LETTERS

PUBLISHED ONLINE: 20 APRIL 2014 | DOI: 10.1038/NCLIMATE2198

Hydrological effects of forest transpiration loss in

bark beetle-impacted watersheds
Lindsay A. Bearup1,2,3*, Reed M. Maxwell1,3,4*, David W. Clow5 and John E. McCray1,2

The recent climate-exacerbated mountain pine beetle
infestation in the Rocky Mountains of North America has
resulted in tree death that is unprecedented in recorded
history. The spatial and temporal heterogeneity inherent in
insect infestation creates a complex and often unpredictable
watershed response, influencing the primary storage and
flow components of the hydrologic cycle. Despite the
increased vulnerability of forested ecosystems under changing
climate1 , watershed-scale implications of interception, ground
evaporation, and transpiration changes remain relatively
unknown, with conflicting reports of streamflow perturbations
across regions. Here, contributions to streamflow are
analysed through time and space to investigate the potential
for increased groundwater inputs resulting from hydrologic
change after infestation. Results demonstrate that fractional
late-summer groundwater contributions from impacted
watersheds are 30 15% greater after infestation and
when compared with a neighbouring watershed that
experienced earlier and less-severe attack, albeit uncertainty
propagations through time and space are considerable. Water
budget analysis confirms that transpiration loss resulting
from beetle kill can account for the relative increase in
groundwater contributions to streams, often considered
the sustainable flow fraction and critical to mountain water
supplies and ecosystems.
In Colorado alone, the mountain pine beetle (MPB) has
impacted over 1.3 million hectares of pine forest2 . Although
evapotranspiration is generally assumed to decrease in beetle-
affected watersheds, tree death also causes competing effects on
evapotranspiration. By the end of the first growing season following
infestation, a killed pine no longer transpires3 , causing the needles
to turn red (identified as red-phase) and begin to drop. Within three
to four years after infestation, most trees have lost all remaining
needles (grey-phase; ref. 4). The resultant loss of canopy cover
increases fluxes of water and energy to the ground surface, causing
changes in soil moisture dynamics and snowmelt processes5,6 that
may offset the effects of reduced evapotranspiration. Increases in
soil moisture6,7 are dependent on the net increase of water inputs due
to losses of transpiration and canopy evaporation balanced against
the net decrease in moisture from higher solar exposure, surface
temperature and ground evaporation6 . The interactions among
these processes are poorly understood across scales, highlighting
the need for better quantification of net transpiration changes from
MPB infestation at the watershed scale.
Transpiration is commonly quantified using sap flux3 , eddy
covariance8 or energy balance formulations9 . Each of these methods
presents limitations. For sap flux and energy-based approaches,
upscaling stand-scale estimates requires spatially comprehensive
measurements to capture the heterogeneity of vegetation and
local energy balances. Eddy covariance methods provide larger-
scale estimations of evapotranspiration but are less reliable in
mountain environments10 and do not separately assess evaporation
and transpiration. The exceptional extent of tree death from the
MPB provides a unique opportunity to evaluate the contribution
of tree processes to the hydrologic cycle at watershed scales, where
water budget perturbations are complex and often combine non-
uniquely. Here, we quantify hydrologic changes in MPB-impacted
watersheds by identifying changes in streamflow contributions
through a chemical and isotopic hydrograph separation analysis.
The importance of transpiration loss is relative to the magnitude
of the other components of the hydrologic cycle, for example,
precipitation, snowmelt, evaporation and soil moisture11 . In the
Rocky Mountains of North America, the effect of transpiration at
the watershed scale may be most apparent during late summer, when
near-surface antecedent soil moisture and snow inputs approach
their annual minima, and the relative importance of subsurface
contributions is greatest12,13 . During this low-flow period, loss
of transpiration may lead to measurable increases in recharge
and groundwater contributions to streamflow, whereas loss of
interception and increased ground evaporation would influence
both surface and subsurface contributions to streamflow, as
conceptualized in Fig. 1. The distribution of late-summer flows
is not commonly studied, but may have important implications
for water supply, water rights, impairment of riverine ecosystems,
and water quality concerns, such as formation of disinfection by-
products in water from MPB-impacted watersheds14 .
Components of the water cycle, including flow paths and water
sources, can be investigated using chemically and isotopically
based hydrograph separation techniques, including endmember
mixing analysis15 (EMMA). A limited number of studies have
applied these techniques to understand flow path alterations from
changes in forest cover, focusing on forest management practices
and fire16,17 . Hydrologic responses to fire and logging are imperfect
analogues to the response of tree die-off from widespread insect
infestation. These disturbances probably represent the upper
bound of hydrologic perturbations. Infestation does not influence
soil compaction or repellency, which affect runoff and recharge
partitioning18,19 . Furthermore, the spatial and temporal patterns
of insect-induced tree death are more complex and variable
across broader landscapes and longer time frames than most
previously studied wildfire or logging events, resulting in different
compensatory and scaling mechanisms with infestation18 . The only

