Effects of defoliation by Douglas-fir tussock moth on timing and quantity of streamflow

Historic, archived document

Do not assume content reflects current
scientific knowledge, policies, or practices.

AN GH | |

INloatTH

Wiest

FOREST AND RANGE
EXPERIMENT STATIO |

PNW-326 December 1978

Effects of Defoliation by Douglas-fir Tussock Moth
on Timing and Quantity of Streamflow

gestir tus,

iS Gg by
3 %
a z Aan :
3 E J. D. Helvey, Principal Forest Hydrologist
% $
% ¢ and

cer. lo pme™
A. R. Tiedemann, Principal Range Scientist

ABSTRACT

Streamflow data from three watersheds in the Blue Mountains
of Oregon were tested for changes in annual runoff, summer
runoff, and peak discharge following the defoliation in 1972
: and 1973 caused by the Douglas-fir tussock moth. Annual runoff
from the Umatilla River watershed in 1974 was 13.2 cm greater
than the predicted value and 2.5 cm greater than the end point
95-percent confidence band for the baseline data. No changes
) in runoff were detected on the North or South Fork of the
Walla Walla River. Defoliation was more extensive on the
Umatilla drainage.

KEYWORDS: Runoff -)vegetation, insect damage (-forest,
defoliation damage, Douglas-fir tussock moth,
Orgyia pseudotsuga.

Acknowledgment
The work leading to this publication was funded in part

by the USDA Douglas-fir Tussock Moth Expanded Research and
Development Program.

JREST SERVICE -U.S. DEPARTMENT OF AGRICULTURE - PORTLAND, OREGON

INTRODUCTION

The purpose of the research reported in this paper was to determine
the effect of defoliation by the Douglas-fir tussock moth (Orgyia
pseudotsugata McDunnough) on annual runoff, seasonal runoff, and peak
discharge from major drainage systems. This information is necessary
for a complete evaluation of the moth outbreak on the affected eco-
systems. If discharge rates during spring runoff increase substantially,
culverts and bridges could be destroyed and stream habitat damaged for
aquatic life. On the other hand, an increase in flow during late
summer could enhance aquatic habitat and provide more water for irriga-
tion agriculture, provided water quality remains acceptable.

The water balance of a drainage basin can be expressed mathematically
as:

News (es i + is) & S

where R is runoff, P is precipitation, T as transpiration, is eimter—
ception, E is evaporation from land surfaces, and S is soil moisture
storage. The units are depth over the drainage area and the time
period usually is a year or a season.

It is generally agreed that vegetation reduction has no measurable
effect on precipitation amounts (McDonald 1960), but transpiration and
interception losses are reduced. Evaporation losses increase somewhat
because of increased soil exposure to solar energy. Soil moisture
storage increases and remains at a higher level than before vegetation
reduction. The net effect of complete vegetation removal, whether by
natural causes or by forest harvest, is an increase in annual runoff.
A partial reduction may or may not produce an increase in runoff,
depending on several factors including the percentage of vegetation
reduction and the balance between soil moisture storage and energy
available for evaporation. A more detailed discussion of the soil
moisture-energy-runoff relationships as they apply to this study will
be presented later.

PREVIOUS WORK (NATURAL DEFORESTATION)

One of the most extensive insect outbreaks in this country
occurred in Colorado between 1941 and 1946. The Engelmann spruce
beetle (Dendroctonus rufipennis (Kirby)) killed practically all of
the Engelmann spruce and lodgepole pine growing on 585 km? within
the White River watershed (Love 1955). Love's analysis of annual
water yield changes indicated that about 5 cm of extra water per unit
area were produced by the White River drainage after the insect attack.
A later analysis by Bethlahmy (1975) indicated that even 25 years
after the attack, runoff was still about 10 percent greater than the
natural values. Apparently the denuded forest was extremely slow in
recovering.

