® .Geomorphology 29 1999 211­233
Anabranching rivers on the Northern Plains of arid central
Australia
Stephen Tooth )
, Gerald C. Nanson
School of Geosciences, Uniersity of Wollongong, Wollongong, N.S.W. 2522, Australia
Received 6 July 1998; accepted 12 January 1999
Abstract
Anabranching rivers are a widespread feature of the Northern Plains in the Alice Springs region of central Australia but
their unusual characteristics previously have not been described. On the Northern Plains, anabranching occurs on rivers
transporting bedloads of coarse sand and gravel and is characterised by channels of variable size and shape which occur
within a broader, typically well-defined, channel-train. Channels are separated by channel-train ridges--narrow, flow-
®aligned, vegetated features--or by wider islands. Ridges and islands are either depositional features formed in situ by
. ® .accretionary processes or erosional features formed by excision from once-continuous areas of floodplain . Vegetation
plays a key role in the initiation, survival and growth of depositional forms through its influence on flow, sediment transport
and ridge and island stability. Anabranching is also related to the influence of tributaries, for some large rivers alternate from
single-thread to anabranching along their length in response to tributary inputs of water and sediment. Tributary inputs occur
during flow events that are either independent from, or in concert with, floods in the trunk channel. Ridges and islands form
in association with tributaries as a result of various hydrological, depositional and erosional processes, including irrigation of
enhanced numbers of in-channel trees and resulting lee-side sediment accretion, floodplain scour, and the formation and
maintenance of deferred-junction tributaries. The change from single-thread to anabranching downstream of tributary
junctions occurs in the absence of any significant change in channel gradient or degree of channel confinement. On the
Northern Plains, anabranching appears to be a stable river pattern that helps to maintain the throughput of relatively coarse
® .sediment in low-gradient typically 0.0005­0.002 channels characterised by an abundance of within-channel vegetation and
subject to declining downstream discharges. q 1999 Elsevier Science B.V. All rights reserved.
Keywords: anabranching; alluvial islands; arid channels; floodplain scour; ridges; riparian vegetation; river pattern; sediment transport;
tributaries
)
Corresponding author. Present address: Department of
Geology, University of the Witwatersrand, East Campus, 1 Jan
Smuts Avenue, Johannesburg, Wits 2050, South Africa. Tel.:
q 27-11-716-2774; Fax: q 27-11-339-1697; E-mail:
065tooth@cosmos.wits.ac.za
1. Introduction
River pattern is a term which describes the plan-
form geometry and implies the processes operating
®within a reach of a river Nanson and Knighton,
.1996 . It represents one means whereby an alluvial
0169-555Xr99r$ - see front matter q 1999 Elsevier Science B.V. All rights reserved.
® .PII: S0169-555X 99 00021-5
( )S. Tooth, G.C. NansonrGeomorphology 29 1999 211­233212
channel can adjust its morphology to prevailing flow
and sediment conditions. Although river patterns
show a continuous variation of form, following the
® .classic work of Leopold and Wolman 1957 they
have traditionally been classified into straight, mean-
dering and braided. However, more recent classifica-
®tions have incorporated anabranching including
. ®anastomosing as a recognisable pattern e.g., Mol-
lard, 1973; Rust, 1978; Schumm, 1981, 1985; Brice,
1984; Knighton and Nanson, 1993; Nanson and
.Knighton, 1996 . A central theme in many of these
classifications is that natural river patterns form a
continuum controlled by interactions amongst a set
of continuously varying factors, such as discharge,
slope, bank strength and sediment load.
Along this continuum of river pattern, many au-
®thors e.g., Slatyer and Mabbutt, 1964; Mabbutt,
.1977; Graf, 1988 consider braided channels to be
particularly characteristic of ephemeral rivers as a
response to common catchment conditions of abun-
dant coarse bedloads, highly variable discharges, and
readily erodible banks. Although these conditions
also apply to many catchments in the Alice Springs
region of arid central Australia, reaches that conform
to traditional descriptions of braided channels are
rare. In this region, rivers tend either to be relatively
low sinuosity, single-thread channels, or to consist of
multiple channels that occur within a broader, some-
times sinuous, channel tract. A preliminary descrip-
®tion of the multiple channel reaches Nanson and
.Knighton, 1996 defined them as `sand-dominated,
® .ridge-forming Type 4 ' anabranching channels.
However, some reaches of channels in central Aus-
tralia also display other forms of anabranching previ-
ously not described.
This paper has three main aims: first, to provide a
comprehensive description of the morphological,
hydraulic and sedimentological characteristics of ana-
branching rivers in the Alice Springs region; second,
to illustrate factors contributing to the formation and
maintenance of anabranching river patterns; and third,
to consider the implications of anabranching for flow
and sediment discharge.
2. Regional setting
® .The `Alice Springs area' Stewart and Perry, 1962
is an arid region in the centre of the Australian
® .continent Fig. 1 . It is dominated by the central
ranges, a west­east trending belt of Proterozoic and
Palaeozoic crystalline and sedimentary rocks which
cuts across the middle of the region. The central
ranges are broken by intermontane lowlands and are
flanked by the extensive Northern Plains and South-
® .ern Desert Basins Mabbutt, 1962 , where uplands
® .are restricted in extent Fig. 1 .
The findings presented here are drawn from study
of rivers on the Northern Plains. Particular attention
is focused on the Sandover, Bundey, Woodforde,
® .Marshall and Plenty Rivers Fig. 1 . Although the
Bundey supplies the vast majority of flow and sedi-
ment downstream of its confluence with the San-
dover, the channel continues as the Sandover River.
Here, it is referred to as the Sandover­Bundey River
in order to avoid confusion with the Sandover River
® .upstream of Ammaroo station Fig. 1 . Rainfall across
the Northern Plains is highly variable but averages
;300 mmryear, and pan evaporation is high
® .;3000 mmryear . As a result, rivers remain dry
for much of the year and only occasionally flow
throughout their length. However, large floods can
® .follow occasional heavy rainfalls )100 mmr24 h
that result from incursions of monsoonal depressions
and cyclones from the north, or from the easterly
passage of frontal weather systems. There are no
rated gauging stations on rivers on the Northern
Plains, but the last few decades have included some
of the largest floods in the 100­110 years of Euro-
®pean settlement in the region Williams, 1970a; Baker
.et al., 1983; Pickup et al., 1988; Pickup, 1991 .
In the upper reaches of river catchments on the
Northern Plains, small, rocky channels transporting
Fig. 1. Ephemeral drainage systems in the Alice Springs region of central Australia, highlighting the main rivers referred to in this paper.
Stippled areas represent land over 750 m in elevation, with these ranges generally rising 100­150 m above the surrounding plains. The
Northern Plains is the extensive area north of the central belt of ranges.
( )S. Tooth, G.C. NansonrGeomorphology 29 1999 211­233 213
( )S. Tooth, G.C. NansonrGeomorphology 29 1999 211­233214
sand and gravel converge on the piedmont to give
rise to well-defined channels flanked by higher allu-
vial surfaces, bedrock or aeolian sandplain. In the
middle reaches, bankfull channel capacities generally
decrease downstream due to declining discharges
resulting from flow transmission losses and flood-
wave attenuation. Channels remain well-defined for
much of their length but, in the lower reaches,
channels eventually disappear and bedload transport
ceases, although occasional large floods continue
across broad, low-gradient, alluvial surfaces known
® .in Australia as `floodouts' Tooth, 1999 . The princi-
pal concerns of this paper are channel forms and
processes in the middle reaches.
3. Anabranching river patterns on the Northern
Plains
For some rivers on the Northern Plains, such as
® .the Bundey Sandover­Bundey , Woodforde and
Marshall, anabranching is the dominant pattern in the
middle reaches. Anabranching is characteristic both
®of large rivers such as the Bundey Sandover­
.Bundey , which anabranches for ;70 km, and
Fig. 2. Schematic illustrations of the typical characteristics of anabranching reaches. Individual channels within the channel-train are
separated either by narrow ridges or broader islands. Definition of individual channels is partly stage-dependent for many ridges or islands
separating channels at low flows may be drowned during higher flows.
