HYDROLOGICAL PROCESSES

Hydrol. Process. 29, 4068–4082 (2015)

Published online 1 April 2015 in Wiley Online Library
(wileyonlinelibrary.com) DOI: 10.1002/hyp.10463

Stemflow and throughfall contributions to soil water recharge

under trees with differing branch architectures
R. Bialkowski1 and J. M. Buttle2*
1
Environmental and Life Sciences Graduate Program, Trent University, Peterborough, ON, Canada, K9J 7B8
2
Department of Geography, Trent University, Peterborough, ON, Canada, K9J 7B8

Abstract:
Trees whose branches slope towards or away from the bole differ in the spatial pattern of water delivery to the soil surface
beneath their canopies. This study examines the implications of these differences for soil water recharge below 1 m depth in a
managed forest on the Oak Ridges Moraine in southern Ontario. Throughfall and vertical profiles of soil water content at varying
distances from the boles of a sugar maple and two red pines of differing ages (36 and 57 years old) were measured during rainfall
inputs in the summer of 2009 and in the spring/early summer of 2010. Stemflow fluxes were estimated using measurements on
nearby trees with similar morphologies. A water balance approach was used to estimate recharge at each distance from the bole,
as well as for the entire sampled area beneath each tree’s canopy. Throughfall was more than 1.3 times above-canopy rainfall
within 0.5 m of the bole of the sugar maple, and stemflow volumes were at least an order-of-magnitude greater than those for the
pines in both 2009 and 2010. These focused inputs resulted in greater near-bole soil water contents and soil water recharge in
excess of above-canopy rainfall within 0.3 m of the sugar maple bole. In contrast, both young and old pines showed no
significant trends in throughfall with distance from the bole, and minor stemflow volumes made insignificant contributions to
recharge near the pine boles. Recharge was significantly greater than 0 beneath the canopy of all three trees in both years and was
generally similar between trees. The exception was recharge beneath the sugar maple in 2010, which exceeded that of the young
red pine as a result of the former’s large stemflow fluxes. Results suggest the transition from red pine plantations to mixed
hardwood-conifer stands will not alter total soil water recharge during spring and summer rainfall inputs; nevertheless, it will
generate marked changes in the spatial pattern of sub-canopy recharge in this forest landscape on the Oak Ridges Moraine.
Copyright © 2015 John Wiley & Sons, Ltd.

KEY WORDS throughfall; stemflow; soil water content; recharge; red pine; sugar maple; water balance
Received 28 August 2014; Accepted 16 February 2015

INTRODUCTION focussing towards the bole (Sato et al., 2011). Species

whose branches slope away from the bole particularly in the
Precipitation reaches the forest floor via diffuse
lower canopy (common for conifers – Steinbuck, 2002)
(throughfall, TF) and point (stemflow, SF) inputs.
focus TF towards the edge of the tree crown (Johnson,
Throughfall can be highly heterogeneous beneath a single
1990). The spatial distribution of TF may also change as
tree (Johnson, 1990; Sato et al., 2011), partly as a result
trees age. Thus, TF was greatest close to tree stems in a
of variations in canopy coverage and branch architecture
young conifer stand, with little or no TF close to the bole and
(Herwitz, 1987; Johnson, 1990; Staelens et al., 2006).
no relationship between total TF and distance from the bole
Studies report increasing (Ford and Deans 1978; Herwitz,
for older conifers (Keim et al., 2006).
1987), decreasing (Johnson, 1990; Beier et al., 1993;
The amount of SF generated by a tree is controlled by
Whelan et al., 1998) and inconsistent (Loustau et al., 1992;
several factors, such as projected crown area, bark
Keim et al., 2006) relationships between TF and proximity
smoothness and storage capacity, and branch orientation
to the tree bole, which may partly reflect inter-species
and inclination. Greater projected crown area is associated
differences in branch geometry. Species whose branches
with increased interception and, thus, with SF production
slope towards the bole (common for deciduous trees –
(Ford and Deans, 1978). Species with smoother bark have
Steinbuck, 2002) promote water movement (Carlyle-Moses
reduced bark storage capacity and resistance to flow and,
and Price, 2006; Alexander and Arthur, 2010) and TF
therefore, produce greater quantities of SF (Alexander
and Arthur, 2010). Branching systems which slope
towards the bole funnel water towards the stem,
*Correspondence to: J M Buttle, Department of Geography, Trent
University, Peterborough, ON, Canada, K9J 7B8. increasing potential for SF generation (Herwitz, 1986;
E-mail: jbuttle@trentu.ca Carlyle-Moses and Price, 2006; Alexander and Arthur,

Copyright © 2015 John Wiley & Sons, Ltd.

 RECHARGE UNDER DIFFERING TREE BRANCH ARCHITECTURES
2010; Xiao and McPherson, 2011). Conversely, de-
creased SF may accompany the tendency for branches
to sag away from the stem as many coniferous species age
(Voigt, 1960; Johnson, 1990; Steinbuck, 2002).
Soil water balance studies in forest ecosystems may
assume spatially-uniform TF inputs (e.g. Buttle et al.,
2014) and can ignore SF completely (e.g. Dripps and
Bradbury, 2010) because of its generally small fraction of
above-canopy precipitation (Liang et al., 2009). Never-
theless, spatial variability in TF results in uneven soil
wetting and heterogeneous soil moisture distributions and
soil water recharge (R) in forests (Guswa and Spence,
2012). This heterogeneity may be more pronounced in the
case of SF, which infiltrates a relatively narrow ring of
soil surrounding the bole (Buttle et al., 2014). Inputs may
subsequently follow preferential pathways created by tree
roots, resulting in heterogeneous soil moisture patterns
and infiltration dynamics (Durocher, 1990; Taniguchi
et al., 1996; Johnson and Lehmann, 2006; Liang et al.,
2009; Germer, 2013). Macropores can quickly channel
water deep into the soil, creating the opportunity for SF to
be an important contributor to R in forested ecosystems
(Chang and Matzner, 2000; Johnson and Lehmann, 2006;
Liang et al., 2009), as well as an important water resource
for tree growth (McKee and Carlyle-Moses, 2010) and
nutrient cycling (Levia and Herwitz, 2000; Alexander and
Arthur, 2010).
The objective of this study was to examine the
influence of differences in branch geometry on TF, SF
Figure 1. The Oak Ridges Moraine, the central and western portions of the Ganaraska Forest and locations of the study stands (see Table I for details)
and rain gauges used to measure above-canopy rainfall (Pg). Air temperature was measured at the open site indicated with *
Copyright © 2015 John Wiley & Sons, Ltd.
4069
and R at the individual tree scale during rainfall. A mature
sugar maple (Acer saccharum – branches slope towards
the bole) is compared with two red pines (Pinus
resinosa – lower branches slope away from the bole) of
different ages, and the following hypotheses were tested:
1. TF and R will increase with proximity to the bole of the
sugar maple and SF will be maximized as a result of its
branch architecture.
2. TF and R will not vary with proximity to the boles of
the pines and SF will be minimized as a result of their
branch architecture.
3. Contributions of SF to R will be less in the older
relative to the younger pine, as branches increasingly
slope away from the bole with greater tree age.
4. The importance of spatially-focussed SF and TF inputs
to R will decrease in the order: sugar maple → younger
red pine → older red pine.
STUDY AREA
This study was conducted in the Ganaraska Forest (GF)
on the Oak Ridges Moraine (ORM) in southern Ontario
(44°05′N, 78°30′W; Figure 1). The ORM is an important
hydrogeological feature composed of a complex strati-
graphic sequence of Quaternary sediments deposited
during a series of glacial and interglacial periods (Gerber
and Howard, 2002). The water table along the crest of the
Hydrol. Process. 29, 4068–4082 (2015)
4070 R. BIALKOWSKI AND J. M. BUTTLE

beech (%)
American
moraine is 35–50 m below the ground surface (Funk,

9.7
1977). Topography in the GF is hummocky (maximum

0
0
elevation of 384 m asl), consisting of sand and gravel hills
and high ridges that comprised interlayered gravels,

maple (%)
sands, silts, clays and minor diamictons (Buttle and

Sugar

48.4
Farnsworth, 2012). The GF has a humid mid-latitude

0
0
climate (K ppen Dfb). Annual average precipitation at
Peterborough airport ~20 km northeast of the GF is
840 mm, 19% of which is snow (Environment Canada,

oak (%)
White

35.5
2011). Mean daily air temperatures in January and July

0
0
are
7.2 and 20.5 °C, respectively (Buttle et al., 2014),
while mean annual evapotranspiration is 540 mm (Brown

pine (%)
et al., 1968). The sandy to sandy loam soils are brunisolic

6.4
Red

100
100
grey brown luvisols, belonging to the Pontypool sand and

Minimum and maximum leaf area index and mean canopy density ranges for the mixed hardwood stand represent the leaf off and leaf on extremes.
Pontypool gravelly sand series [see Hoffman and Acton
(1974) and Greenwood and Buttle (2014) for more

