Professional Documents
Culture Documents
Grazing Impacts On Gully Dynamics Indicate Approaches For Gully Erosion Control in Northeast Australia
Grazing Impacts On Gully Dynamics Indicate Approaches For Gully Erosion Control in Northeast Australia
1
CSIRO Land and Water, scott.wilkinson@csiro.au GPO Box 1700 Canberra, ACT
2
CSIRO Land and Water, Townsville QLD 4814, Australia.
3
ANU Fenner School of Environment and Society, ACT 2601, Australia.
4
CSIRO Land and Water, GPO Box 2583, Brisbane QLD 4001, Australia
Abstract
increase in the amount of fine sediment delivered to the coastal lagoon of the Great
Barrier Reef, but gully erosion rates and dynamics are poorly understood. This study
monitored annual erosion, deposition and vegetation cover in 6 gullies for 13 years,
further 11 gullies in three nearby catchments for four years to investigate the effects
of grazing intensity. The long-term fine sediment yield from the planform area of
This article has been accepted for publication and undergone full peer review but has not
been through the copyediting, typesetting, pagination and proofreading process which may
lead to differences between this version and the Version of Record. Please cite this article as
doi: 10.1002/esp.4339
sediment yield, indicating that gullies were erosion hotspots within the catchment. It
was estimated that gully erosion supplied between 29 and 44% of catchment
sediment yield from 4.5% of catchment area, of which 85% was derived from gully
wall erosion. Under long-term livestock exclusion gully sediment yields were 77%
lower than those of grazed gullies due to smaller gully extent, and lower erosion
rates especially on gully walls. Gully wall erosion will continue to be a major
landscape sediment source that is sensitive to grazing pressure, long after gully
depth and length have stabilised. Wall erosion was generally lower at higher levels of
wall vegetation cover, suggesting that yield could be reduced by increasing cover.
Annual variations in gully head erosion and net sediment yield were strongly
dependent on annual rainfall and runoff, suggesting that sediment yield would also
downstream valley segments of most gullies. This study concludes that reducing
livestock grazing pressure within and around gullies in hillslope drainage lines is a
primary method of gully erosion control, which could deliver substantial reductions in
sediment yield.
Keywords
change.
draining to the Great Barrier Reef (GBR) lagoon contain more than 80,000 km of
gullies (Thorburn and Wilkinson, 2013). Gully erosion can also be a large source of
catchment sediment yield (Poesen et al., 2003). In the 130,000 km2 Burdekin River
basin draining to the GBR, fine sediment yield is derived predominantly from sub-
surface soil (Wilkinson et al., 2013, Wilkinson et al., 2015), and is impairing the
(De'ath and Fabricius, 2010). The Burdekin River basin is the largest single source of
fine sediment to the GBR lagoon (Kroon et al., 2012). However, there has been little
previous research into the rates, processes and drivers of gully erosion in the
including removing woody vegetation (Prosser and Slade, 1994). However, once an
incised channel exists, the rate of gully head retreat can also be heavily influenced
by runoff volume (Oostwoud Wijdenes and Bryan, 2001). The rate of gully extension
decays over time as the catchment area upslope of the headcut, and the runoff
volume delivered from that declines (Graf, 1977). A threshold of catchment slope
and area exists below which gully heads are unlikely to extend at a given point in
time (Montgomery and Dietrich, 1989). That this threshold is to some degree
dependent on vegetation cover (Torri and Poesen, 2014) suggests that vegetation
affects the erosion resistance and the surface runoff volume driving gully erosion.
combined with gully bed level control (e.g., Heede, 1979), or reafforestation of gullied
catchments (e.g., Gomez et al., 2003, Chen and Cai, 2006). However, gully erosion
processes and drivers can vary considerably with terrain, soil and climate conditions.
Site-specific process knowledge can help to inform targeted gully erosion control
production or other land uses. For example, if sediment generation derives from
should address the hydrologic functioning of the landscape and its vegetation. The
dominant headcut erosion process may be scour (Prosser and Soufi, 1998) or mass
failure (Chen et al., 2015). If the gully floor is degrading then engineered check dams
may be necessary for reducing gully sediment yield. If sediment yield derives from
Alternatively if wall erosion occurs through channel floor widening and mass failure
then revegetation may not be effective on its own and reducing runoff may be a
Sediment source tracing studies indicate that 87–96% of fine sediment delivered to
the GBR lagoon is derived from the small proportion of the landscape where
processes eroding sub-surface soil occur such as deep rilling, gully erosion and
streambank erosion (Hughes et al., 2009, Olley et al., 2013, Wilkinson et al., 2013,
Wilkinson et al., 2015). Gully walls and floors together contribute ~40% of fine
Governments have set targets to change land management by 2025 such that
understand the causes and treatment options for gullies in GBR catchments.
This study investigated the erosion dynamics and driving factors affecting gullies in
drainage lines within the Burdekin River basin in the semi-arid tropics of northeast
Australia. These gullies are understood to have developed following the introduction
of cattle grazing in the 1860s. We monitored gully erosion and surface runoff at
selected sites in four catchments for up to 13 years. The research addressed four
questions; (i) the significance of gully erosion as a catchment sediment source, (ii)
whether there are specific spatial elements or processes dominating gully erosion
which may help to inform those undertaking erosion control activities, (iii) how gully
erosion varies over time and the factors influencing that variation, and (iv) the extent
Methods
Study area
The four study catchments are located within the Upper Burdekin catchment of the
130,000 km2 Burdekin River basin (Figure 1). The study area has Granodiorite
lithology (Isbell, 1996, Bartley et al., 2010a), being within the Ravenswood Batholith
which was intruded in the early–middle Ordovician (Hutton et al., 1993). The
soil, which is a duplex textured sandy clay loam. These soils occur predominantly on
hectare contributions of fine sediment to the GBR (McKergow et al., 2005, Bartley et
al., 2015, Furuichi et al., 2016). Gully erosion is more prevalent in Chromosols than
in other soil orders in the Burdekin basin; while they cover 12% of the Burdekin
basin, 25% of the area mapped as having medium or high gully density has
Chromosol soils (Gilad et al., 2012). Chromosol soils are similarly important across
The study catchments have mean-annual rainfall (1900–2012) ranging between 609
mm a-1 and 782 mm a-1 (Table 1). Rainfall intensities are large in this highly seasonal
tropical environment, with more than 70% of annual rainfall occurring within the three
summer months. Annual rainfall and catchment runoff have coefficients of variation
of 40% and ~100%, respectively, which is highly variable in world terms, but typical
The terrain is dissected by many drainage lines, although total relief is low and
median gradients of the four catchments are all in the range 2.2—2.6%. The soil is
The gullies are branching linear features which typically follow drainage lines and
have a wedge shaped planform with width increasing downstream (Heine, 2002).