( )S. Tooth, G.C. NansonrGeomorphology 29 1999 211­233 215
smaller rivers such as the Woodforde, which
anabranches for ;8 km.
Fig. 2 schematically illustrates typical character-
istics of anabranching channels on the Northern
Plains. The channel-train is the entire complex of
® .channels in the cross-section Fig. 2 and is typically
relatively straight, although it sometimes displays a
series of low-amplitude curves. The margins of the
channel-train are laterally stable and are usually
formed by older, indurated alluvial terraces, often in
association with substantial aeolian sands.
The number of channels within the broader chan-
® .nel-train Fig. 2 varies from reach to reach. In
piedmont reaches there are typically between 2 and 4
channels but further downstream there may be as
many as 15 channels. Channels are separated either
by channel-train ridges, which are relatively narrow
®in relation to their length length to width ratio
.)10 or by islands, which are relatively broad in
® .relation to their length length to width ratio -10
® .Figs. 2 and 3 . This distinction between ridges and
islands is arbitrary, however, and a range of interme-
diate forms occurs. Shapes of ridges and islands vary
®widely in plan view but are typically elongate width
.remains roughly constant along their length or tear-
®shaped broader at their upstream end and tapering
.downstream with long axes aligned parallel to the
® .direction of flow Figs. 2 and 3 . Ridges and islands
® .are generally steep sided bank angles )408 and
composed of sand and mud with a vegetation cover
®of trees principally river red gums, Eucalyptus
.camaldulensis , shrubs and grasses. While the tops
of islands generally approximate or exceed bankfull
level of the channel-train, ridges occur at a wide
®range of elevations up to bankfull bankfull being
defined by the elevation of the floodplains at the
. ® .margins of the channel-train Fig. 4 . Islands can
persist for up to several kilometres downstream but
individual ridges are usually not more than a few
hundred metres long. Hence, individual channels
follow roughly parallel courses for varying distances.
Where ridges have been breached, flow and sediment
transport are often diverted from one channel to
another.
Individual channels within the channel-train vary
widely in morphology, both within a given reach and
between different reaches, and have width­depth
ratios that vary between 10 and 200 and bankfull
capacities that vary from 1 to 200 m2
. Many larger
®channels such as those that have a width­depth
2 .ratio )50 or exceed 60 m in capacity typically
possess a number of small, poorly developed ridges
® .Fig. 4b . However, in reaches where anabranching
is well-developed, channel width­depth ratios gener-
ally range between 20 and 40 and bankfull capacities
are less variable. In many channels, bedrock out-
crops and growth of E. camaldulensis are common.
Bed elevations and bed slopes of adjacent channels
often vary widely such that, where channels of dif-
ferent elevation merge at the downstream end of a
dividing ridge or island, bars of sand and gravel with
avalanche faces up to 2 m high are present. Hy-
draulic calculations for anabranching channel reaches
are difficult because variable elevation of the ridges
and islands makes definition of bankfull problematic
and because large numbers of in-channel and ripar-
ian trees create uncertainties in estimation of rough-
ness coefficients. However, although variable, unit
stream powers typically range between 5 and 30
Wrm2
.
On the Northern Plains, anabranching usually oc-
curs over reaches several kilometres in length before
it becomes less well-defined and the rivers revert to
largely single-thread patterns. On short rivers such as
the Woodforde, reversion to a single-thread form
occurs in the middle to lower reaches, but larger
® .rivers such as the Bundey Sandover­Bundey and
Marshall typically alternate from anabranching to
single-thread at several points through their middle
reaches. For all the anabranching rivers on the
Northern Plains, a comparison of 1950s aerial pho-
tographs with those taken in the 1980s shows that
there has been essentially little channel change in
anabranching reaches during this time, despite a
series of large floods. Changes have probably oc-
curred to individual channels or to intervening ridges
or islands but at a scale finer than can be detected
on aerial photographs. As a whole, therefore, ana-
branching here appears to be a stable channel form.
Anabranching patterns similar to those of the
Bundey, Marshall and Woodforde Rivers are
widespread on the Northern Plains but, surprisingly,
there has been little mention of the unusual charac-
teristics of these channels in the literature. Many
®authors e.g., Mabbutt, 1962, 1967, 1977, 1986;
.Stidolph et al., 1988; Pickup, 1991 generally refer
( )S. Tooth, G.C. NansonrGeomorphology 29 1999 211­233216
® .Fig. 3. Oblique aerial views, illustrating the different styles of anabranching channel on the Northern Plains: a Sandover­Bundey River,
® .showing ridge-form anabranching where channels are separated by linear, vegetated ridges flow direction is from bottom right to top left ;
® . ®b Bundey River, showing island-form anabranching where channels are separated by broad islands flow direction is from top right to
.bottom left .
only to `braided' rivers in their commentaries on the
fluvial geomorphology of the Alice Springs region
and, indeed, the Woodforde River has even been
®cited as `a good example' of a braided river Evans,
.1972, p. 5 . However, while braided rivers are char-
acterised by flow which is separated by bars within
( )S. Tooth, G.C. NansonrGeomorphology 29 1999 211­233 217
® .Fig. 4. Examples of surveyed cross-sections from anabranching reaches of rivers on the Northern Plains showing a ridge-form
® . ® . ® .anabranching Sandover­Bundey River and b island-form anabranching Bundey River . The sections correspond to anabranching Types
® . ® .4 and 2, respectively Nanson and Knighton, 1996 and have a different scale but the same vertical exaggeration 12.5 .
the channel, braid bars are generally poorly vege-
tated, and diamond-, rhomboid- or lozenge-shaped in
plan. These characteristics provide stark contrast with
the well-vegetated, very elongate or tear-shaped
ridges and islands of the Bundey, Marshall and
Woodforde Rivers. In addition, while braid bars are
usually submerged at bankfull stage and are fre-
quently reworked, ridges and islands are often ele-
® .vated to bankfull level Fig. 2 and are very stable.
Hence, these rivers cannot be accurately termed
`braided'. Nevertheless, where channel-train ridges
®are not well developed such as in piedmont reaches
.of the Woodforde River , reaches sometimes appear
closer to braiding with broad, low-elevation, mid-
channel bars dividing the channel, but such reaches
are relatively uncommon.
Similar anabranching patterns to those on the
Northern Plains have also been observed in the
®Kimberley region of Western Australia Wende and
.Nanson, 1998 . In their classification of anabranch-
® .ing rivers, Nanson and Knighton 1996 include the
central Australian and Kimberley examples in their
® .`sand-dominated, ridge-forming Type 4 ' category.
They suggest that the distinguishing characteristics
of this style of anabranching are stable, narrow,
steep-sided, vegetated, sandy ridges that divide the
channel-train into a series of relatively straight or
gently curving, subparallel, narrow and deep chan-
® .nels Figs. 3a and 4a . For ease of discussion, this is
referred to here as `ridge-form anabranching'.
On the Northern Plains, in addition to ridge-form
anabranching, some reaches also display other forms
of anabranching. For instance, a 25 km reach of the
Bundey River downstream of the confluence of its
®three main piedmont tributaries upper Bundey River,
.Frazer and Alkara Creeks is characterised by a
broad belt of channels flowing in a south­north
direction. Between three and five major channels up
to 350 m in width can be identified and are separated
®by large islands up to several kilometres long Figs.
.3b and 4b . This pattern corresponds more closely to
® .the `sand-dominated, island-forming Type 2 ' cate-
®gory of anabranching rivers Nanson and Knighton,
.1996 , referred to here as `island-form anabranching'.
The only Type 2 anabranching river studied to
date is Magela Creek, northern Australia, where
( )S. Tooth, G.C. NansonrGeomorphology 29 1999 211­233218
channels are separated by wide, well-vegetated is-
lands and follow subparallel courses with intermit-
tent, relatively wide and sometimes braided reaches.