Mean canopy
detailed descriptions].

density (%)

23–91
68
82
The GF is a mosaic of mixed hardwood-conifer stands
and red and white pine (Pinus Strobus) plantations.
Approximately 42% of the GF was covered by deciduous
forest before reforestation [mainly red oak (Quercus
rubra L.), with smaller amounts of white birch (Betula

index (m2 m
2)

Table I. Study stand characteristics

0.28 – 2.64
papyrifera Marsh.), sugar maple (Acer saccharum

Leaf area
Marsh.) and other hardwoods] (Buttle and Farnsworth,

0.84
1.72
2012). Reforestation began on a large scale in 1948 to
control soil erosion in gullies and prevent wind blowouts
of sandy soils. Plantations consist of purely red and white
pine, as well as combinations of the two, with red pine
Basal area
(m2 ha
1)

32.2
26.9
37.5
being the main species planted (60% of plantations as of
1998). Planting density was ~2100 trees per hectare. Red
pine stands are thinned when basal areas reach ~28–32 m
2
ha
1 (~every 10–12 years); one-third of the trees are
Trees ha
1

removed to achieve a basal area of approximately

775

450
1550

20–22 m2 ha
1. Live and dead branches are pruned up

to 7.4 m height to produce knot-free sawlogs. Thus, there
is no progressive increase in leaf area index and basal area
Mean height

with age in the red pine plantations (Buttle, 2011), which

(±SD) (m)

20.3 ± 4.9
15.0 ± 0.7
26.5 ± 0.4

are characterized by alternating rows of trees and skidder

trails used during forest management. Skidder trails
produce large openings in young red pine stands, which
close with canopy expansion as stands age. Red pine
stands are managed to encourage a transition to a mixed
Age in 2009
(years)

hardwood-conifer forest.
92
36
57

METHODS
Mixed hardwood-conifer

Three stands representing the forest management gradient

in the GF (young red pine → old red pine → mature mixed
hardwood-conifer) were selected (Table I). A level site in
each stand underlain by either Pontypool sand or
Young pine
Study stand

Old pine

Pontypool gravely sand was selected. Areas with minimal

undergrowth were chosen to limit its effects on
interception and increased soil permeability associated

Copyright © 2015 John Wiley & Sons, Ltd. Hydrol. Process. 29, 4068–4082 (2015)
 RECHARGE UNDER DIFFERING TREE BRANCH ARCHITECTURES 4071

with its root networks. Stands were characterized in terms St
S0

of species, basal area, height, density, leaf area index and
canopy density (Buttle and Farnsworth, 2012). Vertical
profiles of bulk density, porosity and soil texture were
determined at the mixed hardwood stand from horizontal
cores (6 cm long, 4.8 cm i.d.), taken from the face of a
1.5 m deep pit dug in the centre of the level site at 0.1 m
intervals beginning at 0.1 m depth to 1 m depth. These
profiles were taken from a vertical core extracted to 1 m
depth at the centre of the pine stands and sectioned every
0.1 m. Analytical methods are described in Buttle et al.
(2014). A single ring infiltrometer and a Guelph
Permeameter (Reynolds and Elrick, 1985) were used to
determine saturated hydraulic conductivity (KH) at the
surface and at depth (every 0.1 m to a depth of 0.8 m),
respectively.
One tree in each stand was used to assess variability in
water inputs, soil water storage and R below 1 m depth
with proximity to the tree bole. A sugar maple was
selected at the mixed hardwood-conifer stand because this
was the most abundant species (Table I). Characteristics
of the selected trees are presented in Table II. A Delta T
PR2/6 Profile ProbeTM was used to measure soil water
content (SWC) at depths of 0.1, 0.2, 0.3, 0.4, 0.6 and
1.0 m. The probe was calibrated as described in Buttle
et al. (2014). Each tree was instrumented with a transect
of six ATL-1 access tubes (1154 mm × 28 mm i.d.),
installed at varying distances from the bole using a PR-
AUG4 auger (Table II). Access tubes were spaced at least
0.2 m apart to avoid sampling overlapping volumes of
soil. Temporal changes in soil water storage in the upper
1.0 m of soil (ΔSt, mm) were determined as:
ΔSt ¼ St
S0
where S is water depth in the upper 1.0 m of soil (mm), St
is depth on the day of measurement and S0 is the initial
depth (23 June 2009 or 4 May 2010). Soil water storage
was calculated as:
S ¼ ∑m
i¼1 ðSWC i zi Þ
where SWCi is soil water content (m3 m
3) at the
midpoint of soil layer i with a thickness zi (mm). Values
of ΔSt significantly different from 0 (p = 0.05) were
determined using Z-scores:
Z ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi (3)
σSt 2 þ σS0 2
where σS is the error in S at the respective time:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 2
σS ¼ ∑m i¼1 σ i (4)
and
σ i ¼ ðSWC i zi Þ0:15 (5)
where 0.15 is the average range in SWC about the
calibration relationship given in Buttle et al. (2014). Non-
significant ΔSt values from Equation 3 were set to 0;
otherwise, ΔSt was determined from Equation 1.
Above-canopy precipitation (Pg) was measured using
standard rain gauges at three open sites (Buttle and
Farnsworth, 2012, Figure 1) and at the Meteorological
Service of Canada station at Tapley (ID 6168525, 44°10′N,
78°30′W, 270 m asl, 7.7 km north of the GF). Estimated Pg
for a stand was taken from the nearest open site (<2 km
distant). TF was measured within 0.3 m of each access tube
with a rain gauge placed at corresponding distances from
the tree bole. Johnson (1990) found that TF did not vary
with direction for a given distance from a bole, so direction
of the sampling transect from the bole was randomized
between trees. SF was measured at four trees in each study
stand (Table III) using clear tygon tubing (25.4 mm i.d.) slit
longitudinally, stapled to the bole in a spiral beginning
~1.5 m above the ground surface and sealed to the bole with
silicone caulking. Collars drained to a collector; no
overtopping of collars or flow beneath collars via the tree
bark was observed during events.
(1) Canopy interception (I) was estimated as:
 
I ¼ Pg
TF w þ SF (6)
where TF w is the weighted mean TF and SF is mean SF
for the stand. TF w was calculated as:
  
(2) n
TF w ¼ ∑i¼1 TF i
Ai
(7)
Atotal
where TFi is TF for distance interval i, Ai is the area of
distance interval i and Atotal is the total area assessed. Ai
was calculated as:

Table II. Characteristics of trees used to estimate soil water recharge

Bole diameter at Bole diameter at Height Mean canopy Projected canopy Distances from tree bole to
Tree breast height (m) ground level (m) (m) radius (m) area (m2) PROFILE PROBE access tubes (m)

Sugar maple 0.38 0.48 23 5.04 80 0.10, 0.30, 0.50, 0.86, 1.10, 2.13
Young pine 0.14 0.17 15 1.26 5 0.17, 0.37, 0.57, 0.87, 1.07, 2.00
Old pine 0.37 0.46 26.5 3.04 29 0.16, 0.36, 0.56, 0.86, 1.06, 2.06

