VOL. 4, NO. 5
WATER RESOURCES RESEARCH
OCTOBER 1968
Conversion of Hardwood-Covered Watersheds to White Pine
Reduces Water Yield
W. T. SWANK AND N. H. MINER
Coweeta Hydrologic Laboratory
Southeastern Forest Experiment Station
Forest Service, U. S. Department of Agriculture, Asheville, North Carolina 28802
Abstract. Mixed mature hardwoods were cleared from two Southern Appalachian experimental watersheds, and the areas were planted with eastern white pine in 1956-1957. Once
the pine crowns began to close, streamflow steadily declined at a rate of 1 to 2 inches per
year. By 1967, water yield was 3.7 inches less from a 10-year-old pine stand on a southfacing watershed than expected water yield from the original hardwood forest. Most of the
water yield reduction occurred during the dormant season and was attributed mainly to
greater interception loss from white pine than from hardwoods. Because interception differences increase as white pine matures, an even greater reduction in streamflow is expected.
(Key words: Water yield; watershed management; interception; basins; white pine)
Perhaps without fully realizing it, the watershed manager may play an important part in
developing water resources through his management of the forests. For example, research
has shown that harvesting forest products can
substantially increase streamflow [Hibbert,
1967; Lull and Reinhart, 1967] and that, unless
care is exercised during road construction and
logging, water quality will suffer [Packer, 1967].
Water yield may also be influenced in less obvious ways; for example, in the choice of tree
species to be grown.
Hewlett [1958] presented reasons why streamflow from pine- and hardwood-covered watersheds might differ and described the studies at
the Coweeta Hydrologic Laboratory, where two
catchments were cleared of native hardwoods
and planted to white pine (Pinus strobus L.).
This paper presents results from ten years of
streamflow measurement following the cover
type conversion of these Southern Appalachian
watersheds. Results provide hydrologists and
municipal watershed managers with conclusive
evidence of some early effects on water yield
when mixed hardwood forests are converted to
pine.
HISTORY OP EXPERIMENTAL AREA
The experimental watersheds are located in
the high precipitation (2% snow) region of
947
western North Carolina and are characterized
by deep, permeable soils. Slopes are steep, and
prior to treatment oak-hickory was the prevalent forest type (Table 1).
Watershed 1 is 40 acres in size and slopes
at 48% to the south. Rainfall averages 68
inches annually, and streamflow during the
10-year calibration period beginning in 1944
averaged 31 inches. In 1954, all vegetation
within the cove type (25% of the watershed
area) was deadened with chemicals. The entire
watershed was clearcut during October-December 1956; slash was scattered and partially
burned, and white pine seedlings (2-0 stock)
were planted at a 6- by 6-foot spacing during
the winter of 1957. Initial survival was only
00%, and dead seedlings were replaced in
1958. Thereafter, competing hardwood sprouts
were cut or sprayed with 2, 4, 5-T as required
to release the pine. In 1967, the pine averaged
20 feet in height (Figure 1), contained 32
square feet basal area with a stocking of 720
trees per acre, and had attained a mean breast
height diameter of 2.9 inches.
Watershed 17, in contrast to Watershed 1,
is 33 acres in size and slopes 57% to the northwest (Figure 2). Annual precipitation averages 76 inches, and streamflow for the four-year
calibration period averaged 27 inches. All shrub
and forest vegetation was cut during January-
SWANK AND MINER
TABLE 1. Cruise Summary of Coweeta Watersheds 1 and 17
Basal Area (Ft2/Acre)
Watershed 1 Watershed 17
Species
(1954 cruise*) (1941 cruise*)
White Oak Group
17
25
20
Red Oak Group
27
7
Hickory
8
Laurel and
Rhododendron
6
8
20
Others
26
Total Original
84
Hardwood Stand
Eastern White
32
Pinef
* Year preceding first treatment,
f Cruised in winter 1967.
SO
42
March 1942 without removal of timber products. Thereafter, annual sprout growth was cutback most years between 1943-1955, and a low
herbaceous and shrub cover developed. White
pines were planted in 1956, one year before
Watershed 1 was planted; survival averaged
over 90%, and dead seedlings were replaced
the following year. Hardwood sprout control
was similar to that on Watershed 1. By 1967,
Watershed 17 had 1/3 more basal area (42
square feet total) than Watershed 1. Stockingwas 685 trees per acre with a mean diameter
of 3.4 inches, and height averaged 25 feet.