However, although the Bundey River has some simi-
larities with Magela Creek, on the Bundey there are
also numerous smaller, often poorly defined channels
separated by low ridges or smaller islands. These
occur both within and adjacent to many major chan-
® .nels Fig. 4b . Hence, the major channels are them-
selves characterised by ridge-form anabranching.
Overall, therefore, this reach of the Bundey River
displays two coexisting styles of anabranching.
The coexistence of two styles of anabranching
® .Types 2 and 4 of Nanson and Knighton, 1996 is
also present to a lesser degree in piedmont reaches of
the Bundey, where ridges sometimes occur within
larger channels separated by wide islands. In addi-
tion, many reaches of the Marshall River display
characteristics very similar to those of the Bundey.
Although the preliminary classification of anabranch-
® .ing channels Nanson and Knighton, 1996 suggested
®that the apparently higher stream powers 15­35
2 .Wrm of Type 4 rivers differentiates them from
® 2 .lower energy 4­8 Wrm Type 2 rivers, data from
the Bundey indicate greater overlap in the stream
powers of these two anabranching styles than previ-
ously has been acknowledged. For example, on the
Bundey, both Type 2 and Type 4 anabranching occur
with bankfull unit stream powers in individual chan-
nels ranging between 1­30 Wrm2
. As knowledge of
anabranching channels expands, the preliminary clas-
® .sification of Nanson and Knighton 1996 will un-
doubtedly be further modified.
4. Factors contributing to anabranching on the
Northern Plains
River patterns on the Northern Plains vary widely
between adjacent catchments. The anabranching pat-
terns of the Bundey and Woodforde Rivers provide
stark contrast with the neighbouring single-thread
Sandover River and while anabranching is character-
istic of many reaches of the Marshall, the adjacent
Plenty River is largely single-thread. These contrast-
ing river patterns are not related to differences in
channel gradients, for bed slopes along all these
rivers are very similar, typically declining from
around 0.002 in the piedmont to around 0.001 or less
through the middle reaches. However, the calibre of
transported bed material varies widely due to differ-
ences in catchment lithology, tributary gradients and
the proportion of tributary-derived drainage. For the
anabranching Bundey, Woodforde and Marshall
Rivers, which drain catchments comprised mainly of
coarse-grained crystalline rocks and receive addi-
tional sediment supplies from moderate- to high-
gradient tributaries in their middle reaches, bed ma-
terial mainly consists of coarse sand to granule gravel,
with median grain size typically 0.5 to y0.5 phi
® . ® .0.71 to 1.4 mm Fig. 5 . For the single-thread
Sandover and Plenty Rivers, which drain catchments
with finer-grained crystalline rocks and receive very
limited sediment supplies from tributaries in their
middle reaches, bed material is finer and mainly
consists of medium to coarse sand, with median
® .grain size typically 2 to 0.5 phi 0.25 to 0.71 mm
® .Fig. 5 . This difference in bed material calibre
appears to be a key factor contributing to the contrast
in river patterns on the Northern Plains, with
anabranching characteristic of rivers transporting rel-
atively coarse bedloads.
Fig. 5. Grain size curves for bed material from rivers on the
Northern Plains. Anabranching patterns are characteristic of those
rivers transporting bedloads of coarse sand and gravel. Each curve
is the average of three samples.
( )S. Tooth, G.C. NansonrGeomorphology 29 1999 211­233 219
Anabranching can also be directly related to pat-
terns of tributary drainage, for both the Bundey
® .Sandover­Bundey and Marshall Rivers alternate
from single-thread to anabranching along their length
in response to tributary inflows. This occurs either
where one small tributary joins the river or where
several tributaries join the river within a short reach.
Fig. 6 illustrates the relationship between tributary
inflows and anabranching in the middle reaches of
the Sandover­Bundey, clearly showing how the
number of channels increases for short distances
below tributary junctions before the channel reverts
to a largely single-thread form. This increase in the
number of channels downstream of tributary junc-
tions occurs in the absence of any significant change
in channel gradient or degree of channel confine-
ment.
In the catchments of the Bundey, Woodforde and
Marshall Rivers, tributaries vary in character from
those transporting coarse sand and gravel as bedload
to those transporting mainly suspended load, yet
ridge- or island-form anabranching can result across
this range. Hence, the processes responsible for the
formation of channel-train ridges and islands may be
tributary-related, evidence for which comes from the
sedimentology of ridges and islands.
5. Sedimentology of ridges and islands
Examples of excavated sections from typical
ridges and islands are presented in Fig. 7. In all
instances, basal coarse sand and granulerpebble
gravel is overlain by a finer overburden varying from
coarse sand through to mud. The top of the basal
sand and gravel is generally elevated above the
surface of the adjacent channel bed. For many is-
lands and floodplains, as well as for a smaller num-
ber of ridges, the finer overburden shows little textu-
ral differentiation or preserved sedimentary struc-
tures. For many other ridges, however, the overbur-
den consists of separate sedimentary units, although
the number, texture and thickness of units vary both
from ridge to ridge and even along the same ridge
® .Fig. 7 . Boundaries between units are usually indis-
tinct with textural changes gradational but in rarer
instances erosional contacts can be identified. Lenses
consisting of partially-decayed leaves and twigs are
Fig. 6. Relationship between tributary inflows and anabranching in the middle reaches of the Sandover­Bundey River, based on assessment
® .of the number of channels at 1 km intervals determined from field surveys and aerial photographs . Although defining the number of
channels across a given section involves a degree of subjectivity, the data nevertheless demonstrate how the number of channels increases
for a short distance downstream of the tributary junctions, resulting in ridge-form anabranching, before the river reverts to a largely
single-thread channel with only occasional ridges and islands.
( )S. Tooth, G.C. NansonrGeomorphology 29 1999 211­233220
Fig. 7. Excavated sections illustrating the typical morphology and sedimentology of ridges and islands. Trenches were excavated
perpendicular to the downstream direction and sediments were described from the face parallel to flow.
( )S. Tooth, G.C. NansonrGeomorphology 29 1999 211­233 221
often found high in the successions but, given rapid
rates of organic decay in the region, these organics
are thought to be too young to be usefully dated by
radiocarbon techniques.
Sedimentary units typically dip from the centre of
ridges towards adjacent channel beds, with the angle
® .of dip generally between 308 and 608 Fig. 8 . In
such instances, ridges are composed of a central core
of coarse sand and granulerpebble gravel which
decreases in thickness towards the ridge margins,
with overlying units generally following the same
pattern. Cross-cutting by overlying units can occa-
sionally be identified, sometimes with textural rever-
sals. Due to bioturbation resulting from growth of
colonising trees, shrubs and grasses, sedimentary
structures are rare but, where preserved, fine-medium
sand shows either weakly developed trough cross-
beds or opposed planar cross-beds with the foresets
® .generally meeting at the ridge centre Fig. 8 .
In some instances, sedimentary units in ridges dip
below the levels of adjacent channel beds and can be
traced up to 1 m away from the ridge margins to
Fig. 8. Photograph and interpretative sketch illustrating the sedimentology of a typical channel-train ridge on the Sandover­Bundey River
® . ® .view into centre of ridge, flow direction towards camera . A central core of structureless coarse sand and granulerpebble gravel unit 1 is
® . ® .overlain by fine to medium sand with planar cross-beds unit 2 , structureless silty sand unit 3 and fine to medium sand with trough
® .cross-beds unit 4 . Units dip towards the ridge margins, with clear signs of interfingering between units 1 and 2. Contacts between units are
both gradational and erosional. The left-hand side of the scale is marked in centimetres and the right-hand side in inches.
( )S. Tooth, G.C. NansonrGeomorphology 29 1999 211­233222
depths of around 15­20 cm below the bed surfaces.
Interfingering between the silt and sand of ridges and
adjacent channel sand and gravel can sometimes be
found. For ridges and islands showing little internal
textural differentiation, however, ridge silt and sand
appears to underlie channel sand and gravel and to
join with neighbouring ridges or floodplains.