Copyright © 2015 John Wiley & Sons, Ltd. Hydrol. Process. 29, 4068–4082 (2015)
4072 R. BIALKOWSKI AND J. M. BUTTLE

Table III. Characteristics of trees used to monitor SF in the three stands

Mixed hardwood - conifer Young pine Old pine

2 2
Tree DBH (m) Height (m) PCA (m ) DBH (m) Height (m) PCA (m ) DBH (m) Height (m) PCA (m2)

1 0.48* 26 163.9 0.14 15 3.0 0.25 26 18.3

2 0.22* 22 40.7 0.13 16 2.8 0.29 27 20.3
3 0.35* 25 60.6 0.08 15 3.2 0.36 26 22.2
4 0.09 16 8.2 0.20 14 7.2 0.43 26.5 39.2

All monitored trees in the mixed hardwood-conifer stand were sugar maples, while all trees monitored at the pine stands were red pines. DBH is diameter
at breast height and PCA is projected canopy area. The bolded tree in the mixed hardwood-conifer stand was used to derive SF inputs for the sugar maple
used to estimate soil water recharge (Table II). Trees marked with asterisks were used to assess the error in SF input for the sugar maple used to estimate
soil water recharge

area (PCA) (Table III). SF for young and old pines was
Ai ¼ πr 2o
πr2i (8) the average of the four trees monitored in each stand.
Canopy-scale ET in 2009 was estimated from the
where ro and ri are the radii of the outer and inner edges
MOD16 global evapotranspiration product (Mu et al.,
of the distance interval, respectively. ri equalled the edge
2011), which combines remotely sensed and modelled
of the tree bole for the innermost distance interval at all
inputs to estimate spatially-distributed 8-day ET at 1 km2
sites, while ro of the outer most distance interval was
resolution using the Penman–Monteith equation. The
equated to the canopy edge for the sugar maple and the
MOD16 grid was overlain on LANDSAT 7 imagery to
old pine and to the distance to the outer-most TF gauge
identify MOD16 pixels that best represented each land
for the young pine. Volume of SF was converted to a
cover type in the GF. Only those pixels composed
depth by dividing by projected crown area for use in
predominantly of a single land cover type were selected to
Equation 6.
minimize convolution resulting from mixed pixels. One
Measurements of Pg, SF, TF and SWC were made nine
pixel was suitable for mixed hardwood-conifer ET
times between Day of Year (DOY) 174 (23 June) and
estimates. Nine MOD16 pixels covering a 3 km × 3 km
DOY 244 (1 September) in 2009 and five times between
area and dominated by red pine were examined. It was not
DOY 124 (4 May) and DOY 193 (12 July) in 2010. The
possible to differentiate between young red pine, old red
2009 study period was during the leaf-on period, while
pine and white pine stands, so pixels likely contained a
the transitional leaf-out period was included in 2010.
mix of the three types. Pixels with a total ET in 2009
Soil water recharge below 1 m depth (R, mm) was
more than one standard deviation above the mean were
calculated for six concentric rings around each bole. The
excluded as outliers, and eight-day ET values for the
boundary of each ring was the midway point between
remaining pixels were averaged to estimate ET for the
two successive access tubes. The bole was used as the
young and old pines. Mixed-hardwood and red pine ET
inner edge of the ring for the access tube closest to the
estimates were divided by eight to obtain daily ET.
bole, and the outer edge of the ring for the outermost
Canopy I was subtracted from MOD16 ET to estimate
access tube was at the access tube itself. Cumulative
direct soil evaporation plus transpiration used in
recharge at time t (Rt) at the access tube closest to each
Equations 9 and 10. MOD16 data were unavailable for
bole was calculated as:
2010 and were derived by regressing 2009 MOD16
 
Rt ¼ ∑nt¼0 ½TF t þ SF t
fET t
I t g
ΔSt (9) estimates against potential ET from the temperature-
based Hamon model (Dingman 2002; Table IV).
and as follows at the other access tubes: Temperature data for ET estimation were obtained from
  a BML-TS-7 Thermilinear Air Temperature Sensor at
Rt ¼ ∑nt¼0 ½TF t
fET t
I t g
ΔSt (10) one of the open sites (Figure 1). Estimated ET was
adjusted by I as for 2009.
where ET is evapotranspiration (mm) and TFt, SFt, ETt,
Total R (Rtotal) beneath each tree was calculated as:
It and ΔSt are cumulative values to time t.
All SF was assumed to reach the soil at the tree base   
n Ai
and was input to the inner-most ring only. SF volume was Rtotal ¼ ∑i¼1 Ri (11)
Atotal
converted to a depth (mm) by dividing by the area of the
innermost ring. The SF input at the sugar maple was where Ri is R for distance interval i and Ai is defined in
assumed to be the same as that from a sugar maple with a Equation 8. Error estimation for Rtotal is described in
similar diameter at breast height and projected canopy Appendix A. Rtotal values and associated errors were used

Copyright © 2015 John Wiley & Sons, Ltd. Hydrol. Process. 29, 4068–4082 (2015)
 RECHARGE UNDER DIFFERING TREE BRANCH ARCHITECTURES
Table IV. Best-fit relationships between 8-day evapotranspiration
(ET) values estimated by MOD16 (y) and 8-day potential ET from
the Hamon model (x) for 2009
Forest type Best-fit relationship r2
Mixed hardwood-conifer y =
0.0043x3 + 0.1896x2 0.87
–1.0756x + 1.9171
Red pine y =
0.0059x3 + 0.2481x2 0.91
–1.4247x + 1.8328
in Z-scores (Winkler and Hays, 1975) to assess the
statistical significance of any differences in Rtotal and
Rtotal:Pg, with and without the inclusion of SF inputs for a
given tree, as well as between trees.
The funnelling ratio (Herwitz, 1986) was used to
assess the efficiency of SF production for the different
trees:
SF volume
Funelling ratio ¼ (12)
Pg Basal Area
The funnelling ratio treats each tree as a TF
collector, with a diameter equal to that of the tree
trunk. Greater ratios indicate an increasing percentage
of Pg converted to SF and more efficient generation
of SF.
RESULTS
Soil characteristics
Soil porosity at all three stands was similar to
elsewhere in the GF (Buttle et al., 2014) and showed a
slight decline with depth at all stands (Figure 2a).
Saturated hydraulic conductivities at all stands were
greatest at the surface and exhibited marked variability
with increasing depth (Figure 2b). Soils were dominated
by sand (Figure 2 c–e), and the large KH values at the soil
surface and often at depth in the profile at each stand
(Figure 2b, KH often > 5 × 10
5 m s
1 or 180 mm h
1)
suggest surface runoff is unlikely and all water inputs
would infiltrate into the soil profile.
Above canopy precipitation, throughfall, stemflow and
interception
Total Pg for the 2009 study period ranged from 317 to
382 mm between the stands (Table V). Rainfall in June was
below the 1981–2010 normal for Peterborough Airport
~20 km northeast of the GF, while July was above normal
and August was slightly above normal. Total Pg for the
2010 study period was more similar between the stands and
was below normal in May, well above normal in June and
slightly above normal in July.
The sugar maple had the smallest total weighted mean
TF for both years (Table V) but significantly more SF,
Copyright © 2015 John Wiley & Sons, Ltd.
4073
either as a total volume or a depth. All but two events (of
14 recorded) for both years generated SF for the sugar
maple, and funnelling ratios (Table VI) showed it to be
much more effective at generating SF than either pine in
both years. Mean funnelling ratios >1 for the young pine
for both years indicated greater ability to generate SF and
deliver it to the bole relative to the old pine. SF was
produced less frequently from the young and old pines
(9 and 10 events, respectively), and total volumes and
depths were consistently less than those for the sugar
maple. SF was significantly correlated with Pg across both
years for all three trees (p < 0.05), and linear regression
relationships between SF volume and Pg (not shown)
indicate that the sugar maple produced SF volumes
almost two orders of magnitude greater than those from
the pines for a given increase in Pg. Total I (both as a
depth and fraction of Pg) was greatest for the sugar maple
and least for the old pine for both study periods.
Nevertheless, SF was a minor fraction of Pg for all trees
for both years (Table V).
Below-canopy TF for the sugar maple showed similar
trends with distance from the bole in 2009 and 2010, with
generally greater TF with increasing proximity to the bole
for all events with Pg >14 mm (Figure 3a). In many cases,
TF near the bole exceeded Pg for the stand. This trend
was not apparent for two smaller events (13 July and 19
August 2009) that did not generate SF. Conversely, there
was generally no trend in TF with distance from the bole
for most events for the young and old pines. There was a
slight increase with proximity to the bole for events with
large Pg and measureable SF (Figure 3b and c); however,
this increase was not as pronounced as that for the sugar
maple. TF adjacent to the sugar maple bole usually
exceeded that near the pine boles (Figure 4a); conversely,
TF at ~2 m from the bole was often greater for the pines
relative to the sugar maple (Figure 4b).
Total TF with distance from the bole for 2009 and 2010
for the three trees supported trends for individual events
(Figure 5). There was greater total TF for the two gauges
closest to the sugar maple bole in 2010 compared with
2009; nevertheless, the slightly lower TF recorded at
more distal gauges in 2010, combined with the greater
surface area that these gauges represented, led to smaller
weighted mean total TF beneath the sugar maple in 2010
(Table V). The gauge at 0.86 m from the bole was
frequently disturbed in 2010 and was not used to estimate
total TF inputs beneath the sugar maple canopy; TF at this
location was interpolated from adjacent gauges. In 2009,
total TF at the three innermost gauges (0.1, 0.3 and 0.5 m)
was 146, 159 and 155% of Pg, respectively, while TF at
the outermost gauges (0.86, 1.10 and 2.13 m) was 74, 64
and 66% of Pg, respectively. In 2010, TF at the 0.1, 0.3,
0.5, 1.10 and 2.13 m gauges was 179, 185, 129, 60 and
59% of Pg, respectively. Conversely, total TF at all
Hydrol. Process. 29, 4068–4082 (2015)
4074 R. BIALKOWSKI AND J. M. BUTTLE