1 HydrologicalScience and Engineering Program, Colorado School of Mines, Golden, Colorado 80401, USA, 2 Department of Civil and Environmental
Engineering, Colorado School of Mines, Golden, Colorado 80401, USA, 3 Integrated GroundWater Modeling Center, Colorado School of Mines, Golden,
Colorado 80401, USA, 4 Department of Geology and Geological Engineering, Colorado School of Mines, Golden, Colorado 80401, USA, 5 Colorado Water
Science Center, US Geological Survey, Denver, Colorado 80225, USA. *e-mail: lbearup@mines.edu; rmaxwell@mines.edu

NATURE CLIMATE CHANGE | VOL 4 | JUNE 2014 | www.nature.com/natureclimatechange 481

2014 Macmillan Publishers Limited. All rights reserved.

LETTERS
When needles fall, storage and
evaporation of intercepted
precipitation are lost
Transpiration ceases in
the first year after attack
After 23 years,
needles fall, increasing
radiative fluxes and
ground evaporation
Postimp
act wate
r-table
Preimp
act wate Increased
r-table groundwat
contributio er
n to stream
flow?
Figure 1 | Conceptual model of water cycle changes with tree death
induced by mountain pine beetles. Under normal circumstances, green
trees use shallow groundwater in late summer for transpiration. Red- and
grey-phase trees cease transpiring, leading to higher water-tables and
greater water availability for groundwater flow to streams. Dying trees
begin to drop their needles, ultimately leading to a loss of interception and
shading. The loss of canopy cover reduces canopy evaporation but also
increases evaporation from the forest floor. Unlike changes in transpiration,
interception and shading losses impact all components of the water budget
at the forest floor, including potential runoff and recharge.
existing study using hydrograph separation techniques to analyse
hydrologic changes from bark beetle infestation analysed large
storm events, and could not distinguish subtle differences at high
flows; however, comparing precipitation with components of the
hydrologic cycle indicated increases in surface and groundwater
runoff in infested watersheds, suggesting important hydrologic
changes20 . Here, we use hydrograph separation and stable-
isotope analysis to determine spatial and temporal alterations to
late-summer streamflow from MPB-induced transpiration loss
at the watershed scale. This work improves understanding of
the interconnections between climate, forest perturbation, and
streamflow generation in the mountain headwaters of important
water supplies.
Two watersheds in Rocky Mountain National Park, Colorado,
USA (RMNP) are compared, the Big Thompson and North Inlet
(Fig. 2). The watersheds have similar geology and soils, dominated
by granite and biotite schist bedrock and associated alluvium,
colluvium or till21 . The forested areas in both watersheds are
dominated by subalpine mixed conifer forest, with more uniform
lodgepole pine forests in the North Inlet watershed than in
the Big Thompson22 . MPB infestation in North Inlet began in
2004, approximately 2 years earlier than in the Big Thompson
(Supplementary Table 1) with no significant new infested area
( 0.01% watershed area) since 2009, whereas the Big Thompson
continued to experience new infestation from 2006 to 2011. The
differing timeline of infestation may be attributed to the watersheds
locations on opposite sides of the Continental Divide (Fig. 2); annual
aerial survey maps document MPB infestation moving west to east
482
2014 Macmillan Publishers Limited. All rights reserved.
NATURE CLIMATE CHANGE DOI: 10.1038/NCLIMATE2198
tte
Pla N
RMNP S.
Denver
Colorado
Arkansas
T E
BT
NI
Snow pit
Rain gauge
Stream gauge
Groundwater well Elevation km
Continental divide 4,345 m
0 5 10
<30% MPB impacted 2,315 m
>30% MPB impacted
Figure 2 | Map of Rocky Mountain National Park and sampling sites in the
Big Thompson and North Inlet watersheds. Mountain pine beetle
(MPB)-impacted area denotes cumulative impacts through 2012. By the
end of the study period, over 30% of the Big Thompson (BT) watershed
experienced MPB infestation whereas the North Inlet (NI) watershed was
less impacted with no new infestation. Relief map based on the National
Elevation Dataset available at http://nationalmap.gov/viewer.html.
across the Colorado Rocky Mountains between 2007 and 20092 .
Precipitation, mostly in the form of snow, generally increases with
elevation. Snowfall in 2012 was similar in both watersheds, with
a watershed average 1 April snow water equivalent of 12.56 cm in
the Big Thompson and 12.96 cm in the North Inlet23 . This paired
watershed approach provides a comparison of two watersheds that
experienced different MPB impact but similar weather in 2012.
Fortuitously, a similar study14 was conducted in the Big Thompson
study area in 1994, which provided a unique opportunity to analyse
data representative of hydrologic behaviour before infestation
(Supplementary Section 1.2 for interannual and interwatershed
weather comparisons). This temporal analysis also controls for
spatial variability in a way that the paired watershed approach
cannot. The combined spatial and temporal approach provides
unique perspective to make traditionally challenging comparisons.
As detailed in the Methods and Supplementary Information,
a three-component hydrograph separation analysis using rain,
snow and groundwater identified differences in subsurface
contributions to streamflow compared spatially and temporally.
Sufficient groundwater records were not available to identify
changes in groundwater storage directly. As depicted in Fig. 3,
the Big Thompson experienced a higher fraction of late-summer
groundwater contributions to streamflow following several years
of MPB-caused tree mortality (2012), than before infestation
(1994). Spatial comparison for 2012 also found higher groundwater
NATURE CLIMATE CHANGE | VOL 4 | JUNE 2014 | www.nature.com/natureclimatechange
NATURE CLIMATE CHANGE DOI: 10.1038/NCLIMATE2198 LETTERS
a Big Thompson b North Inlet
1.0 1.0