Although details of the sedimentology vary even
within any one ridge or island, some general conclu-
sions can be drawn. The complex sedimentology of
many ridges and islands strongly indicates that they
have mainly formed in situ by accretionary processes
® .i.e., they are depositional features . For instance,
interbedded fine­medium sand and silty sand identi-
® .fied in some ridges Fig. 7 may represent deposi-
tional couplets resulting from separate flood events
or from pulses on an individual flood wave. How-
ever, where ridges and islands show little internal
textural differentiation or where their sedimentology
and roughly horizontal sedimentary units differ little
from that of adjacent floodplains, indications are that
they have been cut from areas of formerly continu-
® .ous floodplain i.e., they are erosional features . The
external morphologies of depositional and erosional
ridges and islands are similar and are not readily
distinguishable. Both depositional and erosional types
can occur within a given reach and across a given
section, suggesting that anabranching patterns as a
whole result from a number of depositional and
erosional processes.
While the depositional or erosional origins of
relatively narrow ridges and smaller islands can be
established from their sedimentology, the origins of
larger islands found in some anabranching reaches
are more difficult to ascertain. Due to limited strati-
graphic exposures of larger islands, it is unclear
whether they are primarily depositional or erosional
® .in origin. Anabranching including anastomosing
has usually been defined with reference to islands
®excised from floodplains e.g., Knighton and Nan-
.son, 1993 but recent acknowledgement has been
given to the possibility of islands formed by within-
®channel deposition Nanson and Knighton, 1996; see
.also Sarkar and Basumallick, 1968 . In many cases,
it is hard to establish the origins of islands character-
®ising certain anabranching channels Nanson and
.Knighton, 1996 but, as the following discussion
indicates, it is likely that examples of both large
depositional and large erosional islands can be found
in anabranching channels on the Northern Plains.
6. Formative processes of ridges and islands
The broad distinction between primarily deposi-
tional and primarily erosional forms provides a use-
ful basis for a discussion of the formative processes
of ridges and islands comprising anabranching river
patterns.
6.1. Depositional forms
A number of studies in various fluvial contexts
have documented how flow separation induced by
obstacles such as bridge piers, stranded ice blocks,
boulders and riparian vegetation can result in a vari-
ety of depositional and erosional features collectively
®termed `obstacle marks' Dzulynski and Walton,
.1965; see Allen, 1982, for a summary . Depositional
features are generally known as `current shadows'
®and erosional features as `current crescents' Allen,
.1982 .
Along channels in the Alice Springs region,
growth of in-channel trees, such as E. camaldulen-
sis, is characteristic of many reaches. These trees act
as substantial obstacles and influence local patterns
of flow and sediment transport. Hence, for
anabranching rivers on the Northern Plains, many
®depositional ridges and smaller islands width -10
.m probably originate as current shadows on the
downstream side of in-channel trees. Indeed, the
morphological and sedimentological characteristics
® .of depositional ridges and islands Figs. 7 and 8 are
consistent with current shadow formation in two
main respects. First, ridges and islands range in size
from low, short and narrow up to elevated, lengthy
and wide forms. The sedimentology of different size
depositional ridges and islands shows no consistent
differences, suggesting that small ridges can grow in
size to become larger ridges or islands. Second,
where preserved, internal sedimentary structures are
similar to small current shadows previously de-
®scribed in the literature e.g., Karcz, 1968; Williams,
.1970b; Picard and High, 1973; Sneh, 1983 , some-
times containing opposed foresets of fine­medium
sand.
( )S. Tooth, G.C. NansonrGeomorphology 29 1999 211­233 223
Although ridges and islands may be initiated as
® .current shadows, the great length )100 m and
® .height )2 m of many depositional ridges and
islands suggest that other factors are important in
their continued growth and maintenance, particularly
subsequent colonisation by vegetation, such as
grasses or young saplings. The contribution of in-
channel and riparian vegetation to the stabilisation
and subsequent growth of fluvial deposits has been
®widely reported in a number of contexts e.g., Hadley,
1961; Mabbutt, 1977; Graf, 1978, 1979; Woodyer et
al., 1979; Osterkamp and Costa, 1987; Pickup et al.,
1988; Thorne, 1990; Graeme and Dunkerley, 1993;
.Brooks and Brierley, 1997 , where its influence ap-
pears to be twofold: first, the binding influence of
roots on freshly deposited sediments provides protec-
tion from scour during later floods; and second, by
increasing roughness and reducing flow velocities,
sediment deposition is induced.
On the Northern Plains, although extensive root
networks of colonising trees and shrubs on ridges
and islands are likely to provide considerable local
protection from scour, more extensive protection is
provided by growth of grasses such as Silky Brown-
® . ®top Eulalia fula , Desert Bluegrass Bothriochloa
. ®ewartiana , Queensland Bluegrass Bothriochloa eu-
. ® .rantiara , and Mitchell grass Astreleba pectinata .
Many of these species are tough, stout perennials,
forming dense tussocks and sometimes providing
thick root mats. Hence, they are likely to provide
considerable cohesion for the predominantly sandy
deposits forming ridges and islands. Grass densities
on ridges and islands vary considerably in response
to drought and grazing pressure but, following the
® .method of Dunaway et al. 1994 , the volume of
roots in the top 4 cm of bank sediment for six
samples from the well-grassed Woodforde River and
®Frazer Creek a piedmont tributary of the Bundey
.River ranged from 10­30%. These values are com-
parable to the volume of grass roots in bank sedi-
®ment reported for other fluvial systems Smith, 1976;
.Dunaway et al., 1994 where grasses have been
shown to significantly reduce bank erosion.
The sedimentology of depositional ridges and is-
® .lands Figs. 7 and 8 and the influence of the colonis-
ing vegetation suggest that ridges and islands grow
as a result of three main contemporaneous processes:
downstream progradation, vertical accretion and lat-
®eral accretion. Oblique accretion Nanson and Croke,
.1992 plays a more limited role in their growth. The
initial deposit is typically a small ridge of sand and
granulerpebble gravel, usually on the downstream
® .side of a tree Fig. 9a . Subsequent colonisation by
young saplings encourages deposition of finer sand
and silty sand, representing high saltation or sus-
pended load. Continued growth of the saplings and
possible colonisation by grasses encourages further
® .longitudinal, vertical and lateral growth Fig. 9b of
the ridge or incipient island. Where the cover of
saplings or grasses is sparser, silty sand or mud
deposited during waning flood stages may provide
some protection from the erosive effects of subse-
quent flood flows. These finer-grained deposits,
sometimes containing up to 60% silt and clay, occur
either as individual laminae or coupled with bands of
® .fine to medium sand Fig. 7 . The end result of these
® .processes is a ridge or small island Fig. 9c consist-
ing of a core of sand and gravel enveloped by an
assemblage of finer-grained strata.
In addition to the growth of individual ridges and
islands, flood flow interaction with the whole com-
plex of ridges and islands has to be considered. As a
series of ridges and islands grows, it is possible that
®those in favourable locations in the channel such as
.in the lee of other ridges and islands may survive
and grow, while those in less favourable locations
® .such as between already well-developed ridges are
more likely to be eroded. It is also possible that once
a series of islands and ridges is developing, then
helical secondary flows in the channels may encour-
age further growth by scouring intervening channel
floors and transporting sediment onto neighbouring
ridges or islands, in the manner suggested by flume
®experiments e.g., Nezu et al., 1985; Nezu and Naka-
.gawa, 1989; Tsujimoto, 1989 , and field descriptions
® .of the formation of scroll bars Nanson, 1980, 1981
and sandy channel-train ridges in monsoonal north-
® .ern Australia Wende and Nanson, 1998 . However,
as flow structures during floods in the anabranching
channels are unknown at present, this suggestion can
remain only speculative.
Growth rates of ridges and islands are largely
unknown. In the ridges and islands examined there
was little organic material suitable for dating by
radiocarbon techniques. Nevertheless, the occasional
® .presence of thick ;0.5 m cosets of sedimentary
( )S. Tooth, G.C. NansonrGeomorphology 29 1999 211­233224
( )S. Tooth, G.C. NansonrGeomorphology 29 1999 211­233 225
structures such as trough and planar cross-bedding in
® .some ridges and islands Fig. 8 suggests sometimes
fast rates of vertical and lateral growth during flood
events. However, while growth of many smaller
ridges clearly can be related to colonisation by pre-
sent-day vegetation, the distribution of trees growing
on top of many larger ridges and islands suggests
that initial formation considerably pre-dates the pre-
sent colonising vegetation. For instance, on larger
ridges and islands, there is often little evidence of
any down-ridge succession in the age or size of trees,
as might occur if the present vegetation had estab-
®lished in association with growth of the features cf.