RESULTS
Figures 3 and 4 show the annual difference
between actual and predicted water yield for
calibration and treatment periods on Watersheds 1 and 17, respectively. Analysis is based
on monthly strcamflow prediction equations
derived from a calibration period during which
a control watershed and the watershed to be
treated were in an undisturbed condition. Annual treatment effects are the sums of deviations from twelve monthly calibration regressions (actual minus predicted flow).
The effects on streamflow of clearcutting
these watersheds were discussed in detail by
Hoover in 1944, Hewlett and Hibbert in 1961,
and Hibbert in 1967. The most striking contrast in early results is the nearly threefold
difference in yield increase observed when these
north- and south-facing basins were cut. The
large difference in yield response with aspect is
consistent with findings from other watershed
cutting experiments at Coweeta [Hibbert,
1967]. Swift [1960] calculated that the theoretical extraterrestrial radiation on Watershed 1
is 38% greater than on Watershed 17, and
Douglass [1967] suggested that these radiation
differences ma}' be a major factor in the different yield response to clearcutting.
Changes in streamflow from the treated
catchments remained relatively constant from
the second year after pine was planted until
about the sixth year. Thereafter, streamflow
additions from the clearcutting treatment steadily declined as the crowns of the white pine
closed. By 1966, water yield from Watershed
1 was 1.9 inches less than predicted for the
original hardwood cover. The next year the
white pine basal area had reached 32 square
feet per acre and streamflow dropped 3.7 inches
below that predicted for the original (84 square
feet) hardwood stand. Annual streamflow is
significantly different at the 0.95 probability
level if it differs from the predicted flow by
more than ± 1.2 inches for Watershed 1 in
1967. This error term was derived from the
sum of the variances for the twelve monthly
strcamflow estimates.
Watershed 17 showed a net decrease in
streamflow from the calibration period of only
1.3 inches (±1.7 inches is required for significance) in 1967, but total reduction in water
yield has been greatest on this catchment. Comparison of total yield reductions can be made
if a period can be selected when vegetation was
similar on both catchments. Preceding reforestation, cutting operations were different; at
time of planting, shrub cover was much greater
on Watershed 17.
However, one might assume minimum vegetation differences between catchments in the
third and fourth years following planting, because release measures had reduced shrub cover
to similar levels, and pines were just beginning
rapid growth. Thus, using the average deviation in streamflow for the third and fourth
years as a reference point, the change in streamflow deviations from regression by 1967 was
10.9 inches on Watershed 17 and 5.6 inches
on Watershed 1. The increase in pine basal
area during the same period has also been
greatest on Watershed 17.
The sharp decline in streamflow for the past
Water Yield
949
Fig. 1 yici ten gicmmu =c t on^ thi \\lnte pine phntition on WUei-hed 1 a-\ciages 20
feet in height with a stocking of 720 trees per acre and a basal area of 32 square feet per
acre.
five years has averaged more than 2 inches
per year on Watershed 17 and 1 inch per year
on Watershed 1; thus far, there is no evidence
that the rate of change in annual streamflow
from either watershed has begun to taper off.
In contrast, Kovner [1956] reported a gradual
decline in the rate of streamflow reductions
following early regrowth of a clearcut hardwood forest at Coweeta.
Reductions in monthly water yields for Wa-
tershed 1 in 1966 and 1967 were greatest during
the dormant season, even though most months
contributed to the yearly decrease (Figure 5).
In the months of May and June and November
through March 1967, streamflow was significantly less (0.95 probability level) than predicted
for the previous hardwood cover. A pattern of
increasing monthly streamflow reductions from
year to year was also evident in data from
preceding years. On Watershed 17, a significant
950
SWANK AND MINER
'.Jfl
Fis, 2 An \cinl \u\\ oi Ccmccti Watcishcd 17 shows c ilchmcnt iclief ind genci il structure of the pine stand nine growing seasons after planting compared with adjacent mature
hardwood stands. A 20-year-old test planting of white pine appears in the lower left margin
of the catchment.
decrease in streamflow occurred only in the
month of January; however, the same pattern
of progressively falling yields in most months
for successive years was likewise found on this
catchment (Figure 5). Watershed 17 had a
comparatively short (four-year) calibration period, which contributed to the large confidence
intervals for streamflow estimates. For example,
streamflow decreases in May and April are
large, but they are not yet significant.
DISCUSSION
These experiments provide the first conclusive evidence of how streamflow changes when
hardwood stands are converted to white pine.
The implications are far-reaching. Streamflow
levels began dropping six years after conversion,
and, after only ten years of growth, annual
evapotranspiration losses were greater from
white pine than from the hardwoods it replaced.