Figure 2. Vertical profiles of (a) porosity and (b) saturated hydraulic conductivity. Soil texture of (c) mixed hardwood-conifer, (d) young pine, and (e)
old pine) for the study stands

distances from the bole was less that total Pg for both
pines. Total SF, expressed as a depth over a 0.2 m wide
ring surrounding the sugar maple bole, was similar in
magnitude to TF recorded at the innermost gauge and
represented 180 and 230% of Pg in 2009 and 2010. The
addition of SF to TF near the bole of the sugar maple
resulted in total near-bole water inputs 3–5 × greater than
those for the pines (Table V and Figure 5).
Soil water storage
Soil water storage was consistently greater near the sugar
maple bole relative to greater distances from the bole in both
2009 and 2010 (difference-of-means t-test; p <0.05;
Figure 6a). This reflects greater water inputs near the bole
relative to distal locations beneath the canopy. Larger near-
bole inputs in 2010 (Figure 5) led to greater storage near the
sugar maple bole compared with that in 2009 (p <0.05). Soil
water storage at 2.13 m from the bole was slightly less than
those at similar distances from the pine boles (Figure 6b and
c), which agrees with the smaller amount of TF at this
distance under the sugar maple canopy relative to the pines
(Figures 4 and 5). There were no significant differences in
mean soil water storage near the bole relative to the most
distal location for either pine.
Soil water recharge below 1 m depth
Estimated total ET for the pine stands was 245 and
226 mm in 2009 and 2010, respectively (Table V), and
total ET as a fraction of Pg ranged from 0.64 (old pine,
2009) to 0.71 (young pine, 2010). Estimated total ET
for the mixed hardwood-conifer stand was 240 and
208 mm, representing 76 and 65% of Pg for 2009 and
2010, respectively. Mean daily ET for the pine stands
was 3.5 ± 0.2 mm (range = 3.2–4.0 mm) and 3.3 ± 0.6 mm
(range = 1.8–3.6 mm) for 2009 and 2010, respectively, while
mean daily ET for the mixed hardwood-conifer stand was
3.2 ± 0.6 mm (range = 2.3–4.4 mm) and 3.0 ± 0.6 mm
(range = 1.5–3.3 mm) for 2009 and 2010, respectively.

Copyright © 2015 John Wiley & Sons, Ltd. Hydrol. Process. 29, 4068–4082 (2015)
 RECHARGE UNDER DIFFERING TREE BRANCH ARCHITECTURES 4075

Assumes SF infiltrated within a ring surrounding the bole, the width of which was the mid-point between the closest and second closest Profile Probe access tube to the bole (sugar maple: 0.2 m; young
(mm) (mm) TF+ (mm) (L) by PCA* (mm) SF:Pg I (mm) I:Pg of tree (mm)# base of tree (mm)# (± estimated error, mm) (± estimated error, mm)
Table VI. Mean and range in funnelling ratio for the study trees
for the 2009 and 2010 study periods

RTotal with SF

121 ± 24
114 ± 29
135 ± 35
161 ± 35
82 ± 41
112 ± 53
Year Tree Mean funnelling ratio Range

2009 Sugar maple 6.7 (7.9)+ 0.1–13.8 (3.5–13.8)+

Young pine 1.3 0.1–4.1
Old pine 0.3 0–0.7
2010 Sugar maple 8.5 3.3–13.0
Young pine 1.2 0.4–2.4
Old pine 0.3 0.2–0.4
RTotal without SF

107 ± 23
114 ± 29
134 ± 35
143 ± 33
82 ± 41
112 ± 53 +
Excludes events where Pg was less than 0.5 mm or which did not
generate stemflow from the sugar maple.

Figure 7 presents the monthly MOD16 mixed hardwood-

Pg,, above-canopy rainfall; ET, evapotranspiration; TF, throughfall; SF, stemflow; I, canopy interception; RTotal, soil water recharge below 1 m depth.

conifer and red pine ET values, along with ET for a

5 km × 5 km pixel covering much of the central and western
Table V. Pg, ET, TF, SF, I and RTotal for the 2009 and 2010 study periods

SF input at base TF + SF input at

portions of the GF that is largely forested and contains the

sugar maple and young pine sites. Monthly ET in this pixel
338
376

243
256
1034

1306

was estimated from the Ecological Assimilation of Land and

Canopy Observations (EALCO) model, which simulates
various components of ET (e.g. canopy transpiration,
evaporation of intercepted precipitation and soil evapora-
tion) and has been calibrated and validated using tower flux
estimates of ET at sites across Canada (Wang et al., 2013).
571

735
35
22

14
15

Good agreement between MOD16 and EALCO ET for 2009

suggests the former provides reasonable estimates of ET for
the study sites. Because MOD16 ET includes water
0.30
0.16
0.14
0.37
0.30
0.15

evaporated from the canopy, total ET was reduced by I at

each site to estimate water loss from the soil column via
96
57
53

95
52
119

evaporation and transpiration. The combination of smaller

ET and greater I for the sugar maple relative to the pines
(Table V) leads to greater ET – I for the young and old pine
0.009
0.008
0.001
0.013
0.003
0.001

stands relative to the mixed hardwood-conifer stand in both

years as follows: 188 mm and 192 mm versus 144 mm in
SF SF weighted

2009 and 131 mm and 174 mm versus 89 mm in 2010. These

values were assumed to be uniform with distance from the
0.4

0.3
3
3

4
1

bole for all three trees.

Variations in soil water recharge below 1 m depth at
each distance interval (Rt; Equations 9 and 10) beneath
13
13

5
9
244

315

the trees partly reflect spatial patterns in water delivery to

the soil surface (Figure 8). Thus, Rt near the sugar maple
Areally-weighted mean TF. See text for details.
Mean

bole was much larger than at distal locations, particularly

218
295
329
196
222
287

when SF contributions were considered. In 2009, Rt as a

fraction of Pg at the 0–0.1, 0.1–0.3 and 0.3–0.5 m
240
245
245
208
226
226

intervals was 1.01, 1.14 and 1.09 (when TF was assumed

ET

to be the sole water input to the soil surface), and was

pine: 0.27 m; old pine: 0.26 m).
*PCA, projected canopy area.