Tropical storms sometimes travel along the eastern part of North
America and cause considerable damage to forested areas. A hurricane
in 1938 was one of the most destructive on record as far as the forest
is concerned. It uprooted and broke off vast numbers of trees in two
New England watersheds. Patric (1974) analyzed historical flow
records and concluded that annual runoff increased about 12.5 cm per
unit area during the first year after the hurricane. Because of the
rapid recovery of hardwood forests, runoff returned to normal after
only 5 years.

Wildfire is perhaps responsible for destroying more vegetation in
western forests than any other natural cause. In 1970, about 485 km?
of forested land in north central Washington were blackened by wildfire.
Helvey (1972) reported average water yield increases of 8.4 cm per unit
area (50 percent) and water temperature increases of 5,5° C on the
Entiat Experimental watersheds during the lst year after the fire.
Klock (1972) found that water in the soil profile was 11.5 cm greater
in September 1971 than in September 1970, indicating: that a large part
of the transpiration savings were retained in the soil profile. Water
yield increases in later years were even greater (Helvey 1973), but
record precipitation amounts prevented an accurate determination of
the effect of vegetation reduction alone. The burned areas became
extremely sensitive to precipitation input, and runoff rates were much
higher than before the fire. Debris flows were common during the
second postfire year.

THE STUDY AREA

The study area, located in northeastern Oregon and southeastern
Washington, is part of the Blue Mountains. Topography of the Blue
Mountains varies from undulating plateaus to steep rugged mountains
reaching to 2 500 m elevation. Vegetation varies with elevation i.e.,
big sagebrush (Artemisia tridentata Nutt.) occupies the lowest levels;
and as elevation increases, vegetation changes to ponderosa pine
(Pinus ponderosa Laws.), to Douglas-fir (Pseudotsuga mensiesii (Mirb.)
Franco), to subalpine fir (Abies lasiocarpa (Hook.) Nutt.), and
finally to vegetation typical of above timberline conditions (Hall
S73) :

Precipitation is primarily a cool season phenomenon. Maritime
storms cross the mountains from west to east in the fall and winter.
Orographic lifting cools the moist air and causes precipitation to
fall. Average annual precipitation is 38 cm at the lowest elevations;
over 138 cm at the upper slopes. Approximately 80 percent of the
total annual precipitation falls between October 1 and May 31. At the
upper slopes, a snowpack usually begins to form by late November. It
increases in depth and water content until late March or early April.
Snowmelt begins on lower slopes with south exposure in February, and
the snow line advances to upper elevations. Complete snowmelt varies
from year to year depending on maximum snow accumulation and air
temperature in the early Spring months. The last snow usually is
melted by early June.

Runoff patterns are typical of areas where snow is the dominate
form of precipitation. Figure 1 illustrates average monthly flow rate
for the Umatilla River. Runoff increases during the fall months be-
cause of increased rainfall at the lower elevations, Rapid snowmelt
in March, April, and May produces maximum discharge rates during these
months. Flow rates decrease during summer months because evapo-
transpiration demand greatly exceeds rainfall input.

Oo nNM Ff DOD OO

Average monthly runoff (cm)

Hydrologic year

Figure 1.--Average monthly runoff from the Umatilla River.

METHODS

Runoff changes caused by vegetation reduction usually are deter-
mined by the paired watershed technique. That is, runoff from two or
more watersheds is measured during a 3- to 10-year calibration period
while the watersheds remain undisturbed. The watersheds are purposely
chosen for their similarity in vegetation, soils, and geomorphology.
One watershed is designated the control, and regression equations are
developed during calibration so that runoff from each watershed can be
accurately predicted from values measured on the control unit. After
the watersheds are satisfactorily calibrated, the vegetation on one
or more units is reduced by a predetermined amount while the control
unit remains undisturbed. Runoff changes due to vegetation reduction
are calculated by subtracting the value predicted by the calibration
regression from the measured value. If this absolute difference is
Significantly greater than zero at the accepted level of probability,
the change is attributed to reduced vegetation levels.