While the gully features are individually not large, due to the dissected terrain the
linear density of linear gully features is high, being estimated at 4.5 km km-2 in the
Weany Creek catchment (Heine, 2002). Larger gullies with more variable cross
confinement, such as in alluvium and other deeper soils, although these are
predominant only within particular valleys (e.g., Shepherd, 2010, Shellberg et al.,
2016). Historical air photos indicate that all the monitored gullies were well-
established by 1945.
The four study catchments represent gully erosion under a range of contemporary
grazing pressures. Cattle grazing commenced in the area after 1850 and numbers
increased slowly until the 1940s or later. Reliance on natural surface water would
streams (Abbott and McAllister, 2004), elevating the risk of gully incision. Following
the introduction of Bos inducus breeds and a slump in beef prices in the late 1960s
cattle numbers in Weany and Wheel Creek catchments increased sharply to the mid-
1970s (McKeon et al., 1990). Stocking rates in the Wheel Creek catchment were
approximately 40 head per km2 prior to 2001, and 25 head per km2 since. Stocking
rates in the Weany Creek catchment were approximately 25 head per km 2 prior to
2001 and 5–20 head per km2 since, when pasture resting in alternate summer wet
seasons and forage budgeting were introduced (Bartley et al., 2010a, Bartley et al.,
2014).
By contrast, the Main and Thorton Creek catchments were acquired by the
lower, except during agistment in several drought years in the 1990s, and since 2000
they have been under long-term exclusion, with relatively very low stocking rates
(Bothriochloa pertusa) which replaced native perennial tussock grasses after drought
and grazing pressure in the 1980s (Bartley et al., 2010a). Indian Couch is a
stoloniferous grass which in this environment spreads rapidly after rain but produces
less biomass and is more prone to drought. The ground vegetation in Main and
Thorton Creek today has a biomass several times that in Weany and Wheel Creek,
and remains dominated by native perennial grasses including Black Spear grass
The particle size distribution and dry bulk density was determined for the soil profile
being eroded by gullies by collecting 5 samples between zero and 1.2 m depth from
each of 5 soil cores collected in the Weany Creek and Wheel Creek catchments
(locations shown in Figure 1), and also by prior sampling of gully walls with a hand
sampled and the proportion of silt and clay (<63 µm) in each sample was determined
by dry sieving. This size fraction, herein termed the fine fraction, impacts
and phosphorus (Fabricius et al., 2014). Bulk density was determined for a total of
Gully erosion rates were monitored annually in a total of 17 gullies across the four
catchments (Figure 1; Table 1). In the Weany Creek catchment, gully erosion was
be studied. In the other catchments, which had different grazing intensities, the
monitoring period was 4 water years (2014–2017). Measurements were made in the
winter dry season (April—October), and annual erosion was reported for water years
Gullies were subdivided according to Crouch (1987) into three source areas;
headcut, walls and floor. The linear retreat of the headcut(s) at the upslope end of
each gully was measured with a large set square in the early years, from the surface
break of slope back to steel pin reference markers which were inserted vertically into
the soil surface prior to each wet season, at 30 cm upslope of the headcut. The
reference markers were spaced at 0.5 m intervals around the headcut, and between
10 and 20 markers were used depending on gully width. From 2009/10 headcut
erosion was monitored with annual surveys using a total station theodolite or Real
annual headcut retreat were then calculated. Some gully head segments had forked
heads which were monitored together as compound headcuts. Some gullies in the
Weany Creek catchment had additional side branches with headcuts which were not
depth immediately below the headcuts, and also gully length. Cross sections were
Annual wall erosion and deposition in each gully was monitored at three cross
which net erosion was visually evident, (ii) in a middle segment in which neither
erosion nor deposition were obviously dominant, and (iii) in a valley segment in
which net deposition was predominant (Heine, 2002), down to where the gully met
the stream network. Erosion or deposition at each cross section was quantified using
the local surface at 500 mm spacing. This technique is suited to the small rates of
wall erosion that were expected based on gully size and age (Lawler, 1993). The
pins covered the actively-eroding walls of each cross section excluding the gully
floor. Between 4 and 16 pins (average of 8) were used depending on the cross
section width. A total of 449 pins were installed across the four catchments (Table 1).
In each catchment, each segment type typically had 20–40 pins, depending on the
At each monitoring date the exposed length of erosion pin was measured from a
steel washer placed on the soil surface (see Chaplot, 2013), using digital callipers
with 0.01 mm precision. The mean of measurements taken on opposite sides of the
pin was used. The repeatability of pin measurement was determined by three repeat
measurements of 28 pins by two operators on a single day. Pins that were affected
exceeding 100 mm in any one year) were labelled as channel pins to allow for
differentiation from gully walls with respect to bulk density and the proportion of <63
Scour chains were inserted vertically in the sand bed adjacent to valley cross
sections to measure net changes in gully floor level over each measurement period.
Lengthening of the horizontal portion of the chain over the wet season indicated
erosion (scour), and burial of the chain indicated deposition (Gordon et al., 1992).
The individual magnitudes of scour and deposition also indicated relative differences
in sediment transport between gullies. The chains were reset to floor surface level
each year.
