Modeling Post-Eruption Habitat Changes for Deer at Mount St. Helens using Remote Sensing and GIS - Geotechnologies and the Environment - page 18

Table 2.1 Percent landcover changes in summer and winter ranges within GMU 524 between
1984 and 2002

Summer 1984 1988 1991 1996 1999 2002

Non-Veg 36.49 13.92 7.28 6.69 3.78 2.03

Soil-Veg 1.52 18.39 11.02 9.23 11.16 10.34

Grass-Forb 17.39 15.56 22.95 12.81 11.16 9.47

Shrub-Seedling 20.88 14.03 14.24 13.66 8.89 10.39

Shrub-Seedling 0.00 14.89 18.91 7.19 10.41 10.07

CSP 22.48 22.66 23.64 49.00 53.34 56.42

Shadow 1.25 0.56 1.95 1.42 1.26 1.27

Winter 1984 1988 1991 1996 1999 2002

Non-Veg 12.62 3.37 1.40 1.31 0.86 0.79

Soil-Veg 0.35 6.46 2.73 1.37 6.59 3.47

Grass-Forb 21.49 16.49 13.54 3.77 2.23 3.97

Shrub-Seedling 30.76 16.12 16.02 7.86 3.93 3.42

Shrub-Seedling 0.00 22.71 28.76 7.30 7.10 6.23

CSP 33.42 34.19 37.09 77.88 78.83 81.63

Shadow 1.35 0.65 0.46 0.52 0.48 0.49

1991, and then declining to just under 7 deer/km2 by 2002 (Fig. 2.3 ). Within indi-
vidual years, roads and forage-cover ratios reduced NCC values by >50% ( P < 0.02).
NCC models not corrected for roads and/or forage-cover ratios fit actual population
changes in BTD in 1984 and 1988 better than did models corrected for road and/or
forage-cover ratio effects (Fig. 2.3 ). In contrast, between 1991 and 2002, NCC_Rd,
NCC_FC, and the combined models produced estimates more similar to actual deer
population trends (Fig. 2.3 ) although simple correlations show the NCC model most
associated with deer numbers across all years (Table 2.2 ).

2.4.2 Model Comparisons

NCC and modified estimates for each year along with bootstrap 90% confidence
limits are shown in Table 2.3 . Variation was high, particularly in 1988 and 1991,
with NCC estimates ranging from approximately 5-51 deer/km 2 . Variation was
somewhat reduced in other years although confidence limits were proportionally
high relative to NCC in all models (Table 2.3 ).
Among-year comparisons revealed no statistically significant differences
between 1984 and 1988, while values in 1991, when mean NCC peaked, were
marginally greater than in 1984 though not statistically significant ( P < 0.12)
(Table 2.4 ). A statistically significant decline in NCC occurred between 1991-1999
and 1991-2002 for all models ( P < 0.05), with the greatest difference seen in the
combined model (Table 2.4 ).