A variation of the paired watershed techniaue was used in this
study. Instead of the typical experimental watershed of 40-400
hectares, watersheds used in this study range up to 340 km2. Although

small drainages would have been more desirable, no watersheds within
the defoliated area had been monitored for runoff before defoliation
began. The only alternative was to choose watersheds within the
defoliated area for which the U.S. Geological Survey had collected
and published discharge records.

A map indicating areas of tussock moth defoliation in the Blue
Mountains was supplied by the Umatilla National Forest. Insect damage
was identified on this map as heavy mortality in 1972, heavy mortality
in 1973, top killing in 1973, and light defoliation in 1973. There is
some indication that the damage was not as severe as the initial
survey indicated. After the insects were killed with chemical spray,
some trees, which at first appeared to be heavily damaged, put out a
new set of needles and fully recovered.

Three watersheds which were partially defoliated and which have
been gaged by the Geological Survey were chosen for study (fig. 2).

ima

/
= ‘ Washington
ae ee es C jase
\ Oregon
\
North Fork
Walla Walla :
\
a aaa | N
}
| |
/
Umatilla _/”
River-—
Fa \
\
\
/
Legend

ae Heavy Mortality
Top Killing
Watershed Boundary

Scale
|
0) a 10km

Figure 2.--The study watersheds: Mill Creek, North Fork
Walla Walla River, South Fork Walla Walla River, and
the Umatilla River.

These are (1) the North Fork of the Walla Walla River (113 km2) , the
South Fork of the Walla Walla River (169 km2), and the Umatilla River
(352 km2). Runoff from Mill Creek (160 km2), was used as control
data since defoliation was slight on this drainage (Hicks 1977). A
dot grid was used to estimate the percentage of total watershed area
in each defoliation category listed above.

The analytical procedure used here is illustrated by the following
example: (1) Annual runoff volume from the Umatilla River and Mill
Creek was tabulated for each year between 1950-1971. (2) A scatter
diagram was plotted in which runoff from the Umatilla Basin (partially
defoliated) was the dependent variable (Y) and runoff from Mill Creek
was the independent variable (X). (3) A linear regression was computed
and the least squares line drawn on the scatter diagram. (4) The
location of several points on the 95-percent confidence limits was
calculated for the individual points on the scatter diagram using the
equation:

mr >
I+
Gh

7 y)2
(Residual MS) \(1 + = + Oe

(The interested reader is referred to Freese (1967) for a discussion
of the computation and interpretation of the confidence limits.) (5)
A smooth curve was constructed through the confidence limit points,
(6) Measured runoff from the Umatilla River during each year after
defoliation was plotted on the scatter diagram as a function of con-
currently measured runoff from Mill Creek. If a value after defolia-
tion was inside the confidence bands, we concluded that runoff during
that year was not significantly different from predefoliation values.
If the value was outside the confidence bands, we concluded that runoff
was different from the relationship before defoliation; and we specu-
lated on the cause of the difference:

The same steps as outlined above were followed for annual runoff,
seasonal runoff (April-June, July-September, and September-November) ,
and peak discharge from each of the partially defoliated basins,

Personnel in National Forest Administration were contacted for
information on past insect outbreaks and for timber harvest records.
According to these records, there has been no serious insect defolia-
tion in any of the drainages in recent times; but timber harvest has
proceeded on each watershed since 1950. There is no evidence to indi-
cate that logging on one drainage was enough greater than on the others
to cause measurable changes in water yield. Thus, runoff differences
between watersheds before the tussock moth outbreak are the result of
natural factors and not man's activity. The control watershed (Mill
Creek) serves as a municipal watershed for the city of Walla Walla,
Washington. The city diverts about 0.62 m/sec at a point 4 miles
above the gaging station for municipal use (U.S. Geological Survey
1975). No correction of the records was attempted for this diversion--
it was considered a constant value from year to year. No logging is
permitted on the headwaters of Mill Creek.