The erosion monitoring protocol was adapted in several gullies: (i) One gully (VP0)
had five head segments, each being a branch of the gully, of which the mean erosion
rates were reported; (ii) Two gullies shared a common valley segment (THB and
THC); (iii) one gully (VPXG5) did not have a scour chain until the final two years of
monitoring, so floor scour is not reported; (iv) In the valley segment of two Wheel
Creek, from 2011 the head and middle segments of two gullies (VPXG2 and VPXG5)
had livestock excluded by fencing, and several low porous check dams were
As well as reporting erosion, deposition and yield from each monitored gully element
the whole-of-gully net annual sediment yields were calculated. Gully annual fine
sediment yield (<63 µm), M , was calculated by Equation 1, as the sum of yield
from the headcut, from the three erosion pin cross sections, and from the valley floor
based on the scour chain. In Equation 1 p was the proportion of parent soil with
<63 µm particle size, was the soil dry bulk density, Ahc was the planimetric area
of headcut retreat, Dhc was the mean gully depth at the head cut approximated as
0.667 of the surveyed maximum depth of the headcut (assuming a parabolic cross
section), H , M ,V were the head, middle and valley cross sections respectively, Li
was the gully segment length represented by cross section i , Dij was the erosion or
deposition depth for cross section pin j (10 pins assumed), wij was the cross
section width it represented (0.5 m), pij was the proportion of fine material (<63 µm
particle size) assigned based on whether each cross section pin was deemed to
represent parent soil p or deposited sediment pd , ij was the dry bulk density of
either parent soil or deposited sediment d , and Ds was the net erosion of
sediment from the gully floor in the valley segment as measured by the scour chain
pij and pd =1.0, which removed proportion fine from the equation.
Specific fine sediment yields were calculated by dividing by gully planform area
(Equation 2):
M
SSYGA 1
2 wH wV LH LM LV Equation 2
This metric is appropriate for estimating yield from gully networks across catchments
and for comparison with yields from other landscape elements and scales.
Specific fine sediment yields were also calculated by dividing by the upslope
contributing area draining to the downstream end of each gully ( UCA ; Equation 3):
M
SSYUCA Equation 3
UCA
This metric is suited to assessing the effect on catchment fine sediment yield of
differences in both gully extent and erosion rate between the four monitored
catchments. DEMs derived from airborne LiDAR were used to determine UCA .
Rainfall was measured using tipping bucket rain gauges at multiple locations in each
catchment (Figure 1), and the mean value for each catchment was reported. Surface
runoff in the Weany Creek catchment was monitored using a Parshall flume draining
at the outlets of Weany Creek and Main Creek catchments using regressions of
of surveyed creek cross sections. At those two catchment outlets Total Suspended
Solids (TSS) yield was also estimated using automatic water samplers (Bartley et al.,
paired with coincident site turbidity measurements to fit regressions such that
intervals during runoff events. The particle size distribution of stream suspended
Results
The soil cores representing gully wall soil had an average of 30.9% silt and clay
(n=25 samples from 5 soil cores). The clay content increased rapidly with depth
below. Cores from depositional locations at the bottom of hillslopes had a more
uniform depth profile of 30–40% silt and clay (see Supplementary Figure S1). Gully
channel sediment was 1% silt and clay (n=19 samples). Valley segment deposition in
Wheel Creek also resembled gully channel sediment. The dry bulk density of soil
was 1.54 t m-3, and of gully floor deposits was 1.38 t m-3. The total organic carbon
content of soil cores was very low, being 1.3% within 0–20 cm depth, and 0.5% for
20–100 cm depth.
The mean length of all monitored gullies was 300 m. Gully geometry was more
similar between grazed catchments (Weany and Wheel Creek) and between grazing
exclosure catchments (Main and Thorton Creek) than it was between these land use
groups. For example, mean gully length in Main and Thorton Creek catchments was
approximately half that in Weany and Wheel Creek catchments (Table 1). The
longer gullies, in absolute terms and as a proportion of the area upslope of the
downstream gully ends (Table 1). Individual gully headcuts in the grazed Weany and
Wheel Creek catchments tended to have either smaller contributing catchment area
or lower surface gradients than those in Main and Thorton Creek catchments (Figure
3). Catchment vegetation cover was considerably higher in Main and Thorton Creek,
where livestock grazing was excluded or occurred at low levels (Table 1).
The mean percentage of each gully length represented by the head, middle and
valley segments in Weany Creek was 23%, 38% and 39%, respectively (Heine,
2002, Bartley et al., 2007). Gully widths and depths in head segments were similar
between catchments (Table 1), and much smaller than those in some other studied
regions with deeper and finer or less textured soil (e.g., Oostwoud Wijdenes and
Bryan, 2001, Marzolff et al., 2011, Saxton et al., 2012, Shellberg et al., 2016). Depth
was reasonably consistent along each gully, as observed previously in terrain with
convex hillslope profiles (c.f. Rengers and Tucker, 2014). The longitudinal gradient of
gullies in valley segments was relatively consistent across the catchments, typically
within the range 1.5–2%, and gradient tended to slightly decrease downstream within
Weany Creek gullies were ∼55°, ∼36°, and 15°, respectively (Bartley et al., 2007).
The dimensions of individual gullies are provided as supplementary data (Table S1).
Gully headcuts were typically vertical or slightly over-hanging in the upper part of the
incision profile, down to approximately 0.5–1 m, while the lower part of the headcut
face was typically a concave ‘scoop’ shape exposed to the overflow of surface
runoff.
The mean measured rainfall in Weany Creek over the 13 year monitoring period was
8% below the long-term mean (Table 1; Table 2) noting the difference in methods
between these estimates. Runoff at the Weany Creek catchment outlet was 13% of
mean rainfall over this period (Table 2). The four-year period during which gully
erosion was monitored across the four catchments was drier than the 13 years as a
whole, being 54–78% of the long-term mean depending on catchment (Table 3).
The long-term spatial and temporal patterns in gully erosion and sediment yield are
reported first for Weany Creek gullies in this and the following subsections, based on
the 13 years monitoring there. The grazing impacts on gully erosion are presented
Weany Creek gully headcuts eroded at 30–160 mm yr-1 over 13 years (Figure 4a).