2.81 for the innermost distance interval when SF

317
355
382
319
318
339
Pg

contributions were included. Conversely, Rt:Pg was less

2009 Sugar maple

2010 Sugar maple

than 0.29 at greater distances from the bole. Recharge

Young pine

Young pine
Old pine

Old pine

patterns beneath the tree in 2010 were comparable with

Tree

2009, with Rt:Pg values of 1.51 and 1.57 for the 0–0.1
and 0.1–0.3 distance intervals, and ratios ≤1 for greater
distances, assuming TF was the sole water input.
Year

Inclusion of SF contributions for the innermost distance

+

Copyright © 2015 John Wiley & Sons, Ltd. Hydrol. Process. 29, 4068–4082 (2015)
4076 R. BIALKOWSKI AND J. M. BUTTLE

Figure 3. Changes in throughfall (TF) with distance from the tree bole for two rainfall events in 2009 at the (a) sugar maple, (b) young pine and (c) old
pine. Above-canopy rainfall (Pg) and the volume of stemflow (SF) generated for each event for each tree are indicated

Figure 4. Throughfall (TF) proximal to (a) and distal from (b) the boles of the young and old pines versus TF at corresponding distances from the bole of
the sugar maple

Copyright © 2015 John Wiley & Sons, Ltd. Hydrol. Process. 29, 4068–4082 (2015)
 RECHARGE UNDER DIFFERING TREE BRANCH ARCHITECTURES 4077

Figure 5. Total throughfall (TF) and TF plus stemflow (SF) depths measured at various distances from the boles of the (a) sugar maple (b) young pine and
(c) old pine in the 2009 and 2010 study periods. SF depth was obtained by dividing SF volume by the area of a ring around the bole, the width of which was
the mid-point between the closest and second closest Profile Probe access tube to the bole (sugar maple: 0.2 m; young pine: 0.27 m; old pine: 0.26 m)

Figure 6. Depth of water stored in the upper 1 m of soil at sites that are proximal and distal to the boles of the (a) sugar maple, (b) young pine and (c) old
pine in the 2009 and 2010 study periods. Throughfall (TF) and TF + stemflow (SF) depths recorded for intervals between soil water content
measurements are also indicated for distal and proximal sites, respectively

interval gave a Rt:Pg of 3.82. All of the Rt values minus

Figure 7. Monthly total evapotranspiration for 2009, estimated by the
Ecological Assimilation of Land and Canopy Observations model for a
5 km × 5 km pixel covering western and central portions of the Ganaraska
Forest and by MOD16 for 1 km × 1 km pixels in the Ganaraska Forest
dominated by mixed hardwood-conifers and red pine
the estimated error in Rt (Appendix A) for all distance
intervals for the sugar maple were greater than 0 in 2009
and 2010. Slightly smaller Rt at distal locations beneath
the sugar maple relative to the pines reflects greater mean
TF beneath the red pine canopies in both 2009 and 2010
(Table V). Estimated Rt for various distance intervals
from the young and old pine boles was also greater than 0
in 2009 and 2010, but showed only a minor increase near
the bole. Thus, the maximum Rt:Pg ratio for the 0–0.1
distance interval when SF inputs were included was 0.48
for the old pine and 0.42 for the young pine, both in 2009.
Total recharge below 1 m for the sugar maple was
significantly greater than 0 for both years (Table V), both
with and without the inclusion of SF contributions.
Inclusion of these contributions increased RTotal by 13%;
however, this increase was not statistically significant. Total
recharge was also greater than 0 for the pines in 2009 and