.Osterkamp and Costa, 1987 . As the age of many
ridges and islands may exceed the maximum life-span
®of trees such as E. camaldulensis ;400 years,
.Ogden, 1978 , trees initiating the features have prob-
ably died and been replaced by younger trees.
6.2. Erosional forms
For ridges and islands excised from once-continu-
ous areas of floodplain, explanation of formative
processes lies in a consideration of causes and pro-
cesses of channel avulsion. A number of studies have
shown how channel avulsion is a key factor in the
formation and maintenance of anabranching and
anastomosing river patterns, with channel cutting
involving crevasse splay evolution or processes of
®floodplain scour e.g., Fisk, 1952; Popov, 1962;
Smith and Smith, 1980; Schumann, 1989; Smith et
al., 1989; Brizga and Finlayson, 1990; McCarthy et
.al., 1992; Schumm et al., 1996 .
For anabranching channels on the Northern Plains,
it is difficult to determine the exact erosional pro-
cesses by which any one ridge or island has been
excised from floodplain, as the evidence remaining
after excision usually does not include clues as to the
origin of the episode. Nevertheless, field observa-
tions provide little evidence of channel cutting by
progressive splay development. Along anabranching
reaches, there are few examples of breaches in chan-
®nel banks or `blind anabranching channels' Harwood
.and Brown, 1993 that would occur if splays or
channels were progressively extending through
floodplains. Well-vegetated banklines, general lack
of levees and lateral confinement of channels and
floodplains by alluvial terraces or aeolian sands all
limit the potential for splay development.
In contrast, numerous examples of floodplain
scour can be found in anabranching reaches. Scours
are initiated by overbank flows which result either
from exceedance of bankfull level or by damming
and local ponding of tributary floodwaters by flows
in the trunk channel. Scours generally occur some
5­10 m from channel margins behind prominent
®stands of bankline trees and form small 0.1­0.2 m
.deep , elliptical pits or elongate depressions incised
®on floodplain surfaces and oriented downstream Fig.
.10a . Field observations suggest that these scours
represent the early stages of channel formation with
subsequent channel cutting proceeding by a process
termed linear dissection of floodplains. Linear dis-
section involves selective floodplain erosion along a
preferred line of scour leading to excavation of an
elongate furrow. Continued deepening and extension
of the furrow in the upstream and downstream direc-
® .tions eventually results in channel cutting Fig. 10b .
Subsequent routing of flow and bedload sediment
through this newly excavated channel incorporates it
into the broader anabranching system. Hence, exten-
sive channel cutting by this process results in
anabranching channels separated by relatively nar-
row, elongate, ridges or islands composed of former
® .floodplain Fig. 10c .
There are three main factors promoting linear
dissection of floodplains along anabranching reaches:
first, well-vegetated banklines which protect banks
from laterally extensive stripping and preserve that
part of the floodplain as a ridge or island; second,
confinement of overbank flows over the floodplain
by adjacent alluvial terraces or aeolian sands; and
third, the tendency for floodplain sediments to read-
ily disaggregate in the presence of water. Using the
® .Emerson 1967 Aggregate Test, field assessments of
® .the structural stability of soil aggregates 3­5 mm
.Fig. 9. Series of photographs illustrating growth of depositional ridges and islands: a small current shadow of sand and granules in the lee
® . . ®of a river red gum flow direction is from top right to bottom left ; b small, elongate ridge colonised by grasses flow direction is from top
. . ® .right to bottom left ; c well-developed ridge colonised by large river red gums and shrubs flow direction is towards camera .
( )S. Tooth, G.C. NansonrGeomorphology 29 1999 211­233226
( )S. Tooth, G.C. NansonrGeomorphology 29 1999 211­233 227
showed that in all cases floodplain sediments are
characterised by slaking: aggregates collapsed upon
immersion in water and the water remained clear
® .after 15 min Chan and Abbott, 1993 . The tendency
for floodplain sediments to readily slake in water
results from the relatively low percentage of silt and
® .clay typically -40% which means that unvege-
tated, or sparsely vegetated, sediments have little
cohesion.
Once a preferred line of floodplain scour is initi-
ated, erosion may proceed relatively rapidly, for
scours provide semi-channelised pathways for subse-
quent overbank flows. In some reaches, where shal-
low scours on floodplain surfaces intersect deeper
scours or anabranches, knickpoints up to 30 cm high
are sometimes found at the junction. Hence, an
® .explanation similar to that of Schumann 1989 may
apply in these instances, with further channel cutting
resulting from knickpoint migration along the line of
the scour.
7. The role of tributaries in contributing to
anabranching
Earlier discussion indicated how anabranching on
the Northern Plains is related to patterns of tributary
drainage. Given the different styles of anabranching
® .ridge- and island-form , the different character of
®tributary channels varying from those transporting
relatively coarse bedload to those transporting mainly
.suspended load and the different depositional and
erosional processes involved in ridge and island for-
mation, there is probably a range of mechanisms
whereby tributaries contribute to the formation of
anabranching river patterns. However, field investi-
gations and anecdotal accounts of local pastoralists
indicate that the tributary influence operates under
two main sets of circumstances: first, through inputs
of water and sediment to a dry trunk channel either
prior to, or in the absence of, flows originating
®further upstream `asynchronous tributary flow',
.Thornes, 1977 ; and second, through inputs of water
and sediment to a wet trunk channel during periods
®of flood flow generated further upstream `fully inte-
.grated flow', Thornes, 1977 . In either circumstance,
depositional and erosional ridges and islands can
form as a result of the processes outlined above.
7.1. Asynchronous tributary flow
Although there are no gauged data from which to
assess the frequency or timing of flows in tributaries
in relation to those in trunk channels, the anecdotal
evidence suggests that flows emanating from tribu-
taries often result in trunk channels flowing down-
stream of the junctions, while reaches further up-
stream remain dry. Bankfull capacities of discharges
for tributary channels are usually only minor in
comparison to trunk channel capacities and thus
®tributary inflows only result in low flow sub-bank-
.full events in trunk channels. Nevertheless, it is
possible that the more frequent supply of moisture to
sections of channel immediately downstream of trib-
® .utary junctions `channel-bed irrigation' encourages
denser growth or greater numbers of in-channel trees
which, by acting as obstacles to flow, initiate deposi-
® .tional ridges as current shadows Fig. 9a . However,
the influence of channel-bed irrigation on densities
or numbers of in-channel trees is difficult to ascer-
tain, for although aerial photographs show that many
anabranching reaches are characterised by greater
numbers of trees than single channel reaches, this is
partly a result of large numbers of trees growing on
top of ridges and islands.
Regardless of the influence of moisture supply on
numbers of in-channel trees, once depositional ridges
and islands are initiated, their growth depends on
survival during subsequent flood events. The more
frequent, low-magnitude flows experienced by chan-
nel reaches downstream of tributary junctions may
enable continued growth of ridges through processes
of downstream progradation and vertical and lateral
® .accretion Fig. 9b­c . Furthermore, sediment supply
® .Fig. 10. Series of photographs illustrating channel formation and excision of ridges and islands by linear dissection of floodplains: a
® . ® .elongate scours on a sparsely-vegetated floodplain surface flow direction is away from camera ; b small, shallow channel excavated in
® . ® . ®once-continuous floodplain flow direction is away from camera ; c series of channels and ridges excavated from floodplain flow
.direction is from right to left . The tops of the ridges are accordant and their sedimentology is indistinguishable from that of the adjacent
floodplain.