RESULTS AND DISCUSSION

Table 1 lists the estimated area percentages affected by the
insect in 1972 and 1973. It appears from this tabulation that the
Umatilla drainage was more severely defoliated than the other two.

If we assume that top killing removes about half of the transpiring
surface of a tree and heavy mortality removes all surfaces, the
Umatilla drainage suffered a 25-percent reduction in transpiring sur-
faces in 1972 and 1973 combined. The North Fork of Walla Walla River
lost about 16 percent of its foliage surfaces and the South Fork about
13 percent. Light defoliation in 1973 was about the same on all three
watersheds at 20 percent. The most severe damage occurred along ridge-
tops and upper slopes where soil moisture usually is more limiting than
on lower slopes. This is an important factor because vegetation
removal from upper slopes would be expected to influence runoff less
than an equal reduction on lower slopes. The reason for this conclu-
sion will be discussed later.

Table l1--Douglas-fir Tussock Moth Area-Activity in three
drainages of the Blue Mountains of Oregon

North Fork South Fork

Beery Walla Walla | Walla Walla | U™at+tla

---- + - ee Percent - - ----- -
Heavy mortality in 1972 8 4 i
Heavy mortality in 1973 1 2 10
Top killing in 1973 14 14 30
Light defoliation in 1973 20 21 20

Annual runoff from the Umatilla River is plotted in figure 3 as
a function of annual runoff from Mill Creek. Ninety-five-percent
confidence bands are included, as recommended by Freese (1967), to
illustrate variability in the data before insect defoliation began.
Data points after defoliation are identified on the figure by the year
of measurement.

Our statistical analysis indicated no effect of defoliation on
runoff in 1972 when the insect outbreak began, nor in 1973 when annual
precipitation was extremely low. In 1974, however, when about 25 per-
cent of the transpiring surface was removed and annual precipitation
was near the maximum ever recorded for the study area, annual runoff
was 13.2 cm greater than the predicted value (fig. 3). Although this
value is 2.5 cm greater than the upper end point of the confidence
band, it should be interpreted with caution because the measured value
on Mill Creek in 1974 was 22 cm greater than the largest value in the
calibration data. One basic assumption of regression analysis is that

110

100

90

80

70

60

50

Annual runoff, Umatilla River (cm)

40

e) 40, 50)” 60) (70 80) 590) 100 mat

Annual runoff, Mill Creek (cm)

Figure 3.--Annual runoff from the Umatilla River compared to annual
runoff from Mill Creek. The number beside each solid dot is the
year of measurement for that point.

prediction should be restricted to the range of data used to compute
the regression. Although runoff in 1974 was greater than the upper

end point of the 95-percent confidence band, the uncertainty associated
with the confidence band location prevents a definitive statement about
the effects of defoliation on runoff in 1974.

In 1975 and 1976 when precipitation was slightly above average,
runoff from the Umatilla River was near the predicted value. If the
increase in 1974 was due to decreased evapotranspiration resulting
from defoliation, it seems that increased runoff would have continued
into 1975 and 1976. It could be that tree recovery from the initial
defoliation (greenup) was sufficient to restore transpiration losses
to a level which approximated natural conditions.

The next step in the analysis was to test for seasonal changes in
runoff from the Umatilla River associated with the defoliation. Tests

were made on peak discharge, and total runoff during snowmelt (April-
June), during summer months (July-September), and during the autum
months (September-November). The test indicated no significant change
during snowmelt or the summer months. Runoff during the autumn months
of water year 1974, however, was significantly greater than the pre-
dicted value. Actual runoff during these months was 23.6 compared to
the predicted value of 17.8 cm. This was about 1.3 cm greater than the
upper end point of the confidence band (95-percent level) for the base-
line data.

Peak discharge data were highly variable, and no change could be
detected. This result was expected because even on small watershed
studies where runoff is measured much more accurately than is possible
on rivers, complete clearcutting produces only small increases in peak
flow rates (Harr 1976).

Plottings of annual runoff for the North Fork and the South Fork
of the Walla Walla River revealed no detectable change in runoff after
the defoliation (figs. 4 and 5).