Fractional contribution to streamflow

Fractional contribution to streamflow

0.8 0.8

0.6 0.6
2012

2012
0.4 0.4

Rain
0.2 0.2
Snow
Groundwater

0.0 0.0
Jul. Aug. Sep. Oct. Jul. Aug. Sep. Oct.

Big Thompson
c 1.0 d 1.0

Fractional groundwatercontribution to streamflow

Fractional contribution to streamflow

0.8 0.8

0.6 0.6
gT
2 Bi
1994

201
0.4 0.4
2012 N. Inlet

T
0.2 0.2 1994 Big

0.0 0.0
Jul. Aug. Sep. Oct. Jul. Aug. Sep. Oct.

Figure 3 | Fractional contributions of endmembers to streamflow. ac, Contributions of rain (navy), snow (cyan) and groundwater (orange) to streamflow
in Big Thompson, 2012 (a); North Inlet, 2012 (b); and Big Thompson, 1994 (c). d, Groundwater contributions (orange lines) with propagated uncertainty
(grey shading) for each analysis. Groundwater contributions are greatest in the most actively impacted watershed case, in a. The Big Thompson 2012
groundwater fractions in a and d use the time-varying groundwater endmember and the dashed line represents the constant groundwater endmember. The
1994 methodology is consistent with the dashed line of the 2012 Big Thompson study, and the solid line is consistent with the North Inlet study.