Copyright © 2015 John Wiley & Sons, Ltd. Hydrol. Process. 29, 4068–4082 (2015)
4078 R. BIALKOWSKI AND J. M. BUTTLE
Figure 8. Total soil water recharge (R) below 1 m depth estimated assuming inputs of TF and TF + SF at various distances from the boles of the (a) sugar
maple (b) young pine and (c) old pine in the 2009 and 2010 study periods
2010, and adding SF contributions to TF inputs increased
RTotal for both the young and old pines by less than 1 mm.
Z-scores showed no significant difference in either RTotal or
RTotal:Pg between trees in either 2009 or 2010. The
exception was the sugar maple relative to the young pine
in 2010, where both RTotal and RTotal:Pg were greater at the
sugar maple (with and without SF inputs, p <0.1).
DISCUSSION
Hypothesis 1: Throughfall and R will increase with
proximity to the bole of the sugar maple and SF will be
maximized as a result of its branch architecture.
Our results provide strong support for this hypothesis. TF
at 0.1 m and 0.3 m from the bole regularly exceeded Pg,
while TF beyond 0.5 m was consistently less than Pg. This
reflects a branching system promoting the flow of
intercepted water towards the tree stem. Previous studies
of the heterogeneity of TF beneath deciduous trees
(Carlyle-Moses and Price, 2006; Alexander and Arthur,
2010; Sato et al., 2011) also noted increased values near the
bole. This study sampled TF beneath the inner 40% of the
sugar maple canopy radius, and TF quantities may
continue to decrease with greater distance, suggesting that
I for the sugar maple (Table V) may be an under-estimate of
the whole-tree value.
Total SF depths as a fraction of Pg for the sugar
maple were consistent with values reported for hard-
woods (e.g. Buttle and Farnsworth, 2012), and SF
volumes from the sugar maple were an order of magnitude
greater than those from the pines. The sugar maple had the
largest mean funnelling ratio of the three trees, although it
was considerably less than other values reported for maple
trees (Carlyle-Moses and Price, 2006; Alexander and
Arthur, 2010). Nevertheless, it indicates efficient channel-
ling of water towards the bole. Expression of SF volumes
relative to projected canopy area indicates much smaller
differences between the three trees (Table V) and provides
Copyright © 2015 John Wiley & Sons, Ltd.
little insight into how different tree species partition incident
rainfall and influence soil water dynamics beneath tree
canopies. However, standardization of SF volumes relative
to a focussed infiltration area adjacent to the bole, as in the
cylindrical infiltration model of Tanaka et al. (1996),
resulted in marked differences between the trees. The
assumption that SF infiltrated over a relatively narrow area
(0.2–0.27 m wide ring) around the bole is supported by other
studies (e.g. Voigt, 1960; Navar, 2011; Buttle et al., 2014),
and the addition of SF to TF inputs to the innermost distance
interval around the sugar maple bole resulted in values
3–4 × greater than Pg.
This enhanced water input around the sugar maple bole
led to consistently larger SWCs relative to more distal
locations beneath the canopy (cf. Tang, 1996) and greater
Rt near the bole, as reported elsewhere (Tanaka et al.,
1996; Taniguchi et al., 1996; Chang and Matzner, 2000).
Greater water storage near the sugar maple bole was
likely not the result of inter-stand differences in soil
properties, which were similar between sites (Figure 2).
Ratios of recharge to Pg for comparable forest types
(Bent, 2001; Dripps and Bradbury, 2010; Rosenqvist
et al., 2010) are similar to those found here for the outer-
most distance intervals beneath the sugar maple canopy
(≤0.3). However, the larger sugar maple RTotal:Pg values
reported here (0.38 and 0.50 in 2009 and 2010,
respectively) partly reflect inclusion of focussed TF and
SF as water inputs in water balance estimates of RTotal.
This suggests that neglecting the water-focussing effect of
tree species with branches sloping towards the bole may
lead to underestimation of R in these forest types.
Hypothesis 2: TF and R will not vary with proximity to
the boles of the pines and SF will be minimized as a result
of their branch architecture.
This hypothesis was also supported. No distinct patterns
in TF with distance from the bole were observed for either
pine in 2009 and 2010, suggesting their canopy
Hydrol. Process. 29, 4068–4082 (2015)
 RECHARGE UNDER DIFFERING TREE BRANCH ARCHITECTURES
architecture does not preferentially redirect and redistrib-
ute TF to the same extent as the sugar maple. This agrees
with Loustau et al. (1992), who found no consistent
variation in TF with distance from the bole or the tree
rows in an 18-year-old red pine plantation. Nevertheless,
branch orientation away from the pine boles combined
with their smaller canopy density (Table I) appeared to
promote greater TF at distal locations from their boles
compared with TF at a similar distances from the sugar
maple bole (Figures 4 and 5).
Both pines produced significantly less SF than the sugar
maple, while SF, expressed as a depth relative to the PCA,
was almost an order of magnitude greater for the young
compared with the old pine. Buttle and Farnsworth (2012)
also found greater SF depths from the younger red pine,
the result of similar SF volumes divided by much smaller
PCAs for younger pines. However, the similar SF volumes
from the young and old pines in 2009 and 2010 resulted in
comparable inputs around the tree boles (Table V).
These SF contributions relative to a 0.26–0.27 m wide
ring around their boles resulted in SF depths 12 and 55×
larger than SF relative to PCA for the young and old
pines, respectively. The ratio for the young pine agrees
with Voigt (1960)’s observation for a 35-year-old red
pine stand. Nevertheless, these focussed inputs did not
produce appreciably greater Rt near the bole of either pine
or increased RTotal. Soil water recharge estimates for both
pines in 2009 and 2010 were similar to other growing-
season estimates for conifer forests (Taniguchi et al.,
1996; Dripps and Bradbury, 2010; Rosenqvist et al.,
2010; Buttle et al., 2014).
Hypothesis 3: Contributions of SF to R will be less in the
older relative to the younger pine, as branches increas-
ingly slope away from the bole with greater tree age.
Stemflow funnelling ratios for the pines were much less than
for the sugar maple. Nevertheless, a mean ratio greater than
1 for the young pine (Table VI) indicates focussing of
intercepted water towards the bole. Funnelling ratios less
than 1 for the old pine suggest red pine branches bend away
from the stem and SF contributions to R around the bole
decline between tree ages of 36 and 57 years in the GF. This
agrees with studies of temporal changes in conifer branch
architecture (Johnson, 1990; Steinbuck, 2002). However,
the minor contribution of SF to estimated R for both pines
observed here does not permit a conclusive test of
Hypothesis 3. Buttle et al. (2014) also failed to observe a
consistent change in the role of SF contributions to R with
increasing age in red pine stands in the GF.
Hypothesis 4: The importance of spatially-focussed SF and
TF inputs to R will decrease in the order: sugar
maple → younger red pine → older red pine.
Copyright © 2015 John Wiley & Sons, Ltd.
4079
This hypothesis was not supported. Focussed SF and TF
inputs resulted in increased RTotal for the sugar maple, but
these changes were not statistically significant. Total R and
RTotal:Pg for the sugar maple were not significantly greater
than values for the pines with or without the inclusion of SF
inputs, with the exception of the young pine in 2010.
Focussed inputs near the tree bole also had minimal effect
on soil water recharge for either pine. This contrasts with the
marked contribution that SF made to recharge beneath
Japanese red pine (Tanaka et al., 1996) and 28-year-old red
pine in the GF (Buttle et al., 2014). Further study of younger
red pine in the GF is needed to determine the age threshold at
which SF no longer makes a significant contribution to R.
Other studies have noted greater recharge from
hardwood relative to coniferous stands (Wattenbach
et al., 2007; Komatsu et al., 2008; Rosenqvist et al.,
2010), and Schwärzel et al. (2012) suggested that this
may partly reflect greater SF fluxes from hardwoods
compared with conifers. Our inability to detect enhanced
recharge under the sugar maple canopy relative to the
pines partly reflects larger interception for the former,
such that less TF reached the forest floor under the canopy
relative to the young and particularly the old pine
(Table V). Greater SF and TF inputs near the sugar
maple bole failed to compensate for the reduced water
delivery to the larger sub-canopy area that only received
TF. Nevertheless, RTotal and RTotal:Pg for the sugar maple
exceeded those for the young pine in 2010, likely as a
result of increased SF from the sugar maple. Maximum
and mean rainfall intensity were slightly greater in 2009
(13.1 ± 9.3 mm h
1 and 2.8 ± 1.8 mm h
1) compared with
that in 2010 (10.2 ± 6.5 mm h
1 and 1.9 ± 1.1 mm h
1),
while mean rainfall duration during storms was slightly
less (11.8 ± 9.6 h vs. 17.5 ± 12.1 h). While these results are
not statistically significant, they are consistent with studies
for sugar maple showing a decrease in SF with increasing
rainfall intensity (Carlyle-Moses and Price, 2006), and a
potential increase in the proportion of the tree’s stems and
branches contributing SF to the bole during longer-
duration storms (Hutchinson and Roberts, 1981).
Other factors influencing our ability to discern any
differences in RTotal between the trees are the use of the
water balance approach to estimate recharge and the large
errors associated with recharge estimates (Appendix A).
Recharge was estimated as the residual of the water
balance (Equation 9), where ET was the only component
not measured directly. There is good correspondence
between ET estimates for the mixed hardwood-conifer
stand and values reported for daily ET (Papakyriakou and
McCaughey, 1991; Sun et al., 2008), annual total ET
(Buttle, 1994; Liu et al., 2003) and annual ET : Pg (Sun
et al., 2008) for hardwoods. Estimated mean daily ET,
annual total ET and ET : Pg for the pine stands were
similar to those reported for conifers (Kelliher et al.,
Hydrol. Process. 29, 4068–4082 (2015)
4080 R. BIALKOWSKI AND J. M. BUTTLE
1993; Liu et al., 2003; Restropo and Arain, 2005; Sun
et al., 2008) and slightly exceeded those for the mixed
hardwood-conifer stand used to represent the sugar maple
ET. Sun et al. (2008) reported similar ET from mixed
hardwood and red pine stands using eddy covariance
measurements, while Mackay et al. (2002) found that red
pine transpiration rates generally exceeded (by up to 2×)
those for maple trees. Wattenbach et al. (2007) also noted
a 3.4% decline in ET as pine forests were succeeded by
hardwoods in Germany, comparable with sugar maple–
red pine ET differences found in this study (2% in 2009
and 8% in 2010). Nevertheless, further examination of ET
in the GF is required to verify the ET values used to
estimate R and to reduce the error associated with ET and
other water balance components. Some of these errors were
considerable (Appendix A), such as the relative error in ET
values estimated for 2010 (>0.4) and the mean standard
error of SF (0.56–0.60). Narrowing the uncertainty
associated with water balance fluxes will increase our
ability to detect significant differences in soil water
recharge between trees of differing species and ages.
Future studies of the contributions of SF and TF to soil
water recharge as forest cover in the GF changes from
pine plantations to mixed hardwood-conifer stands must
also address any changes in the spatial variability in these
fluxes for different tree species and a greater range of tree
ages. This study assumed TF and SF inputs were uniform
over the distance interval to which they were applied. TF
does not appear to vary with direction at a given distance
from the bole (Johnson, 1990). However, greater SF fluxes
have been found on downslope relative to upslope sides of
boles (Liang et al., 2007), resulting in increased SWC and
pore water pressure downslope of the tree. Variations in tree
stem inclination may partly control differences in total SF
production (Levia et al., 2015) and may lead to heteroge-
neous SF inputs around tree boles even on relatively flat