( )S. Tooth, G.C. NansonrGeomorphology 29 1999 211­233228
may also be an important factor in the survival and
growth of ridges and islands, especially if tributaries
supply predominantly fine-grained suspended mate-
® .rial Fig. 11a . For instance, during low flows ema-
nating from these tributaries, a veneer of silt and
mud may be deposited on the small, developing
® .Fig. 11. Schematic diagrams illustrating typical situations in which anabranching develops in association with tributary junctions: a
® .formation of ridges and islands by accretion in the lee of in-channel trees; b formation of ridges and islands as a result of
® .deferred-junctions where small tributaries attempt to enter a larger trunk stream; c formation of ridges and islands as a result of
® .deferred-junctions at the confluence of two or more similar-size tributaries. In a tributaries usually supply fine-grained suspended loads but
® . ® .in b and c tributaries supply bedloads of sand and gravel.
( )S. Tooth, G.C. NansonrGeomorphology 29 1999 211­233 229
ridges, thus providing a cohesive coating that pro-
tects them from erosion during subsequent larger
flood events generated upstream. However, water
and fine sediment supplied by tributaries are limited
and are gradually exhausted downstream of tributary
junctions. Here, ridges no longer have the cohesion
to survive larger flood events and the channel reverts
® .to a single-thread form Fig. 11a .
7.2. Fully integrated flow
The main process of anabranch formation during
periods of overbank flow is cutting of erosional
ridges and islands by linear dissection of floodplains
® .Fig. 10a­c . This often occurs where high flows in
the trunk channel cause local ponding of tributary
® .floodwaters `hydrologic damming' which results in
aggressive periods of flow on floodplains adjacent to
the trunk channel. Linear dissection of floodplains
leads to excision of ridges and islands from flood-
plain, resulting in either ridge- or island-form
anabranching.
In other instances, frequent damming of tributary
inflows by high flows in the trunk channel over time
deflects the mouth of the tributary channel several
hundred metres downstream. As such, many
anabranches take the form of deferred-junction tribu-
taries, separated from the trunk channel by large
islands excised from floodplain or developed by
within-channel deposition. Deferred-junction tribu-
taries are commonplace along many anabranching
rivers on the Northern Plains, and usually occur
where a small tributary attempts to join with a larger
trunk channel. They have some similarities with the
® .examples described by Nanson et al. 1993 from
Magela Creek in northern Australia, but do not result
from differential aggradation of tributary and trunk
channels. The vast majority of tributary channels on
the Northern Plains are graded to the level of the
trunk channels. Typical examples of deferred-junc-
tion tributaries are illustrated in Fig. 11b. They com-
monly occur where the attempted junction angle of
tributary and trunk channel is close to a right-angle,
resulting in pronounced downvalley deflection of the
® .lower section of the tributary Fig. 11b, location 1 .
In other instances, the tributary joins an anabranch of
the trunk channel dividing around a large island,
with the head of the island lying upstream of the
® .tributary mouth Fig. 11b, locations 2 and 3 . Expla-
nations for this may involve migration or avulsion of
the tributary mouth from an original position up-
stream of the island head or alternatively, that by
acting as an obstacle to flow, continued deposition at
the island head results in upstream growth to leave
the tributary connected to an anabranch.
In addition to formation of deferred-junction trib-
utaries, anabranches can also develop downstream of
® .the confluence of similar-size channels Fig. 11c . A
prime example is found on the middle Bundey River,
where the upper Bundey River, Frazer Creek and
Alkara Creek converge at a break in a sandstone
range. Despite acute junction angles between the
three channels, beyond the range they do not merge
as one single channel but instead follow subparallel
courses as a broad belt of anabranching channels
® .dividing around large islands Figs. 3b and 4b ,
before eventually merging into a single channel ;25
km downstream. As in the case of deferred-junction
tributaries, frequent ponding of floodwaters by high
flows in neighbouring channels over time might
result in deferral of this type of channel junction.
The origins of the large islands separating the chan-
nels is uncertain but they may form either by exci-
sion from floodplain or by within-channel deposi-
tion.
Earlier discussion demonstrated how anabranch-
ing is characteristic of channels transporting rela-
tively coarse-grained bedload, such as the Bundey,
Woodforde and Marshall Rivers. This is supported
by the fact that deferred junctions of both dissimilar-
and similar-size channels are uniquely associated
with channels transporting bedloads of coarse sand
and gravel. Hence, as for tributaries supplying sus-
® .pended loads to the trunk channel Fig. 11a , bedload
sediment supply may also be an important explana-
tory factor in anabranch development. Possible ex-
planations for this may be found by considering the
implications of anabranching for flow and sediment
discharge.
8. Implications of anabranching for flow and sedi-
ment discharge
The close association between anabranching river
patterns and coarse bedloads on the Northern Plains
invites three related questions. First, what are the
implications of anabranching for flow and bedload
( )S. Tooth, G.C. NansonrGeomorphology 29 1999 211­233230
discharge in these low-gradient catchments ? Second,
do anabranching channels possess any advantages
over other channel forms for the conveyance of flow
and bedload ? Third, if so, how do these advantages
help to explain the maintenance of anabranching
channels on the Northern Plains?
For rivers on the Northern Plains, decreasing
downvalley flow volumes, flood peaks and channel
gradients mean that sediment transport capacity must
also decline downstream. As a result, sediment dis-
charge is transport-limited rather than supply-limited.
® .Recently, Nanson and Knighton 1996 have argued
that a fundamental advantage that anabranching
channels possess over single channels is that, in
situations where it is not possible to increase channel
gradient, division of a single-thread channel into two
or more anabranches concentrates flow and increases
velocity and bed shear stress, which enables the
system to transport water and sediment more effi-
ciently. Through analysis of some basic hydraulic
relationships for alluvial channels which include flow
continuity, roughness and several sediment transport
® .functions, Nanson and Huang 1999 have extended
this argument further. They demonstrate quantita-
tively that conversion from a single to an anabranch-
ing channel through the formation of ridges or is-
lands results in a reduction of total flow width and
an increase in flow depth, mean flow velocity, and
bed shear in each anabranch to levels sufficient to
maintain or enhance water and sediment throughput
without increasing gradient.
The hydraulic and sedimentary advantages that
anabranching appears to confer over single-thread
channels helps to explain anabranching on the North-
ern Plains. In these rivers, there are limited opportu-
nities for increasing gradient in order to maintain or
increase downstream sediment flux, as a result of
several factors. Channels are generally already of
very low sinuosity, thus reducing the possibility of
increasing gradient by further reducing sinuosity.
Furthermore, many channel beds are positioned just
above bedrock and they transport water and sediment
towards stable or gradually aggrading basins, factors
which reduce the possibility of increasing gradient
by downcutting. For rivers such as the Sandover and
Plenty, which transport relatively fine-grained bed-
® .load Fig. 5 , wide and shallow, single-thread chan-
nels are sufficient to transport the supplied bedload
at a rate necessary to maintain stable bed levels.
However, for rivers such as the Bundey, Marshall
and Woodforde, which transport relatively coarse
® .bedload Fig. 5 , a similar channel form appears to
be insufficient to transport bedload at the rate re-
quired over time to maintain stable bed levels. This
is especially likely to be the case where tributaries
provide a supply of water that encourages growth of
obstructing in-channel trees, or where they introduce
®additional supplies of coarse sand and gravel Fig.
.11a­c . However, if a system of ridges and islands
becomes established downstream of tributary junc-
tions, this provides both a temporary store for sedi-
ment and enables flow and bedload to be routed
through a system of relatively narrow and deep
anabranching channels. As a result, total flow width
® .is reduced commonly by up to 20% and velocity,
bed shear stress and rates of bedload transport in
anabranches may be increased to levels sufficient to
ensure onward transport of sediment and mainte-
nance of bed levels.
Adjustments to the flow system which enable the
maintenance of bedload transport will be subtle, for
® .as Nanson and Huang 1999 concede, hydraulic
adjustments more complex than those proposed in
their simple model probably restrain the capacity of
the channels to move significant additional sediment.