contributions to streamflow in the Big Thompson than in the North 2.5

1994 net streamflow 2012 net streamflow
Inlet, where tree mortality was less widespread and less recent.
1994 groundwater 2012 groundwater
Sensitivity analysis of the endmembers identified two methods
of quantifying the groundwater endmember that provided the 2.0
largest range of possible groundwater contributions, including
Daily discharge (mm)

shallow groundwater samples collected biweekly (that is, the time-

varying groundwater endmember) and average pre-melt baseflow 1.5
samples (that is, the constant groundwater endmember). On the
basis of these methods, the 2012 Big Thompson mean fractional
1.0
groundwater contribution ranged from 0.47 to 0.56 0.11,
compared with 0.18 0.08 in 1994 and 0.30 0.04 in the
North Inlet (Fig. 3). The constant groundwater endmember 0.5
approach is consistent with the 1994 study methodology and is
used for further temporal comparisons unless otherwise noted.
Both temporal and spatial analyses found greater fractional 0.0
groundwater contributions to streams in watersheds where MPB Jul. Aug. Sep. Oct.
impact was greater. Endmember compositions naturally vary
owing to differences in spatial characteristics such as elevation Figure 4 | Hydrograph separations presented as partitioning of the total
and subsurface heterogeneity or isotopic processes related to daily stream discharge (in mm; full bar height) for the 1994 (blue) and
interception and snowmelt (Supplementary Table 3). Despite 2012 (orange) seasons. Groundwater discharges to streamflow were
this variability, an in-depth uncertainty analysis described in determined using the constant groundwater endmember. Overlapped
Supplementary Section 4.3 reveals that significant differences shading in 2012 depicts the additional contribution determined from the
between watersheds are still observed by the end of July sensitivity analysis and the time-varying groundwater endmember. Total
(Fig. 3d and Supplementary Fig. 14a), when the signal from annual flow partitioning indicates increased groundwater discharge to
transpiration may be expected to increase relative to that from streams in 2012 despite higher total flows in 1994. Column spacing is based
snowmelt runoff. on stream sampling frequency.

NATURE CLIMATE CHANGE | VOL 4 | JUNE 2014 | www.nature.com/natureclimatechange 483

2014 Macmillan Publishers Limited. All rights reserved.

LETTERS NATURE CLIMATE CHANGE DOI: 10.1038/NCLIMATE2198

a b

n
T Temporal control

) riso
C Constant EM

f. 3 pa
1.5 1.5 S Spatial control

(re com
T Model grey phase

ux
Model red phase

p fl
T T

Sa
C T
Flux change (mm d1)

Flux change (mm d1)

1.0 1.0
C
Temporal control T C C T
C S
S
Temporal control
(Constant EM) C
S T
Spatial control S
0.5 0.5 S
ison S
C T
ar S
mp C
S TS S
co ) C
D IS ef. 9 T
O (r
M C

0.0 0.0
0 20 40 60 80 100 Jul. Aug. Sep. Oct.
Percentage of net trees killed in impacted area

Figure 5 | Watershed evapotranspiration changes from increased groundwater contributions in a simplified water budget. a,b, Calculated fluxes from
transpiration loss compared with the net percentage of trees killed within areas impacted by mountain pine beetles (hatching) that accounts for calculated
flux changes (dashed lines), using Sap Flux3 and MODIS (ref. 9) estimates (a); and seasonal flux trends (T, temporal control with time-varying
groundwater endmember; C, temporal control with constant baseflow endmember (EM); S, spatial control with time-varying groundwater endmembers)
compared with hillslope-scale models28 (b). a assumes 1,000 trees ha1 (ref. 30; grey shading indicates 5%) tree density. In b modelling assumes all
trees are either red or grey, bounding the shaded region.

