WSP00036

• File Number: 8200.750
• Description: San Juan River General
• State: CO
• Basin: San Juan/Dolores
• Water Division: 7
• Date: 3/1/1975
• Author: BOR
• Title: San Juan Ecology Project - Interim Progress Report - March 1975
• Water Supply Pro - Doc Type: Report/Study

Ridge; or (2) the species attained its highest fre-
quency values in stands of medium snow duration. The
reason for the second criterion 1s discussed below.

Species with high positive or negative loadings on a
component are, of course, highly correlated, positive-
ly or negatively, with that component. Stands with
loadings near,tero are either not correlated to that
component, or, are correlated to that component in some
nonlinear way~ not accounted for in this analysis.
The more nearly the loadings of two spe~ies correspond
to each other, the closer their ~ehavior is along the
gradient defined by that component. Species with
highly opposite loadings are negatively correlated
with respect to their response along the gradient of
the component; they are responding in an opposite way
to some common factor. For example, Lathvrus
1eucanthus is highly positively correlated with the
first component in Table 1, as is Vicia americana.
They are responding to the gradien~the first
component in the same way. On the other hand, Luzula
parvif10ra is highly negatively correlated with~
first component; it is responding in an opposite way
from the other two species.

The proportion of the total-variation accounted for by
each component in Table 1 is given at the extreme
bottom of the table (e.g., .3558). This proportion is
the variance or latent root of that component. The
sum of the variances for the first three components is
the proportion of the total variation in the data
accounted for by those components. This proportion is
given at the 'bottom of the "h211 column, and exceeds
.58 in value.

A unique property of PCA is that some sets of data may
be included in such a way t~t they do not appreciably
affect the results of the analysis. This is accom-
plished by scaling the desired data down several
ordets of magnitude so that it might vaty from 0 to
0.1, for example, while the rest of the data varies
from 0 to 100. In this study certain environmental
parameters were included in this way, 8S well as the
density of tree species. The presence of these data
do not appreciably effect the PCA and the statistical
correlation between the results and the environmental
parameters were obtained. The resultant correlation
coefficients'of the mOst important environmental
parameters with the components (their loadings of
those components) are shown near the bottom of Table
1, those of the major tree species are shown near the
top of Table 1.

It can be seen from Table 1 how tree,sapling, and
seedling densities correlate with the first 3 major
gradients in the vegetation. Note, again, that the
analysis is based on understory species frequencies
and that the tree data did not play a part in estab-
lishing the PCA display. Nevertheless, the inclusion
of these data allows the relation of tree species'
densities with major vegetational gradients to be
seen. Spruce densities are very negatively correlated
with the first major vegetational gradient, as is fir,
though to a lesser extent. Aspen is also fairly
strongly correlated with this gradient, though in an
opposite manner to that of spruce and fir.

Of particular interest to this study is snow duration.
Snow duration here is determined by the index

SDI-(P+2C)/PCmax

where P is the first date of partial snow clearance,
C is the first date of total snow clearance, and PCmax
is the largest value of (P+2C) observed in any stand.

A1~ computations were for 1971 snow data since this
is :the only year for which adequate photo coverage is
available. Snow Duration Indices are relative posi-
tions of stands along a snow duration gradient. The
absolute scale of snow duration varies from year to
year. Snow Duration Indices were also calculated for
the few stands covered in the 1973 photographs, and,
although there were very few usable photographs in
1973, the relative positions of the stands along the
snQw duration gradient were not changed from their
positions using only the 1971 data.

The Radiation Index values were obtained from tables
(Frank and Lee, 1966) and are the ratios of the
annual radiation totals to the annual maximum poten-
ti~l solar beam irradiation (the solar constant times
the duration of sunshine for the year). This is more
than a quantification of the aspect parameter; it is
an estimate of an important part of the energy budget
of a site and 1s an estimate of a parameter that has
been shown to be significantly correlated to soil
moisture depletion rates (Stearns and Carlson, 1960).
In.general, a perfectly flat stand would have a
Radiation IndeK of about 0.48 at the latitude of the
st~dy area. Stands with southerly aspects have
Radiation Indices greater than this while those with
nOFther1y aspects have Radiation Indices smaller than
this. The exact magnitude of the Radiation Index for
a giveri latitude depends on the aspect and steepness
of~ the slope.

I

I
II

I

II
I

I

I

The Drainage Class is related to topographic position
and the relative ability of a site to gain or lose
wa:ter from gravitational drainage and runoff. A low
Drainage Class value implies a well drained site that
has little or no recharge from runoff or subsurface
flow; a high Drainage Class value implies a site that
will accumulate runoff or seepage or both.

I

The basic assumption in the interpretation of the
gradients defined by the components is that the dif-
ferences in species frequency at various sites is a
response to differences in the environment in those
sites. If differences in species composition are due
to differences in the environment, and if the differ-
ences of some few environmental parameters account
for much of the differences in species composition,
then major gradients in species composition ought to
be related to gradients of these few environmental
parameters. There ought to be a high correlation
between those factors causing the greatest composi-
tional differences in the vegetstion and the major
g~adients within the vegetation. The PCA is a multi-
variate statistical procedure that defines the
gradients of maximum variation in the data. The
inclusion of environmental parameters in the analysis
allows correlation coefficients of those parameters
with the gradients defined by the analysis.

I

I

I

I

I

As shown in Table 1, there is a high negative cor-
relation (-.91) between snow duration and the
primary gradient in the vegetation (the first compo-
nent). This correlation is so high that the gradient
described by the first component is interpreted as a
snow duration gradient. Radiation Index and Percent
Slope are highly corre1sted with the second component.
As these two factors are not independent (percent
slope is used in the calculation of the Radiation
Index) this might be expected. Therefore the second
component is interpreted as being a gradient of
Radiation Index or simply radiation. The third
component is interpreted as an e1evations1 gradient.

I

I

I

13

I

I