were collected for later identification. A more complete
aerial photographs. We also developed a nomenclature
inventory was conducted concurrently by CRREL
for ecotypes that explicitly relates ecological charac-
(Duffy 1999).
teristics in a terminology that can be easily understood.
For the map verification sites, only vegetation struc-
Because ecosystems are highly complex and vari-
ture and dominant plant species were listed. In addi-
able, it was necessary to aggregate detailed character-
tion, a preliminary ecotype and Alaska Vegetation Clas-
istics described in the field (e.g., soil stratigraphy and
sification class was assigned in the field.
vegetation composition). For each component, we used
a hierarchical approach to aggregation (Fig. 4). For
Classification
geomorphology, we hierarchically aggregated clasts,
Ecosystem classification was approached at two lev-
textures, layers, and lithofacies into geomorphic units
els. First, individual ecosystem components were clas-
(architectural elements) using the approaches of Miall
sified and coded using standard classification systems
(1985) and Brown (1997). Geomorphic units were as-
developed for Alaska (Table 1). Second, these ecosys-
signed to physiographic settings based on their erosional
tem components were integrated to classify ecosystems
or depositional processes (see Appendix A). Surface-
at three spatial scales using a variety of differentiating
forms were simplified into a reduced set of slope ele-
criteria (Table 2).
ments (i.e., crest, upper slope, lower slope, toe, flat).
For vegetation, we used the structural levels of the
Ecosystem components
Alaska Vegetation Classification (Viereck et al. 1992)
Vegetation types initially were classified to Level
because they are more readily identifiable on aerial
IV of the Alaska Vegetation Classification (Viereck et
photography than is floristic composition.
al. 1992), from data collected at sample sites, based on
We identified common relationships among ecosys-
structural and floristic criteria. Geomorphic units were
tem components by looking at graphic profiles and us-
classified according to a system based on landform
ing contingency tables. The contingency tables succes-
soil characteristics for Alaska originally developed by
sively sorted plots by climate zone, physiography, tex-
Kreig and Reger (1982) and modified for this study.
ture, geomorphic unit, drainage, and vegetation type.
During classification of geomorphic units, we also re-
From these tables, common associations were identi-
lied on the geologic map of the Mt. Hayes Quadrangle
fied and unusual associations either were lumped with
(Pw and Holmes 1964), the terrain unit maps in Kreig
those with similar characteristics or excluded as un-
and Reger (1982), and the glacial maps and terrain de-
usual (outliers). Our philosophy was that it was better
scriptions in Pw and Reger (1983). Organic units were
to identify strong relationships that could be used for
those defined in the wetland classification for Canada
prediction and mapping than to create additional rules
(NWWG 1988). Surface-forms were classified accord-
and classes that only increase confusion and degrade
ing to the system developed by Washburn (1973) for
accuracy.
periglacial microtopography. Soils were classified ac-
Ecotype names were based on the simplified eco-
cording to Keys to Soil Taxonomy (Soil Survey Staff
system components. For example, the full name for an
1998).
ecotype for an individual plot would be Boreal Upland
Rocky Moist Mixed Forest, based on climatic, physi-
Ecosystems
ographic, textural, hydrologic (moisture), and vegeta-
Ecotypes (local ecosystems) descriptively integrate
tive components, respectively. Because this generated
the climate, physiography, soil texture, soil moisture,
a large number of specific ecotypes (113) from the 252
and vegetation type of a discrete area. We classified
field plots, we aggregated many similar types into a
ecotypes doing by the following:
reduced set of ecotypes (48). Some textural classes were
grouped (e.g., rocky and loamy) because the vegeta-
Simplying and aggregating detailed ground descrip-
tion classes were similar, or similar vegetation struc-
tions of ecological components.
tures (e.g., open and closed black spruce) were grouped
Identifying ecological relationships among terrain
because species composition was similar. This group-
features by developing graphic profiles of ecosys-
ing relied on identifying the most frequently occurring
tem components along toposequences.
components. Overall, we tried to balance both the need
Deriving a reduced set of ecotypes by identifying the
to differentiate ecological characteristics and the need
most common relationships and central tendencies.
to minimize the number of classes for management
purposes. This approach to classifying ecotypes pro-
In developing the ecotype classes, we also tried to use
vided a reduced set of broader groups, although the
ecological characteristics (primarily geomorphology
grouping can be done in any number of ways and other
and vegetation structure) that could be interpreted from
users may wish to group characteristics in different ways
7
Previous Page