Temporally, our data reveal greater fractions of groundwater
in streams after infestation. As seen in Fig. 4, the fractional
contribution of groundwater to streamflow translates to consistently
higher groundwater-generated stream fluxes, despite interannual
differences in snowfall and consequently higher stream discharge
in 1994 than 2012. On average, groundwater fractions remain
higher after infestation, even when bounding the analysis with the
simplifying assumption that interannual differences in snow water
equivalent translate directly to less streamflow and higher fractional
groundwater contributions compared with 1994 (Supplementary
Section 4.3). Spatially, the late-summer groundwater inputs increase
nonlinearly as the area infested by MPB increases (Fig. 3). In
addition to affecting a larger area in the Big Thompson watershed,
the infestation is more recent, and may still include trees in the
red and grey stages. The North Inlet drainage was impacted earlier
and is experiencing regrowth (Supplementary Fig. 16). Young pines
may exhibit higher rates of evapotranspiration than older, larger,
pines that are limited by hydraulic conductance24 and preferentially
killed by the MPB. In a nearby forest, new recruits (< 3 years old)
were found on half of the unharvested experimental plots 7 years
after the onset of infestation with a greater density of seedlings
than killed trees25 . This potential for a compensatory response
in the North Inlet from remaining and regenerating vegetation
partially offsets the loss of transpiration from the dead pines and
may explain the observed nonlinear relationship between impacted
area and fractional change in groundwater contribution. Similarly,
insignificant increases in streamwater nitrogen have been attributed
to regeneration in impacted watersheds26 . The inherent challenges
of paired watershed approaches, including spatial variability of
forest structure and shallow and deep subsurface connectivity, may
also contribute to the disproportionate effects on groundwater
between these watersheds. Whereas young lodgepole pines may
form taproots27 , understorey growth and seedlings have shallower
root systems that are unable to take up water from the same depth as
mature trees. These differences cause variation in water use within
and among stands of trees of different ages or species composition,
leading to complex responses in subsurface contributions that are
not distinguishable in this model.
The magnitude and consistency of increased groundwater
contributions to streamflow point to a watershed disturbance
responsible for changing streamflow generation processes that
outweighs interannual climate variability and spatial heterogeneity.
The difference between the fractions of streamflow generated from
groundwater in 1994 and 2012 ranges from 0.200.45 across the
sampling period. This difference corresponds to 0.050.45 m3 s1 of
additional flow attributed to groundwater or an additional flux of
0.181.28 mm d1 across the MPB-impacted area, when considering
the Big Thompson watershed in a simplified water budget. Similar
spatial comparisons between the Big Thompson and North Inlet
indicate that the increase in the groundwater fraction to the Big
Thompson is between 0.0620.52, corresponding to an additional
0.280.84 mm d1 .
The magnitude of these increases is comparable to expected
transpiration losses calculated using different approximations and
upscaling methods of evapotranspiration. Sap flux measurements
indicate a single lodgepole pine (average diameter = 24 cm)
transpires a relatively constant 16 l d1 in late summer3 . Uniformly
upscaling these sap flux measurements to the watershed scale
suggests 3054% of trees within the impacted area would need to
be killed to achieve the mean observed additional groundwater flux
to the stream (Fig. 5a). Aerial survey data indicate 245% of trees are
killed annually in the MPB-affected areas across the Big Thompson
watershed2 . Potometers estimate a mature lodgepole stand can
transpire 3.4 mm d1 (ref. 27), suggesting up to 41% of trees in
the impacted area are killed. Evapotranspiration estimates based
on Moderate Resolution Imaging Spectroradiometer (MODIS)
data9 suggest that 4580% of trees within the impacted area have
been killed (Fig. 5a), although the offsetting effects of increased
evaporation may cause overestimation of tree kill (Fig. 5a). The
seasonal trends in the increased groundwater fluxes are consistent

