Figure 6
Figure 7
Table 3
|
RESULTS
The study area contained 7,545 vegetated stands
that support fire according to the HNFR classification. These stands
comprise 90% of the study area as follows:
Age
|
Area
ha
(number of stands)
|
Percentage
of Vegetated Study Area that Supports Fire
|
|
<
200 years (young)
|
124,130.7 (6,645)
|
78%
|
|
≥
200 years (old)
|
35,115.9
(900)
|
22%
|
|
All stands
|
159,246.6 (7,545)
|
100
%
|
A chi-squared test for independence between elevation (predictor variable) and age (response
variable) tests the
the null and associated hypothesis that follows:
H0: Stand
elevation and age are independent,
H1: Stand
elevation and age are dependent.
To test the hypothesis
that elevation and age are not associated, I constructed a two-way contingency
table that includes
values for observed (and expected) results:
|
|
0-1,500 m
|
≥ 1,500 m
|
Row total
|
|
Old
|
297
(193)
|
603
(707)
|
900
|
|
Young
|
1307
(1411)
|
5268
(5164)
|
6575
|
|
Column Total
|
1604
|
5871
|
7475
|
with a χ2 value of 80.9. This test
has 1 degree of freedom and a significance value of 10.83 using an alpha of
0.001.
Since χ2 of 80.9 >
10.83, I reject H0 that elevation and age are independent.
I ran a stepwise
forward logistic regression to assess the importance of elevation,
slope, aspect, solar radiation, and distance from
nearest road as
predictors of age. Elevation, slope, radiation, and distance
from road are all continuous variables and
were not transformed. I
transformed aspect to account for the influence of warm versus cool
aspects as follows:
When aspect>=1 and aspect<=135, aspect_lin = aspect+45+1,
if aspect>=315 and aspect<=360 then aspect_lin = aspect-315+1, and
if aspect<=314 and aspect>=136 then aspect_lin = 360-aspect-45+1.
The resulting values for aspect ranged from cool to warm, 1-181. Variables that best explain stand age are indicated
by p-values < alpha
(.05). The parameters elevation and aspect contributed significantly to
the model while slope, solar radiation,
and distance from road did not.
Increased elevation and warmed aspect caused a significant increase in stand
age (Table 2).
| Model |
Independent Variables |
Test statistics |
| Elevation |
Slope |
Aspect |
Radiation |
Distance from Road |
-2log |
AIC |
| Forward Stepwise |
<.0001
(0.998) |
- |
<.0001
(1.005) |
- |
- |
5388 |
5396 |
Table 2. Model of Factors Influencing Stand Age. Independent variables include
distance from road and four physical
attributes for stands that contain
vegetation and support fire. Model was constructed using a forward step-wise
procedure. Maximum likelihood estimates are presented as p-values and
odds-ratios (in parenthesizes). Model
assessment used -2log likelihood (-2log)
and Akaike’s Information Criterion (AIC).
Stand age differs with elevation and confirm
results of the chi-squared test (Figure 6). Sample sites fall well within range
of young stands that comprise the majority of landscape (Figure 7). Sample sites were located across a range of
elevations within each strata (Table 3).
|
| |