ground. This would suggest that SWC profiles and resulting
recharge estimates obtained here may not completely
represent the potential microscale heterogeneity around
the circumference of the bole. Additional study using TF
and SWC transects in multiple directions from the bole
would provide insight into this issue.
CONCLUSIONS
Differences in branch geometry between sugar maple and
red pine trees partly controlled SF and TF delivery to the
forest floor, and the distribution of point-scale soil water
recharge below 1 m depth with distance from the bole
beneath individual tree canopies in the GF in southern
Ontario. Larger near-bole water inputs as a result of
greater SF and spatial focussing of TF resulted in
enhanced recharge around the bole of a sugar maple,
while recharge was relatively uniform with distance from
Copyright © 2015 John Wiley & Sons, Ltd.
the bole for both young and old red pines. However, there
was no significant difference in total recharge beneath the
canopies of the three trees, which likely reflected the
greater interception efficiency of the sugar maple canopy
combined with the large errors associated with the
recharge estimates. Nevertheless, our results suggest that
the spatial pattern of recharge below 1 m depth during
spring and summer rainfalls may change at the tree and
stand scales in the GF as red pine plantations undergo the
transition to a mixed hardwood-conifer forest, with
implications for water availability to trees and soil
biogeochemical processes in this forest landscape.
ACKNOWLEDGEMENTS
This work was supported by the Natural Sciences and
Engineering Research Council of Canada. We thank the
Ganaraska Region Conservation Authority for their
continued support; Peter Lafleur, Andrew Farnsworth,
Clément Le Saux and Alexandra Ryland for their
assistance; Ken Hill and Marie and John Toon for
permission to install instrumentation on their property;
Qiaozhen Mu (University of Montana) for provision of
the MOD16 data; and Shusen Wang (Canada Centre for
Remote Sensing) for provision of the EALCO data. We
also thank two anonymous reviewers for their construc-
tive comments on an earlier version of this paper. The
authors state that they have no conflict of interest to
declare with respect to this submission.
REFERENCES
Alexander H, Arthur M. 2010. Implications of a predicted shift from
upland oaks to red maple on forest hydrology and nutrient availability.
Canadian Journal of Forest Research 40: 716–726.
Beier C, Hanson K, Gunderson P. 1993. Spatial variability of throughfall
fluxes in a spruce forest. Environmental Pollution 81: 257–267.
Bent GC. 2001. Effects of forest-management activities on runoff
components and ground-water recharge to Quabbin Reservoir, central
Massachusetts. Forest Ecology and Management 143: 115–129.
Brown DM, McKay GA, Chapman LG. 1968. The Climate of Southern
Ontario, Climatological Studies 5. Meteorological Branch, Department
of Transport: Toronto.
Buttle JM. 1994. Hydrological response to reforestation in the Ganaraska
River basin, southern Ontario. The Canadian Geographer 38: 240–253.
Buttle JM. 2011. Streamflow response to headwater reforestation in the
Ganaraska River basin, southern Ontario, Canada. Hydrological
Processes 25: 3030–3041.
Buttle JM, Farnsworth AG. 2012. Measurement and modeling of
canopy water partitioning in a reforested landscape: the Ganaraska
Forest, southern Ontario, Canada. Journal of Hydrology 466-467:
103–114.
Buttle JM, Toye HJ, Greenwood WJ, Bialkowski R. 2014. Stemflow and
soil water recharge during rainfall in a red pine chronosequence on the
Oak Ridges Moraine, southern Ontario, Canada. Journal of Hydrology
517: 777–790.
Carlyle-Moses D, Price AG. 2006. Growing-season stemflow production
within a deciduous forest in southern Ontario. Hydrological Processes
20: 3651–3663.
Hydrol. Process. 29, 4068–4082 (2015)
 RECHARGE UNDER DIFFERING TREE BRANCH ARCHITECTURES
Chang S-C, Matzner E. 2000. The effect of beech stemflow on spatial
patterns of soil solution chemistry and seepage fluxes in a mixed
beech/oak stand. Hydrological Processes 14: 135–144.
Dingman SL. 2002. Physical Hydrology, 2nd edition. Prentice-Hall:
Upper Saddle River, NJ; 646.
Dripps WR, Bradbury KR. 2010. The spatial and temporal variability of
groundwater recharge in a forested basin in northern Wisconsin.
Hydrological Processes 24: 383–392.
Durocher M. 1990. Monitoring spatial variability of forest interception.
Hydrological Processes 4: 215–229.
Environment Canada. 2011. National Climate Data and Information
Archive: Canadian Climate Normals 1971–2000: Peterborough A.
Website accessed August 23, 2011: http://www.climate.weatheroffice.
gc.ca/climate_normals
Ford E, Deans J. 1978. The effects of canopy structure on stemflow,
throughfall and interception loss in a young sitka spruce plantation.
Journal of Applied Ecology 15: 905–917.
Funk G. 1977. Geology and water resources of the Bowmanville, Soper
and Wilmot Creeks IHD Representative Drainage Basin, Water
Resources Report 9a, Ontario Ministry of the Environment,Water
Resources Branch, Toronto, ON, 113 p
Gerber R, Howard KWF. 2002. Hydrogeology of the Oak Ridges Moraine
aquifer system: implications for protection and management from the
Duffins creek watershed. Canadian Journal of Earth Sciences 39:
1333–1348.
Germer S. 2013. Development of near-surface perched water tables during
natural and artificial stemflow generation by babassu plams. Journal of
Hydrology 507: 262–272.
Greenwood WJ, Buttle JM. 2014. Effects of reforestation on near-surface
saturated hydraulic conductivity in a managed forest landscape,
southern Ontario, Canada. Ecohydrology 7: 45–55.
Guswa A, Spence C. 2012. Effect of throughfall variability on recharge:
application to hemlock and deciduous forests in western Massachusetts.
Ecohydrology 5: 563–574.
Herwitz S. 1986. Infiltration-excess caused by stemflow in a cyclone-
prone tropical rainforest. Earth Surface Processes and Landforms 11:
401–412.
Herwitz S. 1987. Raindrop impact and water flow on the vegetative
surfaces of trees and the effects on stemflow and throughfall generation.
Earth Surface Processes and Landforms 12: 425–432.
Hoffman DW, Acton CJ. 1974. The soils of Northumberland county.
Report No. 42, Ontario Soil Survey Research Branch, Agriculture
Canada and the Ontario Agricultural College, Ottawa, ON, 117 p.
Hutchinson I, Roberts MC. 1981. Vertical variation in stemflow
generation. Journal of Applied Ecology 18: 521–527.
Johnson R. 1990. The interception, throughfall and stemflow in a forest in
highland Scotland and the comparison with other upland forests in the
U.K. Journal of Hydrology 118: 281–287.
Johnson M, Lehmann J. 2006. Double-funneling of trees: Stemflow and
root-induced preferential flow. Ecoscience 13: 324–333.
Keim R, Skaugset A, Weiler M. 2006. Storage of water on vegetation
under simulated rainfall of varying intensity. Advances in Water
Resources 29: 974–986.
Kelliher F, Leuning R, Schulze E. 1993. Evaporation and canopy
characteristics of coniferous forests and grasslands. Oecologia 95: 153–163.
Komatsu H, Kume T, Otsuki K. 2008. The effect of converting a native
broad-leaved forest to a coniferous plantation forest on annual water
yield: a paired-catchment study in northern Japan. Forest Ecology and
Management 255: 880–886.
Levia DF, Herwitz SR. 2000. Physical properties of water in relation to
stemflow leachate dynamics: implications for nutrient cycling. Cana-
dian Journal of Forest Research 30: 662–666.
Levia DF, Michalzik B, Näthe K, Bischoff S, Richter S, Legates DR.
2015. Differential stemflow yield from European beech saplings:
the role of individual canopy structure metrics. Hydrological Processes
29: 51–53.
Liang W-L, Kosugi K, Mizuyama T. 2007. Heterogeneous soil water
dynamics around a tree growing on a steep hillslope. Vadose Zone
Journal 6: 879–889.
Liang W-L, Kosugi K, Mizuyama T. 2009. A three-dimensional model of
the effect of stemflow on soil water dynamics around a tree on a
hillslope. Journal of Hydrology 366: 62–75.
Copyright © 2015 John Wiley & Sons, Ltd.
4081
Liu J, Chen J, Cihlar J. 2003. Mapping evapotranspiration based on
remote sensing: an application to Canada’s landmass. Water Resources
Research 39: SWC41–SWC415.
Loustau D, Berbigier P, Granier A, Hadj MF. 1992. Interception loss,
throughfall and stemflow in a maritime pine stand. I. Variability of
throughfall and stemflow beneath the pine canopy. Journal of
Hydrology 138: 449–467.
Mackay D, Ahl D, Ewers B, Gower S, Burrows S, Samanta S, Davis J.
2002. Effects of aggregated classifications of forest composition on
estimates of evapotranspiration in a northern Wisconsin forest. Global
Change Biology 8: 1253–1265.
McKee AJ, Carlyle-Moses DE. 2010. Stemflow: A potentially important
point source of water for growth. Link 11: 11–12.
Mu Q, Zhao M, Running S. 2011. Improvements to a MODIS global
terrestrial evapotranspiration algorithm. Remote Sensing of Environment
115: 1781–1800.
Nàvar J. 2011. Stemflow variation in Mexico’s northeaster forest
communities: its contribution to soil moisture content and aquifer
recharge. Journal of Hydrology 408: 35–42.
Papakyriakou T, McCaughey J. 1991. An evaluation of the water balance
technique for the estimation of evapotranspiration for a mixed forest.
Canadian Journal of Forest Research 21: 1622–1631.
Restropo NC, Arain MA. 2005. Energy and water exchanges from a
temperate pine plantation forest. Hydrological Processes 19: 27–49.
Reynolds WD, Elrick DE. 1985. In situ measurement of field saturated
hydraulic conductivity, sorptivity, and the alpha-parameter using the
Guelph permeameter. Soil Science 140: 292–302.
Rodrigo A, Àvila A. 2001. Influence of sampling size in the estimation of
mean throughfall in two Mediterranean holm oak forests. Journal of
Hydrology 243: 216–227.
Rosenqvist L, Hansen K, Vesterdal L, van der Salm C. 2010. Water
balance in afforestation chronosequences of common oak and Norway
spruce on former arable land in Denmark and southern Sweden.
Agricultural and Forest Meteorology 150: 196–207.
Sato AM, de Avelar AS, Netto C. 2011. Spatial variability and temporal
stability of throughfall in a eucalyptus plantation in the hilly lowlands of
southeastern Brazil. Hydrological Processes 25: 1910–1923.
Schwärzel K, Ebermann S, Schalling N. 2012. Evidence of double-
funneling effect of beech trees by visualization of flow pathways using
dye tracer. Journal of Hydrology 470-471: 184–192.
Staelens J, De Schrijver A, Verheyen K, Verhoest N. 2006. Spatial
variability and temporal stability of throughfall water under a dominant
beech (Fagus sylvatica L.) tree in relationship to canopy structure.
Environmental Pollution 142: 254–263.
Steinbuck E. 2002. The influence of tree morphology on stemflow in a
redwood region second-growth forest – Master’s thesis. California State
University at Chico, Chico, CA, 55 p.
Sun G, Noormets A, Chen J, McNulty S. 2008. Evapotranspiration
estimates from eddy covariance towers and hydrologic modeling in
managed forests in northern Wisconsin, USA. Agriculture and Forest
Meteorology 148: 257–267.
Tanaka T, Taniguchi M, Tsujimura M. 1996. Significance of stemflow in
groundwater recharge. 2: a cylindrical infiltration model for evaluating
the stemflow contribution to groundwater recharge. Hydrological
Processes 10: 81–88.
Tang C. 1996. Interception and recharge processes beneath a Pinus elliotii
forest. Hydrological Processes 10: 1427–1434.
Taniguchi M, Tsujimura M, Tanaka T. 1996. Significance of stemflow in
groundwater recharge. 1: evaluation of the stemflow contribution to recharge
using a mass balance approach. Hydrological Processes 10: 71–80.
Voigt G. 1960. Distribution of rainfall under forest stands. Forest Science
6: 2–10.
Wang S, Yang Y, Luo Y, Rivera A. 2013. Spatial and seasonal variations
in evapotranspiration over Canada’s landmass. Hydrology and Earth
System Sciences 1&: 3561–3575.
Wattenbach M, Zebisch M, Hattermann F, Gottschalk P, Goemann H,
Kreins P, Badeck F, Lasch P, Suckow F, Wechsung F. 2007.
Hydrological impact assessment of afforestation and change in tree-
species composition – a regional case study for the federal state of
Brandenburg (Germany). Journal of Hydrology 346: 1–17.
Whelan MJ, Sanger LJ, Baker M, Anderson JM. 1998. Spatial patterns of
throughfall and mineral ion deposition in a lowland Norway spruce
Hydrol. Process. 29, 4068–4082 (2015)
4082 R. BIALKOWSKI AND J. M. BUTTLE