Over the sample of gullies, the linear retreat (headcut erosion) rate, was not
and surface slope, or on headcut depth (R2 values < 0.1, reported in Supplementary
Table S2). The product of headcut catchment area and surface slope was used
because surface runoff tends to increase with both variables, and the product has
been used in prior studies (e.g., Saxton et al., 2012). We included each of the 5 head
Gully wall erosion was typically 1–7 mm yr-1, being generally largest at head cross
sections and smaller at middle cross sections (Figure 4b; Table 2). At valley cross
sections the gully walls experienced net deposition in 5 of the 6 Weany Creek
gullies, consistent with their definitions as deposition zones. This deposition occurred
on the walls either side of the active channel, and was more significant in drier years
suggesting it was partly sourced from erosion elsewhere on the valley walls. Fluvial
deposition of sediment from further upstream along the gully would have contributed
was 0.35 mm, which compared favourably with that in previous studies (Boardman et
al., 2015). With each gully segment in each catchment being represented by 20–40
pins, the precision in annual wall erosion for each segment in each catchment was
measured rates and their standard errors based on variability between gullies
Creek gullies (Figure 4c). Gullies with the largest changes were VP0 which aggraded
by 255 mm over the 13 years, and VPXG2 which degraded by 270 mm; the latter
the monitored gully, or by low check dams in that gully whose effects are the subject
with the scour chains typically detecting about 35 mm of fluvial scour and a similar
depth of sediment deposition in each year. Annual changes in the valley floor
Every monitored Weany Creek gully had a positive net sediment yield over the 13
years of monitoring, averaging 11.9 t yr-1 per gully, of which 3.27 t yr-1 was fine
sediment (silt and clay; Table 2). Relative to gully planform area, the mean specific
sediment yield ( SSYGA ; Equation 2) was 22.1 t ha-1 y-1 for total sediment, of which 6.1
t ha-1 yr-1 was fine sediment. While the largest erosion rates occurred at gully head
cuts (Figure 4), wall erosion in the gully head and middle segments made larger
contributions to gully erosion volume and consequently sediment yield, due to the
area over which wall erosion occurred being much larger than the head cut faces
(Figure 4d). The rankings of individual gullies differed between erosion rates and
yields due to differences in gully segment length, width and head depth (reported in
Table S1). For example VPXG2 had the largest head wall erosion rate but the third
contributed from walls in the head and middle segments (total of 85%), secondly
from headcut extension (15%) and almost none from gully floor erosion (Table 2).
Deposition on gully walls in valley segments reduced the net gully fine sediment yield
to 66% of the total amount of fine sediment eroded elsewhere in the gully (from 9.3
to 6.1 t ha-1 yr-1; Table 2). The mean sediment yield of valley wall segments was
negative accounting for the erosion rate and segment length and width of each gully,
although the arithmetic mean erosion rate across the six valley wall segments was
Variables which best (although not strongly) explained variations in total sediment
yield (t y-1) between Weany Creek gullies were the longitudinal gradient of the valley
segment (R2=0.44), the catchment area upslope of each gully head cut (R2=0.20),
and the product of that catchment area and slope (R2=0.39); see Supplementary
Table S2. Variations in wall erosion rates between gullies were inversely correlated
with wall vegetation cover (R2=0.43, 0.97 and 0.92 for head, middle and valley
segments, respectively). However, grazing pressure was similar across all Weany
gullies, leaving the possibility that differences in erosion rate caused the differences
The Weany Creek catchment specific fine sediment yield for the monitoring period,
calculated from stream discharge and TSS concentration monitoring, was 0.84 t ha-1
yr-1. The specific fine sediment yield from the gully area (6.1 t ha-1 yr-1) was 7.3 times
the catchment yield, indicating that gullies are erosion ‘hotspots’ within the
catchment (Table 2). Multiplying gully specific yield by the total area of gullies across
sediment yield during the study period (1,260 t yr-1). Total gully area was estimated
at 59 ha, based on the product of gully density (4.5 km km-2; Heine, 2002),
catchment area (13 km2) and the mean width of monitored gullies (10 m). A second
estimate was also made by multiplying the mean-annual average erosion rate per
monitored gully headcut and head segment combined (2.5 t yr-1) by the 220 gully
headcuts mapped across the Weany Creek catchment (Heine, 2002, Bartley et al.,
2007), giving an estimate of 550 t yr-1 or 44% of the Weany Creek catchment yield.
An alternative assessment of gully sediment yield was also explored, the net fine
specific sediment yield of monitored gullies relative to the area of the gullies and their
contributing catchments ( SSYUCA ). This was estimated at 0.22 ± 0.17 t ha-1 yr-1, or 27
± 20% of the catchment specific yield. This included an estimated 0.06 t ha-1 yr-1
sediment yield from 8 additional unmonitored side branch headcuts across the
monitored gullies. The uncertainty in SSYUCA was the standard error based on
variation between gullies. That gully SSYUCA is smaller than catchment sediment yield
suggests that most of the monitored gullies may be relatively mature, and that the
mean SSYUCA of monitored gullies may under-estimate that of all gullies in the
catchment as a whole.
Annual erosion from the headcuts and head segment walls across the six Weany
Creek gullies was dependent on annual runoff, and to a lesser extent on annual
rainfall (see R2 values in Figure 5). Variable annual deposition in the valley sections
resulted in net gully sediment yield being less dependent on runoff and rainfall.
strongly dependent on rainfall than are sidewall erosion rates (Marzolff et al., 2011).
A wide range of annual rainfall totals were experienced during the 13 years (range
319–1224 mm), with 8 years being below the long-term mean and 5 years being
above the mean of 686 mm yr-1 (Table 1). Net gully sediment yields were negative in
three years. However, headcut and headwall erosion was positive in all years (Figure
5a), with temporary storage on gully floors being visually apparent in dry years.
The sequence of annual variations in the mean fine sediment yield estimated from
erosion pin monitoring across the monitored Weany Creek gullies was mimicked by
the annual variation in catchment fine sediment yield (Supplementary Figure S3).
Weany Creek TSS comprised 70% clay, 20% silt and 10% sand. The sand
Over the shorter 4-year monitoring period common across the four catchments, the
mean rates of gully headcut linear extension in Main Creek were much lower than in
the grazed Weany and Wheel Creek gullies (Table 3; Figure 6a). The Thorton Creek
mean headcut erosion rate was similar to or slightly higher than that in grazed
catchments, despite its mean headcut upslope contributing area being larger (Table
1; Table 3), which suggested that the exclusion of grazing supressed headcut
erosion rates there also, relative to what they might have been under grazing. One
outlier annual headcut linear extension in a Main Creek gully was removed from the
Wheel Creek gullies, while the grazing exclosure Main and Thorton gullies had small
magnitudes of net deposition in middle and valley segments on average (Figure 6b).