90

80

70

60

50

40

30

20

10

Annual runoff, North Fork Walla Walla River (cm)

Top 20..:50 60, 70 80 90 100 110

Annual runoff, Mill Creek (cm)

Figure 4.--Annual runoff from the North Fork of Walla Walla River
compared to annual runoff from Mill Creek. The number beside
each solid dot is the year of measurement for that point.

160

150

140

130

120

110

100

South Fork Walla Walla River (cm)

Annual runoff,
0
S)

Annual runoff, Mill Creek (cm)

Figure 5.--Annual runoff from the South Fork of Walla Walla River
compared to annual runoff from Mill Creek. The number beside
each solid dot is the year of measurement for that point.

From these and previously published results, it appears that at
least three factors are involved in determining the amount of annual
runoff increase following vegetation reduction. These are (1) percent
of the total drainage area deforested, (2) location of the deforested
area with respect to the stream channel, and (3) current annual precip-
itation as a percent of the longterm mean value. Hibbert (1965) con-
cluded from his world wide literature survey that at least 20 percent
of a basin must be deforested before runoff significantly increases.
Hibbert gave two reasons for this result: First, removing a smaller
percentage of vegetation, such as thinnings, allows remaining trees to
increase their water use rates, especially in areas such as eastern

10

Oregon where potential evapotranspiration in late summer usually exceeds
available water supplies. Second, an increase must be larger than the
experimental error associated with the baseline data before a statisti-
cally significant change is indicated. Accuracy of the runoff data
used in this analysis is rated "good" by the U.S. Geological Survey.
This means that about 95 percent of the daily discharge values are
within 10 percent of the true value. Therefore, an increase smaller
than 10 percent in this study cannot be detected.

In areas where potential evapotranspiration exceeds available soil
moisture supplies, removing riparian vegetation has a larger effect on
runoff than an equal area of cutting on upper slopes. For example,
Rowe (1963) reported an increase in flow equal to 35 cm over the area
treated when riparian vegetation was removed from a drainage in
southern California. On the other hand, the riparian effect could not
be demonstrated in western North Carolina where precipitation during
all seasons exceeded potential evapotranspiration (Helvey and Hewlett
19627) .

Bethlahmy (1974) showed that increases in runoff after deforesta-
tion are directly related to current annual precipitation, i.e., in-
creases were much larger during wet than during dry years. Therefore,
the indicated runoff increase in 1974 from the Umatilla River probably
was caused by insect defoliation because this was the year of maximum
defoliation and an ample moisture supply.

SUMMARY

The trees on 16 percent of the North Fork of Walla Walla River
Basin, 13 percent of the South Fork of Walla Walla River Basin, and
25 percent of the Umatilla Basin were defoliated by Douglas-fir
tussock moth between 1972 and 1974. The integrated effects of this
natural activity on the water balance were determined by regression
analysis of runoff data. Runoff records from an adjacent basin (Mill
Creek) which received only minor activity, served as control data.

Because of the great variability in runoff data before defoliation
began, a rigorous test of runoff changes caused by insect activity could
not be made. Annual runoff from the Umatilla River in 1974, however,
was 13.2 cm greater than the predicted value and 2.5 cm greater than
the upper end point of the 95-percent confidence band for the baseline
data.

No changes in annual runoff were detected from the lightly defoli-

ated basins, and no effect of defoliation on peak discharge was detected
on any of the watersheds.

1l

Bethlahmy ,
1974.

LITERATURE CITED

N.
Water supply as affected by micro- and macro-watershed
management decisions on forest lands. Northwest Science
48(1):1-8.

Bethlahmy, N.

WTS

Reese ihe
1967.

A Colorado episode: beetle epidemic, ghost forests, more
streamflow. Northwest Science 49(2):95-105.

