489
1) if gopher
tionships
of herbaceous
etation. Three
fferences
of new
ovided the
(mounds
for mounds).
Commission errors occurred when other objects in the photographed
scene were misinterpreted as gopher mounds. The commission errors resulted
from similar confusions contributing to the omission errors; misinterpretation
of old gopher mounds, bare soil patches with apparent similar microrelief as
gopher mounds, and infrequently some low-growing shrubs.
Omission errors were high and commission errors were low for all
interpreters, but they were consistent. In fact, there were no significant
differences (P = .05) among interpreters for percent correct identification
of fresh gopher mounds, percent omission errors or percent commission errors.
Although the correlation coefficient between photo mound counts and ground
mound counts was marginally acceptable (r values between 0.7 and 0.8), the
ed to compare
tointerpreted
t.
lack of significant differences among interpreters provided credence that the
photo technique could be used to estimate pocket gopher populations. The high
degree of correlation (r = 0.9) among interpreters for percent mounds counted
as well as mounds omitted indicated the interpreters were consistently interpreting
the photographs in the same way.
ctives of
dence.
There were no significant differences (P = .05) in interpreter
results due to apparent vegetation differences among plot locations or between
treatments within locations. Obliterating the mounds before the photo mission
may have caused some interpreter confusion. The obliteration of mounds resulted
600 scale
the
o the soil
entifying
und-based
based on
in the interpretation of some of the disturbed areas as fresh mounds. There
were obvious examples in the photographs which may have caused the interpreters
to be excessively cautious.
Based on the fact that interpreter errors were similar and there
were negligible effects due to plot treatment and vegetation height, the actual
photointerpreted mound counts were adjusted to ground mound counts. This was
done for interpretation results for each interpreter on a plot-by-plot basis
ariously
old mounds,
relative
ric, although
imes in
d film
m bluish-green,
s were light
and old
texture,
image.
and the resulting ratio coefficient was applied to the Reid et al. (1966)
equation in the following form:
'y = .6582 / RM log (RM + 1)
/N
where: Y = gopher population estimate per unit area
R = ratio coefficient: ground mound counts
photointerpreted mound counts
M = photointerpreted mound counts per unit area
The mean gopher population estimates for all plots by the Reid et al.
(1966) ground-sign count technique was 34 animals per 0.4 ha (Table 2). The
erpreters
ission errors
ittle surface
e mounds
plants although
on interpretation
looked because
ated low
population estimate derived through photointerpretation of 1:600 scale color
infrared aerial photographs was 33 animals per 0.4 ha or 97 percent as accurate
as the estimate derived from the ground count technique. This relationship was
developed after the photointerpreted data had been normalized to ground counts
of earth mounds.
There was considerable variation in the photointerpreted gopher estimates
among individual plots. However, the close relationship of the collective values
comparing the ground based estimates and photointerpreted estimates indicates
increased sample size is accounting for among-plot variations. Therefore, the
photointerpretation technique described is a valid procedure to estimate gopher
populations provided ground truth is available to establish the coefficient
between known ground mound numbers and photointerpreted mound numbers.