484 NATURE CLIMATE CHANGE | VOL 4 | JUNE 2014 | www.nature.com/natureclimatechange

2014 Macmillan Publishers Limited. All rights reserved.

NATURE CLIMATE CHANGE DOI: 10.1038/NCLIMATE2198
with hillslope-scale integrated model results of evapotranspiration
changes in beetle-impacted forests28 (Fig. 5b). In August and
September, fluxes from the spatial and temporal comparisons begin
to converge and show consistent decreasing trends, suggesting that
the effects of MPB-induced mortality on transpiration in the late
summer exceed those related to meteorological and geographical
differences. The observed decreasing trends in the differential fluxes
are probably due to an evapotranspiration decrease in the fall, as
energy limitations become influential28 .
Here, we connect climate-exacerbated insect infestation and
the subsequent watershed-scale transpiration loss to late-summer
streamflow generation processes. Both the paired watershed and
temporal comparisons indicate the potential for increased ground-
water contributions to streamflow after infestation. Our combined
approach provides spatial and temporal controls on inherently chal-
lenging field heterogeneities that may be improved only by numer-
ical modelling of flow paths in impacted watersheds. In RMNP,
new regeneration and continued growth of the remaining vegetation
seem to offset this loss of transpiration within approximately 8
years after the onset of infestation. Ultimately, understanding these
changes in streamflow generation provides needed insight for water
resource management in MPB-infested watersheds and for changing
forested landscapes throughout the region.
Methods
MPB-impacted area was quantified from US Forest Service aerial survey data2
and includes all species killed by the MPB (that is, lodgepole pine, ponderosa
pine and limber pine). Total area affected was determined by summing the
annually impacted areas through 2012 and omitting overlapping area. We were
not able to evaluate the distribution of sizes, ages and species of pines and other
forest vegetation in the two watersheds. Precipitation, snow, groundwater and
stream water samples were collected during the late summer of 2012 throughout
RMNP (Fig. 2). Precipitation isotope samples were collected in polycarbonate
rain gauges, using mineral oil to prevent evaporation. Sampling occurred weekly
to biweekly from July to October, depending on rain events. Precipitation
chemistry was available through the National Atmospheric Deposition Program
and National Trends Network using weekly averaged data at the Beaver Meadows
site. Snow isotopic compositions were obtained from a snow pit sampled near
peak snow accumulation in early April 2012. The site was selected to be outside
the drip line of surrounding trees, and protected from radiation and wind
exposure. Samples were taken every 10 cm using a snow density cutter and the
average snow pack isotopic composition was used for subsequent analysis. Bulk
snow chemistry was provided at the site through the US Geological Survey
(USGS) Rocky Mountain Regional Snowpack Chemistry Monitoring Study
(available at http://co.water.usgs.gov/projects/RM_snowpack/html/data.html).
Shallow (1 m deep) groundwater wells from a wetland and riparian monitoring
study that maintained flow through the season were selected to characterize the
shallow groundwater contribution to adjacent streams. Samples were collected
approximately biweekly from July to the end of October. Isotopic composition of
the shallow groundwater near North Inlet was compared to a nearby water well
and found to be indistinguishable, suggesting good agreement with the
groundwater fingerprint, despite possible hyporheic zone mixing. Stream water
isotope and chemistry samples were collected weekly to biweekly and were
analysed using standard USGS methods29 . Samples from the 2012 season were
analysed for 18 O at the Colorado School of Mines stable isotope laboratory.
Stream and snow samples were analysed at the USGS research laboratory in
Boulder, Colorado. Shallow groundwater chemistry was analysed at Colorado
School of Mines in Golden, Colorado.
The three-component hydrograph separation utilizes 18 O compositions and
electrical conductivity and assumes streamflow contributions from rain, snow and
groundwater, consistent with ref. 12. The resulting set of equations used to
describe endmember contributions to streamflow is:
Qs ECs = Qr ECr + Qn ECn + Qg ECg (1)
Qs 18 Os = Qr 18 Or + Qn 18 On + Qg 18 Og (2) 11. Stednick, J. D. Monitoring the effects of timber harvest on annual water yield.
Qs = Qr + Qn + Qg
