson, J. Kagan, C. Chappell, J. M.
Omernik, S. H. Azevedo, and A. J. Woods. 1998. Ecoregions of west­
ern Washington and Oregon. U.S. Geologic Survey, Reston, Virginia.
Randall, W. K., 1996. Forest tree seed zones for western Oregon. Oregon
Department of Forestry, Salem.
Randall, W. K. and P. Berrang. 2002. New Washington tree seed zones.
Washington Department of Natural Resources, Olympia.
Ruetz, W., R. Dimpflmeier, J. Kleinschmit, J. Svolba, H. Weisgerber, and
H. M. Rau. 1990. The IUFRO Abies procera provenance trial in the
Federal Republic of Germany: field results at age 9 and 10 years.
Pages 2.276–2.289 in Proceedings of the Joint International Union
of Forest Research Organizations (IUFRO) meeting. Weyerhaeuser
Company, Tacoma, Washington.
Scott, J. M., and M. D. Jennings. 1997. A description of the national gap
analysis program. Biological Resources Division, U.S. Geological Sur­
vey, Washington, D.C. Available from http://www.gap.uidaho.edu/
about/overview/gapdescription/default.htm (accessed May 2002).
Scott, J. M., and M. D. Jennings. 1998. Large-area mapping of biodiversity.
Annals of the Missouri Botanical Garden 85:34–47.
Scott, J. M., F. Davis, B. Csuti, R. Noss, B. Butterfield, C. Groves, H.
Anderson, S. Caicco, F. D’Erchia, T. C. Edwards Jr., J. Ulliman, and G.
Wright. 1993. Gap analysis: a geographic approach to protection of
biological diversity. Wildlife Monographs 123.
Sorensen, F. C., R. K. Campbell, and J. F. Franklin. 1990. Geographic vari­
ation in growth and phenology of seedlings of the Abies procera/A.
magnifica complex. Forest Ecology Management 36:205–232.
Yanchuk, A. D. 2001. A quantitative framework for breeding and conser­
vation of forest tree genetic resources in British Columbia. Canadian
Journal of Forest Research 31:566–576.
Yanchuk, A. D., and D. T. Lester. 1996. Setting priorities for conserva­
tion of the conifer genetic resources of British Columbia. Forestry
Chronicle 72:406–415.
Conservation Biology
Volume 18, No. 2, April 2004
Download