Net deposition was also measured on the two Wheel Creek valley segment wall
cross sections, but these were found to poorly represent valley wall erosion in that
catchment. Valley floor deposition was found to extend across much of these pin
cross sections due to wall slopes being very small. Arrays of erosion pins further
from the centre of these segments measured mean-annual erosion of 2.6 mm yr-1
(MV1) and 0.22 mm yr-1 (MV2). Substantial historical wall erosion in Wheel Creek
gullies is also indicated by the widths of these gullies being almost twice those in
Weany Creek (Table 3), and by the low vegetation cover (Table 3), especially during
drought periods (Figure 7a). Therefore, the reported valley segment sediment yield
for those gullies (Table 3, Figure 6d) was determined as the average of yields
calculated from the cross sections and the pin arrays. The deposited material on
Wheel Creek valley cross sections was coarse sand delivered from upstream, which
Differences in wall erosion rates between catchments were inversely correlated with
vegetation cover in the head and middle segments, which is consistent with
established sheetwash and rill erosion behaviour (Renard et al., 1997). The
dependency was weaker in valley segments, which typically had higher levels of
vegetation cover than head or middle segments (Figure 8; Table 3). This is not
surprising given that erosion is less sensitive to cover at higher cover levels. The
positive correlation in valley segments was associated with the deposition measured
in Wheel Creek valley segment cross sections, which was not representative as
from this analysis, since parts of them were subject to temporary grazing exclosures.
Our interest in comparing gully fine sediment yields between catchments was in off-
site water quality impacts at catchment scale, which are influenced by the area of
gullies as well as their erosion rate. Therefore, we compared specific gully fine
sediment yields relative to the area draining to the downstream end of each gully
( SSYUCA ; Equation 3), rather than relative to gully planform area. The total, and
specific, fine sediment yield of gullies ( SSYUCA ) in either grazing exclosure catchment
(Main and Thorton) was less than half that of gullies in either of the grazed (Weany
and Wheel) catchments (Figure 6d; Table 3). The mean gully fine specific sediment
yield across the two grazing exclosure catchments ( SSYUCA =0.071 t ha-1 yr-1) was
77% lower than that across the grazed catchments over the 4 year period
( SSYUCA =0.31 t ha-1 yr-1). These are weighted means accounting for the respective
gully sediment yields (Table 3) and areas (Table 1) in each catchment. The large
difference in measured gully sediment yields between grazed and grazing exclosure
catchments was similar to the difference in longer term catchment specific sediment
yields, which were 62% lower in Main Creek than in Weany Creek over 13 years
(Table 3).
Sediment yields from headcut erosion over the 4 year gully monitoring period were
comparable across all catchments independent of land use, but wall erosion rates
and fine sediment yields were much lower in grazing exclosure catchments (Figure
6). Wall erosion contributed more than 70% of gully fine sediment yield in both
grazed catchments (weighted mean of 86%), which was similar to the 85%
less than 30% of sediment yield in both grazing exclosure catchments (mean of
22%).
Discussion
The results have confirmed that more than 150 years after the introduction of
sediment yield in the study area. The density of the hillslope gullies studied here
makes them important sediment sources at river basin scale. There are about
13,000 km of gullies in the 40,000 km2 Upper Burdekin River catchment (Gilad et al.,
2012), and more than 80,000 km of gullies across the 423,000 km 2 of the GBR
catchments (Thorburn and Wilkinson, 2013). Gully erosion also induces additional
soil surface erosion from persistently denuded areas (scalds) upslope of some gully
persistently denuded areas are known to yield sediment at 3.9 t ha-1 yr-1 (Bartley et
al., 2010a). The concentration of surface runoff by gully erosion is likely to also
increase erosion within the channel network downstream of gullies, including through
channel incision (DiBiase and Whipple, 2011). The significance of gully erosion
found here is supported by previous studies in the Weany Creek catchment which
indicated (based on fewer years data) that the major sources of fine sediment were
2014).
The supply of fine sediment from concentrated hotspots of severe erosion such as
(Wilkinson et al., 2013, Hancock et al., 2014, Wilkinson et al., 2015), which indicate
that approximately 90% of fine sediment yield is derived from sub-surface soil
sources, and that gully erosion is the largest contributor to that. Other GBR basins
are similarly dominated by sub-surface soil sources (Hughes et al., 2009, Olley et al.,
erosion processes and sediment yields in northeast Australia in a similar way to the
The mean-annual rates of gully head retreat in Weany Creek over 13 years were
small relative to a global median of 0.89 m yr-1 (Vanmaercke et al., 2016). Larger and
deeper gullies often have larger rates of expansion and sediment yields than those
at our study sites, including those in deep alluvium in northeast Australia (Shellberg
et al., 2016). However, the gully erosion rates reported here may not be atypical
The observed sensitivity of gully erosion to annual rainfall and surface runoff means
that it will remain a chronic source in this landscape in coming decades. Our finding
that the annual sediment yield of hillslope drainage line gullies depends on annual
We found that gully wall erosion was the largest sediment source in gullies under
grazing landuse, which extends similar findings from gully studies in the grazing
lands of southeast Australia (e.g., Blong et al., 1982, Crouch, 1987). That the widths
of Wheel Creek gullies were larger than those of Weany Creek gullies, where
stocking rates were historically somewhat lower (Table 1), suggests that gully wall
The relative contribution of gully wall erosion to sediment yield tends to increase over
time as gullies become longer and larger in area (Blong et al., 1982), and as the rate
of gully linear extension decays in line with the decline in contributing area above
Sloping gully walls are obviously exposed to direct rainfall impact and rill erosion in
grazed catchments (Figure 2; Figure 7). Larger wall erosion rates occurred at lower
proportions of vegetation cover, particularly at the lower cover levels of head and
middle segments (Figure 8). This is consistent with studies of sheetwash and rill
erosion in GBR catchments and elsewhere (Renard et al., 1997, Silburn et al., 2011).
The prevalence of sloping gully walls also indicates that the coarse-textured soil
(Kirkby and Bull, 2000). The species composition of vegetation is likely to also be
important in determining its effect on wall and floor erosion. Under current conditions,
foliar cover in wetter years, but less persistent ground cover than native perennials
which can be a more important driver of rangeland erosion (Hernandez et al., 2017).