Elementary statistical methods for foresters. Forest
Products Labs» USDA’ Form. Sexy. AenicemHandb), ysl ano mape

Hallils Eredernnuck Ge

UTS «

Plant communities of the Blue Mountains in eastern Oregon
and southeastern Washington. USDA For. Serv. Pac. North-
west Region 6, Area Guide 3-1, 62 p.

Harr, R. Dennis

IDO.

InENWEN 5 dS
ISIVZ

nUWEY 5 dc
NOB.

IsUWESY 4 Vo
NIZE

Hibbert, A.

IGS

Forest practices and streamflow in western Oregon. USDA
For. Serv. Gen. Tech. Rep. PNW-49, 18 p. Pac. Northwest
For. and Range Exp. Stn., Portland, Oreg.

D.
First-year effects of wildfire on water yield and stream
temperature in north-central Washington. Jn S. C. Scallany,
T. G. McLaughlin, and W. D. Striffler (Eds.), Watersheds in
Transition, Amer. Water Resour. Assoc. and the Colorado State
Unive) Pi. S08 32 =
DE
Watershed behavior after forest fire in Washington. Pro-
ceedings of the Irrigation and Drainage Division Specialty
Conference, Amer. Soc. Civil Eng., Fort Collins, Colloz, |
p. 403-422.

Ds, ginGl JN Ws istewilerce.
The annual range of soil moisture under high rainfall in
the Southern Appalachians. J. For. 60(7) :485-486.

R.

Forest treatment effects on water yield. Jn W. E. Sopper
and H. W. Lull) (Eds), Int: Symp. For: Hydrology. pi.) 527 -
543. Pergamon Press, New York.

Hicks, Martin Ed.

OWT

WA

Effects of deforestation by the Douglas-fir tussock moth

on the quality of streamflow and stream productivity param-
eters. M.S. Thesis, Central Washington State College,
Ellensburg, Wa., 79 p.

Klock, -G. 10.
1972. Soil moisture trends on mountain watersheds following
forest fire. (Abstr.) 45th Annu. Meet., Northwest
Serentif£ie Assoc., p. 7.

Love, L. D.

1955. The effect of streamflow on the killing of spruce and pine

by the Engelmann spruce beetle. Trans. Amer. Geophys.
Union 36(1):113-118.

McDonald, J. E.
1960. The evaporation-precipitation fallacy. Weather 17(5):
168-177.

Prrierii(es lc Jae
1974. River flow increases in Central New England after the
hurricaneoL 1958. Ji. For. 72 (1) 221-25.

Rowe, L. R.
1963. Streamflow increases after removing woodland riparian

vegetation from a southern California watershed. J. For.

61:365-370.

U.S. Geological Survey.
1975. Water resources data for Washington. Water Data Report

WA-75-1. 684 p.

GPO 987-1

20

tS

The mission of the PACIFIC NORTHWEST FOREST
AND RANGE EXPERIMENT STATION is to provide the
knowledge, technology, and alternatives for present and
future protection, management, and use of forest, range, and
related environments.

Within this overall mission, the Station conducts and
stimulates research to facilitate and to accelerate progress
toward the following goals:

1. Providing safe and efficient technology for inventory,
protection, and use of resources.

2. Developing and evaluating alternative methods and
levels of resource management.

3. Achieving optimum sustained resource productivity
consistent with maintaining a high quality forest
environment.

The area of research encompasses Oregon, Washington,
Alaska, and, in some cases, California, Hawaii, the Western
States, and the Nation. Results of the research are made
available promptly. Project headquarters are at:

Fairbanks, Alaska Portland, Oregon
Juneau, Alaska Olympia, Washington
Bend, Oregon Seattle, Washington
Corvallis, Oregon Wenatchee, Washington

La Grande, Oregon

Mailing address: Pacific Northwest Forest and Range
Experiment Station
P.O. Box 3141
Portland, Oregon 97208

Agriculture is dedicated
ation’s forest resources
life, and recreation.

and private forest
ional Grasslands, it
ly greater service to

yportunity Employer.
iven equal consideration

f