(Picea abies) plantation at the plot scale. Atmospheric Environment
32: 3493–3501.
Winkler RL, Hays WL. 1975. Statistics: probability, inference, and decision.
nd
2 Ed., Holt, Rinehart and Winston, New York, NW, 889 p + appendices.
Xiao Q, McPherson EG. 2011. Rainfall interception of three trees in
Oakland, California. Urban Ecosystems 14: 755–769.
The error in TF (σ TF t ) on day t was estimated as 20% of
TF. This is a conservative estimate based on Rodrigo and
Àvila (2001), who found a coefficient of variation for TF
measured by two gauges to be approximately 15%. The
error in SF (σ SFt ) on day t was estimated as:

APPENDIX A. ERROR ANALYSIS FOR ESTIMATED σ SFt ¼ SF t SE (A10)

SOIL WATER RECHARGE (R)
The error in Rtotal ðσ Rtotal Þ was calculated as: where SE is the mean standard error in SF from all days of
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi measurement. Three trees were used to assess SE for SF

ffi for the sugar maple studied for soil water recharge
n
σ Rtotal ¼ ∑i¼1 σ 2Ri (A1) (Table III), while four trees in each of the young and old
pine stands were used to estimate SF SE for the pines
where σ Ri , the error in Ri, is: studied for soil water recharge. The SEs of SF at the
  mixed hardwood-conifer, young pine and old pine stands
σ Rt
σ Ri ¼ Ri (A2) were 0.59, 0.56 and 0.60, respectively.
Rt The σ ET t for day t in 2009 was estimated as 25% of ET
where σ Rt , the error in Rt, at the proximal access tube to based on the mean absolute bias in MOD16 estimates
each bole is: reported in Mu et al. (2011). The σ ET t for day t in 2010
was calculated as:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
σ Rt ¼ σ 2TF þ σ 2SF þ σ 2ET þ σ 2I þ σ 2ΔS (A3) qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
σ ET t ¼ σ 2ET Hamont þ σ 2ET rel (A11)
and:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
where σ ET Hamont is the error in the Hamon ET estimate and
σ Rt ¼ σ 2TF þ σ 2ET þ σ 2I þ σ 2ΔS (A4)
σ ET rel is the mean relative error in the regression estimates
at the other access tubes, where σ TF, σ SF, σ ET, σ I and σ ΔS of ET (Table IV). σ ET Hamont was estimated to be 20% of the
are errors in TF, SF, ET, I and ΔS on day t. These error Hamon ET estimates, while σ ET rel was taken as 48 and
terms were calculated as: 42% for the mixed hardwood-conifer and pine stands,
respectively, based on the average range in MOD16 ET
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

values about the respective lines of best-fit.
σ TF ¼ ∑nt¼0 σ 2TF t (A5) The error in I for day t (σ I t ) was calculated as:
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

ffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
σ SF n
¼ ∑t¼0 σ SFt 2
(A6) σIt ¼ σ 2Pg t þ σ 2TFt þ σ 2SFt (A12)

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi where σ Pg t , the error in Pg on day t, was assumed to be

n  
σ EF ¼ ∑ σ 2ET t (A7) 20% of Pg (the average coefficient of variation in Pg from
t¼0 all of the open sites).
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi The error in storage (σ ΔSt ) on day t was assumed to be

15% of storage on day t, based on the relative variability
σ I ¼ ∑nt¼0 σ 2I t (A8) in measured SWCs about the calibration relationship
given in Buttle et al. (2014).
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

ffi
n
σ ΔS ¼ ∑t¼0 σ ΔSt 2
(A9)

Copyright © 2015 John Wiley & Sons, Ltd. Hydrol. Process. 29, 4068–4082 (2015)