The smaller yields of gullies under grazing exclosure (Figure 6d) are associated with
generally shorter gully lengths, as well as generally lower rates of wall erosion (net
deposition in middle and valley segments). The differences in gully lengths between
cover (Table 1) and function (Roth, 2004), which is consistent with previous studies
of the gully head slope-area threshold (Torri and Poesen, 2014). Presumably
vegetation cover controls headcut erosion rates through its effect on surface runoff
timescales (Figure 5b). Within the grazed Weany Creek catchment, headcut retreat
rates were not obviously related to the headcut catchment area, slope or headcut
depth (Table S2), which is in contrast to several prior studies (Burkard and
Kostaschuk, 1997, Vandekerckhove et al., 2003, Marzolff et al., 2011, Saxton et al.,
2012). However, the differences in these attributes between Weany Creek gullies
were minor relative to the differences between study catchments, and other variables
such as soil properties and terrain convergence also influence headcut erosion.
The concave ‘scoop’ profile of the lower portion of some gully heads indicates that
scour is an important process in this part of the headcut (Rengers and Tucker,
2015), which undercuts the upper portion of the gully heads. It can be expected that
the clay-rich layer 30 cm below the soil surface inhibits infiltration in this and other
duplex soils, increasing the frequency and duration of soil saturation and runoff. As
with the exception of valley segment wall erosion in Wheel Creek gullies. It was
important to monitor multiple gullies, given the variability in erosion rates between
source also reinforces the importance of monitoring headcut, gully wall and floor
erosion to define gully erosion processes and contemporary yield, rather than
Although erosion pins are an established method to monitor channel wall erosion
(Lawler, 1993), the reliance on erosion pins could be regarded as a weakness of this
study. However, reported wall erosion rates are averages in each type of segment
across several gullies in each catchment and over several years, which improves the
represented adequately, with generally higher rates in head segments than middle
segments, and net deposition in most valley segments (Figure 4, Figure 6),
consistent with expected patterns. The main source of uncertainty in mean wall
erosion is therefore the number of gullies monitored and variability between them
(standard errors of 2–4 mm yr-1; Figure 4b), rather than measurement precision. We
acknowledge that a larger number of erosion pins would be required to monitor wall
Marzolff and Poesen, 2009; Koci et al., 2017) and terrestrial laser scanning can
today monitor wall erosion with more comprehensive sampling. However, while such
available at the commencement of this study, and erosion pins measured the small
magnitudes of annual erosion in the study area more precisely than would have
The dependence of gully sediment yield on gully extent and gully vegetation cover
found in this study identifies these as important variables for modelling spatial
patterns in gully yield across river basins. While gully extent is included in the
vegetation cover in models of gully sediment yield using remote sensing or field
measurement datasets is a priority coming from this study, including for representing
Australia also merits further investigation, such as through historical air photographs.
A key question is whether gully extent continues to expand linearly over time since
gully networks were initiated (e.g., Saxton et al., 2012, Shellberg et al., 2016), or
Given the importance of wall erosion to gully sediment yields in grazing land,
increasing vegetation cover within gully features would be an effective approach for
reducing river sediment loads and their associated impacts on freshwater and
Physical revegetation activities such as sowing seed may accelerate the recovery
Increasing vegetation cover on gully floors can also help to reduce gully sediment
transport capacity and enhance deposition (Zierholz et al., 2001), even though the
general stability of most gully floor levels in the study catchments indicates that
controlling bed level is not critical for stabilising gullies. Combined with reducing
grazing pressure, small check dams in strategic locations can deliver large
reductions in sediment yield at the scale of individual gullies and small catchments
(Heede, 1979). Active revegetation of gully walls and floors can also be effective
erosion. In contrast, installing grade control structures in primary gully heads will
address only a small proportion of gully sediment yield in grazed landscapes of the
study area.
manage gully erosion, given the dependence of gully headcut retreat and head
segment erosion on annual runoff (Figure 5). Any form of vegetative cover in the
gully catchment reduces surface runoff by obstructing flow and reducing the sealing
of the soil surface by rain drop impact (Bridge et al., 1983). High soil infiltration
capacity (and hence lower runoff) in the Burdekin basin has been associated with
soil biological activity, particularly presence of worm castings and absence of crusts
with high biomass of perennial native grasses such as those present in Thorton and
Main Creek catchments. At the hillslope scale, a grazing exclosure with these
features had just 40% of the runoff of an adjacent grazed hillslope dominated by the
al., 2008). In contrast, gully sediment yield may increase if more intensive land uses
such as cropping are introduced (Frankl et al., 2012). Reducing landscape runoff will
also reduce the progression of channel incision and widening downstream of the
In the long term lower grazing pressure can benefit forage production (Wilkinson et
al., 2014b) and thus farm profitability (O'Reagain et al., 2011), in addition to
requires constraining the forage consumed each year to within 25% of the amount
grown, or periodically resting pastures during the wetter growing season (Ash et al.,
2011). Accounting for climate variability, and in particular runs of wet and dry years,
is challenging but is critical for the long-term sustainability of grazing (Stafford Smith
et al., 2007).
pasture resting in Weany Creek had no apparent effect on annual hillslope runoff
volumes (Bartley et al., 2014). Gradual increases in the perennial composition and
biomass of forage during that study indicates that long-term recovery of hydrologic
function is possible, but may take up to several decades following land use change
have more rapid effect on gully sediment yields than increasing cover in the
Water quality within the GBR lagoon is responsive to annual variations in fine
sediment inputs (Fabricius et al., 2014). Considering that gully sediment yields are
sensitive to annual rainfall, and that they are well-connected to streams in the
Burdekin River basin (Wilkinson et al., 2013), it could be expected that coastal water
quality would improve within years if gully sediment yields could be reduced across
the basin.