A greater, perhaps much greater, reduction in
Water Yield
951
1
^
^
Ci
--J
^
§
CALIBRATION PERIOD
| 5
is
1$
•=c
1 9 4 5 4 6 4 7 4 8 4 9 5 0 5 1 52 53 54 55 56 57 5859 60 6 I 62 63 64 65 66 67
YEAR
Fig 3. Actual minus predicted annual streamflow for Coweeta Watershed 1 during calibration and treatment periods (May-April water year).
streamflow is expected as
ture. We suggest that the
primarily from increased
the pine.
Summaries of forest
the plantations mareduced flow results
interception loss by
interception
studies
[Zinke, 1967; Kittredge, 1948] indicate that
higher interception losses could be expected
from white pine during the dormant season,
since the intercepting surface in a hardwood
stand is drastically reduced following leaf fall.
Uj ^
j* "- L
193839404142434445464748495051 5253545556575859606! 626364656667
" VEAR
Fig. 4. Actual minus predicted annual streamflow for Coweeta Watershed 17 during calibration and treatment periods (May-April water year).
SWANK AND MINER
952
COWEETA WATERSHED I
M i l l !
MAY JUN JULAUGSEPOCTNOVOECJAN.FEBMAR APR. MAYJUNJUL AUG.SEROCTNOVDEC.JANFEB.MARAPR.
Fig. 5. Actual minus predicted monthly strcamflow for Coweeta Watersheds 1 and 17
during 1966 and 1967 water years.
This is confirmed by experiments conducted
near Coweeta, which show that 10-year-old
white pine intercepts more precipitation than
mature hardwoods [Hdvey, 1967], and most of
the difference occurs during the period when
hardwoods are leafless. Monthly differences in
interception loss for these two forest types
were calculated from equations derived by.
Helvey [1967] and Helvey and Patric [1965].
Since interception varies with frequency and
amount of rainfall, actual precipitation received on Watershed 1 during 1967 was used
in calculating 10-year-old pine and mature
hardwood values (Figure C). The figure shows
that interception differences between young
pine and mature hardwoods are closely correlated with observed reductions in water yield
from Watershed 1 during November through
LEGEND
••ACTUAL MINUS PREDICTED STREAMFLOW
£52) INTERCEPTION DIFFERENCEIMATURE HARDWOOD MINUS 10-YEAR OLD WHITE PINE)
I I INTERCEPTION DIFFERENCEIMATURE HARDWOOD MINUS 60-YEAR OLD WHITE PINE)
JUNE JULY AUG. SEPT. OCT. NOV. DEC. JAN. FEB. MAR APR.
r
.1.0-s
12Fig. 6. Comparison of actual minus predicted monthly stroamflow for Watershed 1 in
1967 and differences in monthly interception for 10- and 60-year-old white pine plantations
versus mature hardwoods.
Water Yield
March. Water yield reductions are insignificant
in the period July through October, when both
stands are in full leaf and interception differences are small. Streamflow deviations and
interception differences for the period of July
through March are highly correlated (r =
0.92). Watershed 1 was selected for the above
discussion because the precision of predicting
yield changes is greater there than on Watershed 17. The interception-water yield reduction
pattern was similar but less distinct on Watershed 17.
Interception loss appears to account for only
part of the negative deviations in streamflow
during April, Ma}', and June (April 1967 represents an unusual event, since previous years
showed consistent reductions in water yield for
this month). Streamflow reductions during
these months might be partially attributed to
transpiration differences, since pine foliage is
transpiring water before and during leafing out
of hardwoods. Differences in stage of development of transpiring surfaces have largely disappeared by the end of June. From Raber's
[1937] summary of transpiration studies, itappears that during summer transpiration rates
of hardwoods per unit of leaf area are 2 to 3
times greater than conifers. However, Kramer
and Kozlowski [1960] point out that, for trees
of similar size, transpiration per tree may be
higher for conifers because of a greater total
leaf area. Quantitative data on transpiration
differences between conifers and hardwoods
during winter months are limited, but most
studies indicate that rates per unit surface area
for conifers are only slightly greater than for
hardwoods [Kozlowski, 1943; Weaver and Mogensen, 1919]. Comparative measurements of
seasonal transpiration and transpiring surface
area are needed before convincing statements
can be made about the effect of transpiration
differences on water yield. As these catchment
experiments at Coweeta continue, physiological
studies will be made in an effort to clarify
further the relative importance of transpiration
and interception.