where Q denotes flow, EC is electrical conductivity (S cm1 ), 18 O is stable
isotope composition (h), and the subscripts s, r, n and g, represent stream, rain,
NATURE CLIMATE CHANGE | VOL 4 | JUNE 2014 | www.nature.com/natureclimatechange
2014 Macmillan Publishers Limited. All rights reserved.
LETTERS
snow and groundwater respectively. Using the mass balance based on flow, the
equations can be rearranged to calculate the fraction of stream water that each
endmember contributes to streamflow. This approach also facilitates comparisons
between watersheds and seasons, inherently accounting for differences in total
flow. Full EMMA (ref. 15) confirmed that three endmembers are appropriate to
describe the variability in the 2012 Big Thompson stream water chemistry.
Hydrograph separations performed using the EMMA projections are comparable
to those using 18 O and EC to define the endmembers. The reduction of required
chemistry parameters is useful to compare different data sets through time and
space (Supplementary Section 2).
Spatially, two watersheds, as described previously, provide information on the
response of stream contributions based on the timing and extent of outbreak.
Temporally, the data collected in this study were compared with hydrograph
separations performed using data from a previous study in the Big Thompson
watershed in 1994, before infestation12 . The primary methodological difference
between the 1994 study and this study is the use of a constant groundwater
endmember in 1994, determined on the basis of pre-melt baseflow stream
concentrations. Although nearby springs agreed well with baseflow
concentrations for small watersheds, the use of this method may underestimate
groundwater contributions, particularly during snowmelt12 . For comparison, the
temporal analysis was repeated with a constant baseflow endmember for the 2012
season (Supplementary Section 4.3). National Atmospheric Deposition Program
data were used to provide 1994 precipitation electrical conductivity data and to
be consistent with the 2012 analyses. All other data were taken from the
published study12 . Uncertainty analysis was performed using estimates of
uncertainty and variability for each tracer/endmember combination and
propagating that error through the hydrograph separation equations using
first-order Taylor expansion (Supplementary Section 4.3).
The differences in the subsurface-derived fraction from the spatial and
temporal analyses were multiplied by total streamflow measured at the time of
sampling to estimate an increased groundwater flow to the stream. The change in
flow was distributed over the MPB-impacted area in the Big Thompson watershed
to estimate a flux. The flux was used to compare the increased groundwater
contributions to traditional estimates of evapotranspiration. This simplified water
budget approach assumes a total tree density of 1,000 trees ha1 , consistent with
previous observations near Cub Lake in the Big Thompson watershed30 .
Received 22 November 2013; accepted 17 March 2014;
published online 20 April 2014
References
1. Anderegg, W. R. L., Kane, J. M. & Anderegg, L. D. L. Consequences of
widespread tree mortality triggered by drought and temperature stress. Nature
Clim. Change 3, 3036 (2012).
2. Colorado State Forest Service, 2012 Report on the Health of Colorados Forests
(CSFS, 2013).
3. Hubbard, R. M., Rhoades, C. C., Elder, K. & Negron, J. Changes in
transpiration and foliage growth in lodgepole pine trees following mountain
pine beetle attack and mechanical girdling. For. Ecol. Manage. 289,
312317 (2013).
4. Wulder, M. A., Dymond, C. C., White, J. C., Leckie, D. G. & Carroll, A. L.
Surveying mountain pine beetle damage of forests: A review of remote sensing
opportunities. For. Ecol. Manage. 221, 2741 (2006).
5. Biederman, J. A. et al. Multiscale observations of snow accumulation and peak
snowpack following widespread, insect-induced lodgepole pine mortality.
Ecohydrology 7, 150162 (2014).
6. Mikkelson, K. M. et al. Bark beetle infestation impacts on nutrient cycling,
water quality and interdependent hydrological effects. Biogeochemistry 115,
121 (2013).
7. Clow, D. W., Rhoades, C., Briggs, J., Caldwell, M. & Lewis Jr, W. M. Responses
of soil and water chemistry to mountain pine beetle induced tree mortality in
Grand County, Colorado, USA. Appl. Geochem. 26, S174S178 (2011).
8. Brown, M. G. et al. Evapotranspiration and canopy characteristics of two
lodgepole pine stands following mountain pine beetle attack. Hydrol. Process.
http://dx.doi.org/10.1002/hyp.9870 (2013).
9. Maness, H., Kushner, P. J. & Fung, I. Summertime climate response to mountain