Conclusions
This study has monitored the annual erosion and sediment yield of mature drainage
line gullies in granite-derived soil in northeast Australia. We conclude that more than
150 years after the introduction of livestock grazing, gully networks continue to be a
large sediment source at landscape scale in this area, contributing an estimated 29–
44% of catchment fine sediment yield. We also conclude that wall erosion is the
largest source of fine sediment within gullies in grazed rangeland . Annual gully
sediment yields are strongly dependent on annual rainfall and runoff. The
exclusion (very low stocking rates) were less than half those under grazing.
cover within gully networks is the most effective approach to reducing gully sediment
Physical interventions and planting willbe required in some areas. Broad application
of this approach has the potential to substantially reduce catchment fine sediment
Acknowledgements
We are grateful to the landholders who hosted this research on their properties,
being Rob and Sue Benetto, John Ramsay and family, and the Department of
Defence. This research was funded by the CSIRO in collaboration with the
and Paddock to Reef Program), by Meat and Livestock Australia (B.NBP.0546), and
by Department of Defence (Main Creek and Thorton Creek). Support was provided
by NQ Dry Tropics and The State of Queensland. Soil survey data were provided by
LiDAR DEM of Weany Creek was provided by Dan Tindall (Queensland Department
monitoring was provided by Rex Keen (CSIRO). John Gallant, Alexey Sidorchuk and
Abbott BN, McAllister RRJ. 2004. Using GIS and satellite imagery to estimate the
Ash AJ, Corfield JP, McIvor JG, Ksiksi TS. 2011. Grazing management in tropical
Ayele GK, Gessess AA, Addisie MB, Tilahun SA, Tebebu TY, Tenessa DB,
Langendoen EJ, Nicholson CF, Steenhuis TS. 2016. A Biophysical and Economic
Bartley R, Corfield JP, Abbott BN, Hawdon AA, Wilkinson SN, Nelson B. 2010a.
Bartley R, Corfield JP, Hawdon AA, Kinsey-Henderson AE, Abbott BN, Wilkinson
SN, Keen RJ. 2014. Can improved pasture management reduce runoff and sediment
loss to the Great Barrier Reef? The results of a 10 year study in the Burdekin
Bartley R, Croke J, Bainbridge ZT, Austin JM, Kuhnert PM. 2015. Combining
contemporary and long-term erosion rates to target erosion hot-spots in the Great
DOI: 10.1016/j.geomorph.2006.10.001
Bartley R, Wilkinson SN, Hawdon AA, Abbott BN, Post DA. 2010b. Impacts of
Blong RJ, Graham OP, Veness JA. 1982. The role of sidewall processes in gully
badland sites in the Karoo, South Africa. Earth Surface Processes and Landforms
Bridge NK, Mott JJ, Hartigan RJ. 1983. The formation of degraded areas in the dry
10.1071/SR9830091
Burkard MB, Kostaschuk RA. 1997. Patterns and controls of gully growth along the
shoreline of Lake Huron. Earth Surface Processes and Landforms 22: 901-911. DOI:
10.1002/(SICI)1096-9837(199710)22:10<901::AID-ESP743>3.0.CO;2-O
Chaplot V. 2013. Impact of terrain attributes, parent material and soil types on gully
characteristics of their impact factors in the Yuanmou Valley, China. Catena 125: 82-
induced sediment yield. Science in China, Series D: Earth Sciences 49: 176-192.
DOI: 10.1007/s11430-005-0177-4
https://www.environment.gov.au/marine/gbr/long-term-sustainability-plan
and macroalgae on the Great Barrier Reef. Ecological Applications 20: 840-850.
DOI: 10.1890/08-2023.1
DiBiase RA, Whipple KX. 2011. The influence of erosion thresholds and runoff
variability on the relationships among topography, climate, and erosion rate. Journal
Fabricius KE, Logan M, Weeks S, Brodie J. 2014. The effects of river run-off on
water clarity across the central Great Barrier Reef. Marine Pollution Bulletin 84: 191-
10.1016/j.geomorph.2012.06.011
Furuichi T, Lewis SE, Bainbridge ZT, Olley JM, Wilkinson SN, Burton J. 2016. Paired
geochemical tracing and load monitoring analysis for identifying sediment sources in
a large catchment draining into the Great Barrier Reef Lagoon. Geomorphology 266:
Gilad U, Denham R, Tindall D. 2012. Gullies, Google Earth and the Great Barrier
Reef: A remote sensing methodology for mapping gullies over extensive areas. In
Gomez B, Branbury K, Marden M, Trustrum NA, Peacock DH, Hoskin PJ. 2003.
Gully erosion and sediment production: Te Weraroa Stream, New Zealand. Water
Gordon ND, McMahon TA, Finlayson BL. 1992. Stream hydrology, an introduction
Graf WL. 1977. Rate law in fluvial geomorphology. American Journal of Science 277:
178-191.
Hancock GJ, Wilkinson SN, Hawdon AA, Keen RJ. 2014. The use of fallout tracers
Be-7, Pb-210 and Cs-137 to distinguish the form of sub-surface soil erosion
IAHS Publ. 325, Schmidt J, Cochrane T, Phillips C, Elliott S, Davies TR, Basher L
Hernandez M, Nearing MA, Al-Hamdan OZ, Pierson FB, Armendariz G, Weltz MA,
Spaeth KE, Williams CJ, Nouwapko SK, Goodrich DC, Unkrich CL, Nichols MH,
Holifield Collins CD. 2017. The Rangeland Hydrology and Erosion Model: A Dynamic
Hughes AO, Olley JM, Croke JC, McKergow LA. 2009. Sediment source changes
over the last 250 years in a dry-tropical catchment, central Queensland, Australia.
Hutton LJ, Hartley JS, Rienksl I, P. 1993. Geology of the Charters Towers region. In
Jones DA, Wang W, Fawcett R. 2009. High-quality spatial climate data-sets for
Kirkby MJ, Bull LJ. 2000. Some factors controlling gully growth in fine-grained
10.1016/S0341-8162(99)00077-6
Koci J, Jarihani B, Leon JX, Sidle RC, Wilkinson SN, Bartley R. 2017. Assessment of
Kroon FJ, Kuhnert PM, Henderson BL, Wilkinson SN, Kinsey-Henderson A, Abbott
B, Brodie JE, Turner DR. 2012. River loads of suspended solids, nitrogen,
phosphorus and herbicides to the Great Barrier Reef lagoon. Marine Pollution
Lawler DM. 1993. The measurement of river bank erosion and lateral change: a
10.1002/esp.3290180905
use changes in vegetation cover and sidewall erosion in a gully head of the Penèdes
region (northeast Spain). Earth Surface Processes and Landforms 34: 1927-1937.