A major factor influencing this restraint is likely to
be the extra roughness generated by large numbers
of trees lining the ridges and islands in anabranching
reaches, which would limit predicted increases in
flow velocity and sediment discharge. For rivers on
the Northern Plains, a lack of flow gauging data and
difficulties in assessing roughness in heavily-treed
® .channels Graeme and Dunkerley, 1993 means that
changes in roughness through anabranching reaches
are largely unknown at present. Similarly, whilst a
lack of bedload transport measurements makes it
impossible to test these assertions using field data,
® .the model results of Nanson and Huang 1999 pro-
vide a logical, possible explanation for anabranching
on the Northern Plains.
9. Conclusion
River patterns on the Northern Plains of the Alice
Springs region range from single-thread to ana-
branching due to contrasts in bed material calibre
( )S. Tooth, G.C. NansonrGeomorphology 29 1999 211­233 231
and patterns of tributary drainage. On the Sandover
and Plenty Rivers, which transport bedloads of
medium to coarse sand, channels are typically
single-thread and relatively wide and shallow. On the
Bundey, Woodforde and Marshall Rivers, which
transport bedloads of coarse sand to granules, many
reaches are characterised by anabranching channels
occurring within a broader channel-train. The indi-
vidual anabranches are relatively deep and narrow
and are separated by channel-train ridges--narrow,
flow-aligned, vegetated features--or by wider is-
lands. Ridges and islands are either depositional
® .features formed in situ by accretionary processes or
®erosional features formed by excision from once-
.continuous areas of floodplain . Vegetation plays a
key role in the initiation, survival and growth of
depositional forms through its influence on flow and
sediment transport. As such, these distinctive fluvial
landforms contribute to the growing body of knowl-
®edge termed `biogeomorphology' Viles, 1988; Hupp
.et al., 1995 .
Anabranching is also related to the influence of
tributaries, for both the Bundey and Marshall Rivers
alternate from single-thread to anabranching along
their length in response to tributary inputs of water
and sediment. Tributary inflows occur either inde-
pendently from, or in concert with, flows in trunk
channels. Ridges and islands form downstream of
tributary junctions as a result of various hydrologi-
cal, depositional and erosional processes, including
irrigation of enhanced numbers of in-channel trees
and resulting lee-side sediment accretion, linear dis-
section of floodplains, and by formation and mainte-
nance of deferred-junction tributaries. Through their
influence on river patterns, tributary inflows on many
rivers on the Northern Plains have a greater effect
than has previously been described for ephemeral
channels. Furthermore, the importance of sub-bank-
full flows emanating from tributaries as a contribu-
tory factor to the initiation, stabilisation and growth
of some ridges and islands contrasts with the often
overriding dominance attributed to large floods in
shaping ephemeral channel morphology. Ridges and
islands forming downstream of tributary junctions
provide temporary storage sites for sediment and
appear to help maintain the throughput of relatively
coarse sediment in these low-gradient channels,
which are characterised by an abundance of within-
channel vegetation and subject to declining down-
stream discharges. Anabranching channels are a
common feature of arid Australia and thus similar
tributary-related mechanisms may contribute to
anabranching in other low-gradient, arid catchments.
Acknowledgements
Research was supported by a postgraduate schol-
arship from the University of Wollongong to S.
Tooth and with grants from the Quaternary Environ-
® .ments Research Centre University of Wollongong
and the Australian Research Council to G.C. Nan-
son. Fieldwork assistance from Richard Walsh, Steve
Beaman, Amanda Hajnal, He Qing Huang and Alice
Turkington is gratefully acknowledged. The com-
ments of two anonymous referees are appreciated.
References
Allen, J.R.L., 1982. Sedimentary Structures: Their Character and
Physical Basis, Vols. I and II. Elsevier, Amsterdam.
Baker, V.R., Pickup, G., Polach, H.A., 1983. Desert palaeofloods
in central Australia. Nature 301, 502­504.
Brice, J.C., 1984. Planform properties of meandering rivers. In:
® .Elliot, C.M. Ed. , Proceedings of the Conference--Rivers
'83. American Society of Civil Engineers, New Orleans, pp.
1­15.
Brizga, S.O., Finlayson, B.L., 1990. Channel avulsion and river
metamorphosis: the case of the Thompson River, Victoria,
Australia. Earth Surface Processes and Landforms 15, 391­
404.
Brooks, A., Brierley, G.J., 1997. Geomorphic responses of the
lower Bega River to catchment disturbance, 1851­1926. Geo-
morphology 18, 291­304.
Chan, K.Y., Abbott, T.S., 1993. Physico-chemical properties con-
® .trolling soil structure. In: Hazelton, P., Koppi, A.J. Eds. , Soil
Technology--Applied Soil Science. A Course of Lectures,
2nd edn. Australian Society of Soil Science, NSW Branch and
Department of Agricultural Chemistry and Soil Science, Uni-
versity of Sydney, Sydney, pp. 61­76.
Dunaway, D., Swanson, S.R., Wendel, J., Clary, W., 1994. The
effect of herbaceous plant communities and soil textures on
particle erosion of alluvial streambanks. Geomorphology 9,
47­56.
Dzulynski, S., Walton, E.K., 1965. Sedimentary Features of Fly-
sch and Greywackes. Elsevier, Amsterdam.
Emerson, W.W., 1967. A classification of soil aggregates based
on their coherence in water. Australian Journal of Soil Re-
search 5, 47­57.
( )S. Tooth, G.C. NansonrGeomorphology 29 1999 211­233232
Evans, T.G., 1972. 1:250,000 Geological Series--Explanatory
Notes. Napperby, Northern Territory, Sheet SFr53-9 Interna-
tional Index. Australian Government Publishing Service, Can-
berra.
Fisk, H.N., 1952. Mississippi River valley geology in relation to
river regime. American Society of Civil Engineers Transac-
tions 117, 667­689.
Graeme, D., Dunkerley, D.L., 1993. Hydraulic resistance by the
river red gum, Eucalyptus camaldulensis, in ephemeral desert
streams. Australian Geographical Studies 31, 141­154.
Graf, W.L., 1978. Fluvial adjustments to the spread of tamarisk in
the Colorado Plateau region. Geological Society of America
Bulletin 89, 1491­1501.
Graf, W.L., 1979. Fluvial adjustments to the spread of tamarisk in
the Colorado Plateau region: reply. Geological Society of
America Bulletin 90, 1183­1184.
Graf, W.L., 1988. Fluvial Processes in Dryland Rivers. Springer,
Berlin.
Hadley, R.F., 1961. Influence of riparian vegetation on channel
shape, northeastern Arizona. United States Geological Survey
Professional Paper 424C, 30­31.
Harwood, K., Brown, A.G., 1993. Fluvial processes in a forested
anastomosing river: flood partitioning and changing flow pat-
terns. Earth Surface Processes and Landforms 18, 741­748.
® .Hupp, C.R., Osterkamp, W.R., Howard, A.D. Eds. , 1996. Bio-
geomorphology, Terrestrial and Freshwater Systems. Geomor-
phology 13, 347 pp.
Karcz, I., 1968. Fluviatile obstacle marks from the wadis of the
® .Negev southern Israel . Journal of Sedimentary Petrology 38,
1000­1012.
Knighton, A.D., Nanson, G.C., 1993. Anastomosis and the contin-
uum of channel pattern. Earth Surface Processes and Land-
forms 18, 613­625.
Leopold, L.B., Wolman, M.G., 1957. River channel patterns:
braided, meandering and straight. United States Geological
Survey Professional Paper 282B, 39­85.
Mabbutt, J.A., 1962. Part VII. Geomorphology of the Alice
® .Springs area. In: Perry, R.A. Ed. , General Report on Lands
of the Alice Springs Area, Northern Territory, 1956­1957.
CSIRO Land Research Series 6, 163­184.
Mabbutt, J.A., 1967. Denudation chronology in central Australia:
structure, climate and landform inheritance in the Alice Springs
® .area. In: Jennings, J.N., Mabbutt, J.A. Eds. , Landform Stud-
ies from Australia and New Guinea. Australian National Uni-
versity Press, Canberra, pp. 144­181.