pine beetle disturbance in British Columbia. Nature Geosci. 6, 6570 (2013).
10. Pypker, T. G. et al. Can carbon isotopes be used to predict watershed-scale
transpiration? Wat. Resour. Res. 45, W00D35 (2009).
J. Hydrol. 176, 7995 (1996).
12. Sueker, J. K., Ryan, J. N., Kendall, C. & Jarrett, R. D. Determination of
(3)
hydrologic pathways during snowmelt for alpine/subalpine basins, Rocky
Mountain National Park, Colorado. Wat. Resour. Res. 36, 6375 (2000).
13. Clow, D. W. et al. Ground water occurrence and contributions to streamflow in
an alpine catchment, Colorado Front Range. Ground Wat. 41, 937950 (2003).
485
LETTERS
14. Mikkelson, K. M., Dickenson, E. R. V., Maxwell, R. M., McCray, J. E. &
Sharp, J. O. Water-quality impacts from climate-induced forest die-off. Nature
Clim. Change 3, 218222 (2013).
15. Hooper, R. P. Diagnostic tools for mixing models of stream water chemistry.
Wat. Resour. Res 39, 10551067 (2003).
16. Tetzlaff, D., Malcolm, I. A. & Soulsby, C. Influence of forestry, environmental
change and climatic variability on the hydrology, hydrochemistry and
residence times of upland catchments. J. Hydrol. 346, 93111 (2007).
17. Jung, H., Hogue, T. & Rademacher, L. Impact of wildfire on source water
contributions in Devil Creek, CA: Evidence from end-member mixing analysis.
Hydrol. Process. 23, 183200 (2009).
18. Adams, H. D. et al. Ecohydrological consequences of drought- and infestation-
triggered tree die-off: insights and hypotheses. Ecohydrology 5, 145159 (2012).
19. Pugh, E. & Gordon, E. A conceptual model of water yield effects from
beetle-induced tree death in snow-dominated lodgepole pine forests. Hydrol.
Process. 27, 20482060 (2013).
20. Beudert, B., Klcking, B. & Schwarze, R. Grosse Ohe: Impact of bark beetle
infestation on the water and matter budget of a forested catchment. For. Hydrol.
Res. Ger. Russ. 4163 (2007).
21. Braddock, W. & Cole, J. Geologic Map of Rocky Mountain National Park and
Vicinity, Colorado (US Geol. Surv. Map I-1973, 1990).
22. Rocky Mountain National Park Vegetation Project Field Plot Locations (US Natl
Park Serv., 2006).
23. National Operational Hydrologic Remote Sensing Centre, Snow Data Assim.
Syst. (SNODAS)Data Prod. NSIDC, [April 1, 2012](National Snow and Ice Data
Center, 2004).
24. Hubbard, R. M., Bond, B. J. & Ryan, M. G. Evidence that hydraulic
conductance limits photosynthesis in old Pinus ponderosa trees. Tree Physiol.
19, 165172 (1999).
25. Collins, B. J., Rhoades, C. C., Hubbard, R. M. & Battaglia, M. A. Tree
regeneration and future stand development after bark beetle infestation and
harvesting in Colorado lodgepole pine stands. For. Ecol. Manage. 261,
21682175 (2011).
486
2014 Macmillan Publishers Limited. All rights reserved.
NATURE CLIMATE CHANGE DOI: 10.1038/NCLIMATE2198
26. Rhoades, C. C. et al. Biogeochemistry of beetle-killed forests: Explaining a
weak nitrate response. Proc. Natl Acad. Sci. USA 110, 175660 (2013).
27. Knight, D. H., Fahey, T. J. & Running, S. W. Water and nutrient outflow
from contrasting lodgepole pine forests in Wyoming. Ecol. Monogr. 55,
2948 (1985).
28. Mikkelson, K. M. et al. Mountain pine beetle infestation impacts: Modeling
water and energy budgets at the hill-slope scale. Ecohydrology 6, 6472 (2013).
29. Fishman, M. J. Methods of analysis by the US Geological Survey National Water
Quality Laboratory: Determination of Inorganic and Organic Constituents in
Water and Fluvial Sediments, vol. 93125 (US Geol. Surv., 1993).
30. Sibold, J. S. et al. Influences of secondary disturbances on lodgepole pine stand
development in Rocky Mountain National Park. Ecol. Appl. 17,
16381655 (2007).
Acknowledgements
This material is based on work supported by the National Science Foundation
under Grant No. WSC-1204787, the USGS-National Institute of Water Resources
under Grant No. 2011CO245G Subaward G-2914-1 and through the Water, Energy,
and Biogeochemical Budgets program and USGS/NPS Partnership program.
Author contributions
L.A.B. and R.M.M. conceived the study, L.A.B. and D.W.C. collected and analysed
the data, and L.A.B., R.M.M., D.W.C. and J.E.M. interpreted results and contributed
to writing.
Additional information
Supplementary information is available in the online version of the paper. Reprints and
permissions information is available online at www.nature.com/reprints.
Correspondence and requests for materials should be addressed to L.A.B. or R.M.M.
Competing financial interests
The authors declare no competing financial interests.
NATURE CLIMATE CHANGE | VOL 4 | JUNE 2014 | www.nature.com/natureclimatechange