DOI: 10.1002/esp.1870
Marzolff I, Ries JB, Poesen J. 2011. Short-term versus medium-term monitoring for
McKeon GM, Day KA, Howden SM, Mott JJ, Orr DM, Scattini WJ, Weston EJ. 1990.
Barrier Reef World Heritage Area. Marine Pollution Bulletin 51: 200-211. DOI:
10.1016/j.marpolbul.2004.11.029
Montgomery DR, Dietrich WE. 1989. Source areas, drainage density, and channel
10.1029/WR025i008p01907
dominates the supply of fine sediment to rivers draining into Princess Charlotte Bay,
10.1016/j.jenvrad.2013.04.010
arid valley floor in Kenya: A case study into temporal variation and sediment
10.1002/esp.225
Petheram C, McMahon TA, Peel MC. 2008. Flow characteristics of rivers in northern
10.1016/j.jhydrol.2008.05.008
environmental change: importance and research needs. Catena 50: 91-133. DOI:
10.1016/S0341-8162(02)00143-1
Prosser IP, Slade CJ. 1994. Gully formation and the role of valley-floor vegetation,
7613(1994)022<1127:GFATRO>2.3.CO;2
Prosser IP, Soufi M. 1998. Controls on gully formation following forest clearing in a
10.1029/98WR02513
Prosser IP, Winchester SJ. 1996. History and processes of gully initiation and
91-109.
Renard KG, Foster GA, Weesies DK, McCool DK, Yoder DC. 1997. Predicting soil
erosion by water: A guide to conservation planning with the Revised Universal Soil
Washington, DC.
Rengers FK, Tucker GE. 2014. Analysis and modeling of gully headcut dynamics,
Rengers FK, Tucker GE. 2015. The evolution of gully headcut morphology: a case
study using terrestrial laser scanning and hydrological monitoring. Earth Surface
Roth CH. 2004. A framework relating soil surface condition to infiltration and
DOI: 10.1002/esp.1104
DOI: 10.1016/j.geomorph.2007.01.009
Saxton NE, Olley JM, Smith S, Ward DP, Rose CW. 2012. Gully erosion in sub-
10.1016/j.geomorph.2012.05.030
influences on alluvial gully erosion along the Mitchell River fluvial megafan.
River fluvial megafan by alluvial gully erosion increased by post-European land use
10.1016/j.geomorph.2016.04.021
Shepherd B. 2010. Changes in gully erosion along the Upper Burdekin River
Society.
Silburn DM, Carroll C, Ciesiolka CAA, deVoil RC, Burger P. 2011. Hillslope runoff
Influences of cover, slope, and soil. Soil Research 49: 105-117. DOI:
10.1071/SR09068
Stafford Smith DM, McKeon GM, Watson IW, Henry BK, Stone GS, Hall WB,
Howden SM. 2007. Learning from episodes of degradation and recovery in variable
Thorburn PJ, Wilkinson SN. 2013. Conceptual frameworks for estimating the water
10.1016/j.agee.2011.12.021
10.1016/j.earscirev.2013.12.006
Valentin C, Poesen J, Li Y. 2005. Gully erosion: Impacts, factors and control. Catena
How fast do gully headcuts retreat? Earth-Science Reviews 154: 336-355. DOI:
10.1016/j.earscirev.2016.01.009
Wasson RJ, Mazari RK, Starr B, Clifton G. 1998. The recent history of erosion and
555X(98)00019-1
Wilcox BP, Huang YUN, Walker JW. 2008. Long-term trends in streamflow from
impacts in large river basins. Science of The Total Environment 468–469: 1210–
Wilkinson SN, Hancock GJ, Bartley B, Hawdon AA, Keen RJ. 2013. Using sediment
Burdekin River basin, Australia. Agriculture Ecosystems & Environment 180: 90-102.
DOI: 10.1016/j.agee.2012.02.002
Sediment source tracing with stratified sampling and weightings based on spatial
gradients in soil erosion. Journal of Soils and Sediments 15: 2038-2051. DOI:
10.1007/s11368-015-1134-2
Wilkinson SN, Stokes CJ, Hawdon AA, Nicholas DM. 2014b. Fat cows improve water
quality: the benefits of higher forage levels in GBR rangelands. In 7th Australian
Townsville; 116-122.
Zierholz C, Prosser IP, Fogarty PJ, Rustomji P. 2001. Instream wetlands and their
significance for channel filling and the catchment sediment budget, Jugiong Ck, New
Table 1. Characteristics of each study catchment and the monitored gullies. Values in bracket
are standard errors based on variability between gullies in each catchment.
relative to the extent of Chromosol soils in the Burdekin basin. The gully probability
mapping is reproduced with permission from Gilad et al. (2012). One soil core
years of above-average rainfall; (a) One of the five headcuts; (b) the head cross
section, note the grey bedrock exposure; (c) middle cross section looking
contributing areas.
survey), (b) mean-annual gully wall erosion by segment (measured by erosion pins);
(c) mean-annual valley segment floor erosion (measured by scour chains); (d) mean
annual yield of total sediment (clay, silt and sand) by gully element, and the net total.
Negative values indicate net deposition. Error bars show the standard error on the
Creek gullies relative to annual rainfall, and runoff measured at the catchment outlet.
Sediment yield is reported relative to the total planform area of monitored gullies.
Parts (a) and (b) show annual yields from headcuts and head segments only. Parts
four catchments with different grazing pressure, for water years 2014–2017: (a)
wall erosion by gully segment (measured by erosion pins); (c) mean-annual valley
segment floor erosion (measured by scour chains); (d) mean-annual fine sediment
yields of gully elements, and the net total. Negative values indicate deposition. Error
bars show the standard error on the mean considering variation between gullies in
each catchment. In (b) the open circle is erosion measured by additional arrays of
Creek gully MV2, showing very low vegetation cover under heavy cattle grazing and
in the foreground a rilled side branch and; (b) looking obliquely upstream across the
valley segment of Main Creek gully MC3, showing perennial grass cover under
livestock exclusion. Photos were taken in May 2015 after three years of below-
average rainfall.
Data points are mean values over four years (2014–2017). Error bars show the
standard error on the mean considering variation between gullies in each catchment.