We hypothesize that streamflow from these
white-pine-covered watersheds will continue to
decline, and that even greater differences in
water use are inevitable. There is evidence
from previous Coweeta studies that as hardwood stands regrow in diameter and height
953
after cutting, evapotranspiration increases and
streamflow progressively decreases [Douglass,
1967; Hewlett and Hibbert, 1961; Hibbert,
1967; Kovner, 1956]. A similar response is
expected as the pine stands mature. The hypothesis is further supported by Helvey's [1967]
interception work, which shows an average annual interception difference of 9 inches between
60- and 10-year-old white pine stands. The data
in Figure 6 indicate that large interception
differences between pine and hardwoods may
eventually appear during both winter and
summer months.
The results and supporting evidence presented here are of practical significance to
watershed managers. Conversion from mixed
hardwoods to white pine has lowered water
yield, and all evidence indicates that substantially larger reductions in yield must be expected. Although results reflect a specific set
of climatic, vegetative, and soil conditions, and
identical water yield reductions would not be
expected elsewhere, the evaporative processes
involved are universal. Thus, a trend toward
reduction in total water yield should also be
expected in other regions.
Most of the flow reduction has come during
the dormant season, when high flows fill storage
reservoirs. In drought years, when streamflow
is insufficient to refill reservoirs, water shortages
will be aggravated by hardwood to pine conversion. The ultimate impact of conversion on
the low flow period is uncertain, but continuation of the current- monthly trend of flow reduction will adversely affect flow during this
period. Thus, where water is the resource of
primary concern, as it is on municipal watersheds, it would appear wise to consider carefully the effect on water yield of converting
from hardwood to white pine.
REFERENCES
Douglass, J, E., Effects of species and arrangement, of forests on evapotranspiration, Proceedings oj the International Symposium on
Forest Hydrology, 451-461, Pergamon Press,
Inc., New York, 1967.
Hclvcy, J. D.. Interception by eastern white
pine, Water Resources Res., 3, 723-729, 1967.
Helvey, J. D., and J. H. Pat.ric, Canopy and
litter interception of rainfall by hardwoods of
eastern United States, Water Resources Res., 1,
193-206, 1965.
Hewlett. J. D., Pine and hardwood forest water
954
SWANK AND MINER
yield, J. Soil Water Conserv., 13, 106-109, 1958.
Hewlett, J. D., and A. R. Hibbert, Increases in
water yield after several types of forest cutting,
Bull, intern. Assoc. Sci. Hydrol, 6, 5-17, 1961.
Hibbert, A. R., Forest treatment effects on water
yield, Proceedings of the International Symposium on Forest Hydrology, 527-543, Pcrgamon Press, Inc., New York, 1967.
Hoover, M. D., Effect of removal of forest vegetation upon water yields, Trans. Am. Geophys.
Union, 6, 969-975, 1944.
Kittredge, J., Forest Influences, McGraw-Hill
Book Company, New York, 99-114, 1948.
Kovner, J. L., Evapotranspiration and water
yields following forest cutting and natural regrowth, Proc. Soc. Am. Foresters, 106-110, 1956.
Kozlowski, T. T., Transpiration rates of some
forest tree species during the dormant season.
Plant Physiol, IS, 252-260, 1943.
Kramer, P. J., and T. T. Kozlowski, Physiology
•oj Trees, McGraw-Hill Book Company, New
York, 1960.
Lull, H. W., and K. G. Reinhart, Increasing water
yield in the Northeast by management of for-
ested watersheds, U. S. Forest Serv. Res. Paper
NE-66, 45 pp., 1967.
Packer, P. E., Forest treatment effects on water
quality, Proceedings oj the International Symposium on Forest Hydrology, 687-699, Pergamon
Press, Inc., New York, 1967.
Raber, 0., Water utilization by trees with special
reference to the economic forest species of the
North Temperate Zone, U. S. Dept. Agr. Misc.
Publ. No. 257, 97 pp., 1937.
Swift, L. W., Jr., The effect of mountain topography upon solar energy theoretically available for evapotranspiration, M. S. thesis, North
Carolina State University, Raleigh, 1960.
Weaver, J. E., and A. Mogensen, Relative transpiration of coniferous and broadleaved trees in
autumn and winter, Bot. Gaz., 68, 393-424, 1919.
Zinke, P. J., Forest interception studies in the
United States, Proceedings oj the International
Symposium on Forest Hydrology, 137-161, Pergamon Press, Inc., New York, 1967.
(Manuscript received May 2, 1968;
revised May 27, 1968.)