Mabbutt, J.A., 1977. Desert Landforms. Australian National Uni-
versity Press, Canberra.
® .Mabbutt, J.A., 1986. Desert lands. In: Jeans, D.N. Ed. , Australia
--A Geography, 1: The Natural Environment. Sydney Univer-
sity Press, Sydney, pp. 180­202.
McCarthy, T.S., Ellery, W.N., Stanistreet, I.G., 1992. Avulsion
mechanisms on the Okavango fan, Botswana: the control of a
fluvial system by vegetation. Sedimentology 39, 779­795.
Mollard, J.D., 1973. Airphoto interpretation of fluvial features. In:
Proceedings of the 9th Canadian Hydrology Symposium. Na-
tional Research Council of Canada, Edmonton, Alberta, pp.
341­380.
Nanson, G.C., 1980. Point bar and floodplain formation of the
meandering Beatton River, northeastern British Colombia,
Canada. Sedimentology 27, 3­29.
Nanson, G.C., 1981. New evidence of scroll bar formation on the
Beatton River. Sedimentology 28, 889­891.
Nanson, G.C., Croke, J.C., 1992. A genetic classification of
® .floodplains. In: Brackenbridge, G.R., Hagedorn, J. Eds. ,
Floodplain Evolution. Geomorphology 4, 459­486.
Nanson, G.C., Huang, H.Q., 1999. Anabranching rivers: divided
efficiency leading to fluvial diversity. In: Miller, A., Gupta, A.
® .Eds. , Varieties of Fluvial Form. Wiley, New York, pp.
477­494.
Nanson, G.C., Knighton, A.D., 1996. Anabranching rivers: their
cause, character and classification. Earth Surface Processes
and Landforms 21, 217­239.
Nanson, G.C., East, T.J., Roberts, R.G., 1993. Quaternary stratig-
raphy, geochronology and evolution of the Magela Creek
catchment in the monsoon tropics of northern Australia. Sedi-
mentary Geology 83, 277­302.
Nezu, I., Nakagawa, H., 1989. Self forming mechanism of longi-
tudinal sand ridges and troughs in fluvial open-channel flows.
In: Proceedings 23rd IAHR Congress, Ottawa, Canada. pp.
B65­B72.
Nezu, I., Nakagawa, H., Tominaga, A., 1985. Secondary currents
in a straight channel flow and the relation to its aspect ratio.
In: Turbulent Shear Flows 4. Springer, Berlin, pp. 246­260.
Ogden, J., 1978. On the dendrochronological potential of Aus-
tralian trees. Australian Journal of Ecology 3, 339­356.
Osterkamp, W.R., Costa, J.E., 1987. Changes accompanying an
extraordinary flood on a sand-bed stream. In: Mayer, L., Nash,
® .D. Eds. , Catastrophic Flooding. Binghampton Symposium in
Geomorphology 18. Allen and Unwin, Boston, pp. 201­224.
Picard, M.D., High, L.R. Jr., 1973. Sedimentary Structures of
Ephemeral Streams. Developments in Sedimentology 17. Else-
vier, Amsterdam.
Pickup, G., 1991. Event frequency and landscape stability on the
floodplain systems of arid central Australia. Quaternary Sci-
ence Reviews 10, 463­473.
Pickup, G., Allan, G., Baker, V.R., 1988. History, palaeochannels
and palaeofloods of the Finke River, central Australia. In:
® .Warner, R.F. Ed. , Fluvial Geomorphology of Australia. Aca-
demic Press, Sydney, pp. 105­127.
Popov, I., 1962. Application of morphological analysis to the
evolution of channel deformation of the River Ob. Soviet
® .Hydrology American Geophysical Union 3, 267­324.
Rust, B.R., 1978. A classification of alluvial channel systems. In:
® .Miall, A.D. Ed. , Fluvial Sedimentology. Canadian Society of
Petroleum Geologists, Calgary, pp. 187­198.
Sarkar, S.K., Basumallick, S., 1968. Morphology, structure, and
evolution of a channel island in the Barakar River, Barakar,
West Bengal. Journal of Sedimentary Petrology 38, 747­754.
Schumann, R.R., 1989. Morphology of Red Creek, Wyoming, an
arid-region anastomosing channel system. Earth Surface Pro-
cesses and Landforms 14, 277­288.
Schumm, S.A., 1981. Evolution and response of the fluvial sys-
tem, sedimentologic implications. Society of Economic Pale-
ontologists and Mineralogists, Special Publication 31, 19­29.
( )S. Tooth, G.C. NansonrGeomorphology 29 1999 211­233 233
Schumm, S.A., 1985. Patterns of alluvial rivers. Annual Review
of Earth and Planetary Sciences 13, 5­27.
Schumm, S.A., Erskine, W.D., Tilleard, J.W., 1996. Morphology,
hydrology, and evolution of the anastomosing Ovens and King
Rivers, Victoria, Australia. Geological Society of America
Bulletin 108, 1212­1224.
Slatyer, R.O., Mabbutt, J.A., 1964. Hydrology of arid and semi-
® .arid regions. In: Chow, V.T. Ed. , Handbook of Applied
Hydrology. McGraw-Hill, New York, pp. 24-1­24-46.
Smith, D.G., 1976. Effect of vegetation on lateral migration of
anastomosed channels of a glacial meltwater river. Geological
Society of America Bulletin 86, 857­860.
Smith, D.G., Smith, N.D., 1980. Sedimentation in anastomosed
river systems: examples from alluvial valleys near Banff,
Alberta. Journal of Sedimentary Petrology 50, 157­164.
Smith, N.D., Cross, T.A., Dufficy, J.P., Clough, S.R., 1989.
Anatomy of an avulsion. Sedimentology 36, 1­23.
Sneh, A., 1983. Desert stream sequences in the Sinai Peninsula.
Journal of Sedimentary Petrology 53, 1271­1279.
Stewart, G.A., Perry, R.A., 1962. Part I. Introduction and sum-
mary description of the Alice Springs area. In: Perry, R.A.
® .Ed. , General Report on Lands of the Alice Springs Area,
Northern Territory, 1956­1957. CSIRO Land Research Series
6, 9­19.
Stidolph, P.A., Bagas, L., Donnellan, N., Walley, A.M., Morris,
D.G., Simons, B., 1988. Elkedra SF53-7: 1:250,000 Geologi-
cal Map Series--Explanatory Notes. Department of Mines and
EnergyrNorthern Territory Geological Survey, Darwin.
Thorne, C.R., 1990. Effects of vegetation on riverbank erosion
® .and stability. In: Thornes, J.B. Ed. , Vegetation and Erosion:
Processes and Environments. Wiley, Chichester, pp. 125­144.
Thornes, J.B., 1977. Channel changes in ephemeral streams:
® .observations, problems and models. In: Gregory, K.G. Ed. ,
River Channel Changes. Wiley, Chichester, pp. 317­335.
Tooth, S., 1999. Floodouts in central Australia. In: Miller, A.,
® .Gupta, A. Eds. , Varieties of Fluvial Form. Wiley, Chi-
chester, pp. 219­247.
Tsujimoto, T., 1989. Longitudinal stripes of alternate lateral sort-
ing due to cellular secondary currents. In: Proceedings 23rd
IAHR Congress, Ottawa, Canada. pp. B17­B24.
® .Viles, H.A. Ed. , 1988. Biogeomorphology. Blackwell, Oxford.
Wende, R., Nanson, G.C., 1998. Anabranching rivers: ridge-form
alluvial channels in tropical northern Australia. Geomorphol-
ogy 22, 205­224.
Williams, G.E., 1970a. The central Australian stream floods of
February­March 1967. Journal of Hydrology 11, 185­200.
Williams, G.E., 1970b. Piedmont sedimentation and late Quater-
nary chronology in the Biskra region of the northern Sahara.
Zeitschrift fur Geomorphologie, Supplementband 10, 40­63.¨
Woodyer, K.D., Taylor, G., Crook, K.A.W., 1979. Depositional
processes along a very low-gradient, suspended-load stream:
the Barwon River, New South Wales. Sedimentary Geology
22, 97­120.