Earth-Science Reviews 53 Ž2001.

149–196
www.elsevier.comrlocaterearscirev

Anastomosing rivers: a review of their classification, origin and

sedimentary products
Bart Makaske )
The Netherlands Centre for Geo-Ecological Research (ICG), Faculty of Geographical Sciences, Utrecht UniÕersity, Heidelberglaan 2,
P.O. Box 80115, 3508 TC Utrecht, Netherlands
Accepted 7 November 2000

Abstract

Anastomosing rivers constitute an important category of multi-channel rivers on alluvial plains. Most often they seem to
form under relatively low-energetic conditions near a Žlocal. base level. It appears to be impossible to define anastomosing
rivers unambiguously on the basis of channel planform only. Therefore, the following definition, which couples floodplain
geomorphology and channel pattern, is proposed in this paper: an anastomosing riÕer is composed of two or more
interconnected channels that enclose floodbasins. This definition explicitly excludes the phenomenon of channel splitting by
convex-up bar-like forms that characterize braided channels.
In present definitions of anastomosing rivers, lateral stability of channels is commonly coupled with their multi-channel
character. Here, it is suggested that these two properties be uncoupled. At the scale of channel belts, the terms ‘straight’,
‘meandering’ and ‘braided’ apply, whereas at a larger scale, a river can be called anastomosing if it meets the definition
given above. This means that, straight, meandering and braided channels may all be part of an anastomosing river system.
Straight channels are defined by a sinuosity index; i.e., the ratio of the distance along the channel and the distance along the
channel-belt axis is less than 1.3. They are the type of channel that most commonly occurs in combination with anastomosis.
The occurrence of straight channels is favoured by low stream power, basically a product of discharge and gradient, and
erosion-resistant banks.
Anastomosing rivers are usually formed by avulsions, i.e., flow diversions that cause the formation of new channels on
the floodplain. As a product of avulsion, anastomosing rivers essentially form in two ways: Ž1. by formation of bypasses,
while bypassed older channel-belt segments remain active for some period; and Ž2. by splitting of the diverted avulsive flow,
leading to contemporaneous scour of multiple channels on the floodplain. Both genetic types of anastomosis may coexist in
one river system, but whereas the first may be a long-lived floodplain-wide phenomenon, the latter only represents a stage in
the avulsion process on a restricted part of the floodplain.
Long-lived anastomosis is caused by frequent avulsions andror slow abandonment of old channels. Avulsions are
primarily driven by aggradation of the channel belt andror loss of channel capacity by in-channel deposition. Both processes
are favoured by a low floodplain gradient. Also of influence are a number of avulsion triggers such as extreme floods, log
and ice jams, and in-channel aeolian dunes. Although some of these triggers are associated with a specific climate, the
occurrence of anastomosis is not. A rapid rise of base level is conductive to anastomosis, but is not a necessary condition.

)
Department of Landscape and Spatial Planning, Alterra, Wageningen University and Research Centre, P.O. Box 47, 6700 AA
Wageningen, The Netherlands.

0012-8252r01r$ - see front matter q 2001 Elsevier Science B.V. All rights reserved.
PII: S 0 0 1 2 - 8 2 5 2 Ž 0 0 . 0 0 0 3 8 - 6
150 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196

Anastomosing rivers can be considered an example of equifinality, since anastomosis may result from different combinations
of processes or causes.
Anastomosing river deposits have an alluvial architecture characterized by a large proportion of overbank deposits, which
encase laterally connected channel sand bodies. Laterally extensive, thick lenses of lithologically heterogeneous, fine-grained
avulsion deposits can be an important element of the overbank deposits of anastomosing rivers. These deposits may also
fully surround anastomosing channel sandstones. Anastomosing channel sand bodies frequently have ribbon-like geometries
and may possess poorly developed upward-fining trends, as well as abrupt flat tops. The overbank deposits commonly
comprise abundant crevasse splay deposits and thick natural levee deposits. Lacustrine deposits and coal are common in
association with anastomosing river deposits. None of these characteristics is unique to anastomosing river deposits, and in
most cases, anastomosis Žcoexistence of channels. cannot be demonstrated in the stratigraphic record. q 2001 Elsevier
Science B.V. All rights reserved.

Keywords: anastomosing rivers; straight rivers; channel pattern; floodplains; fluvial sedimentation; avulsion

1. Introduction wide Že.g. Smith, 1973, 1983, 1986; Smith and Smith,
1980; Rust, 1981; Nanson et al., 1986; Schumann,
Anastomosing rivers represent a major group of 1989; Schumm et al., 1996; Makaske, 1998; and
rivers that are currently of interest in fluvial geomor- many others.. It is realized now that ancient anasto-
phology and sedimentology. For a long time, they mosing river deposits may have a high preservation
were frequently confused with braided rivers, which potential in the stratigraphic record, and that they can
roughly have a comparable planform. Nowadays, the have much economic importance ŽSmith and Put-
term ‘anastomosing’ is reserved for a type of river nam, 1980; Putnam, 1993; Smith, 1986.. Modern
with multiple, interconnected, coexisting channel anastomosing river floodplains provide an extensive
belts on alluvial plains ŽFig. 1., which most often potential of arable and pasture land for rapidly grow-
seems to form under relatively low-energetic condi- ing populations in developing countries Že.g.
tions near a local base level. Makaske, 1998, p. 245.. Putting this potential into
In past three decades, a substantial number of use requires control measures in the natural river
anastomosing river systems has been reported world- system to prevent flooding and to restrain channel

Fig. 1. The anastomosing Tonle´ Sap River in central Cambodia.

 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196 151

dynamics. In addition, some anastomosing river ‘anastomosing’; Ž2. what drives river anastomosis;
floodplains support rare and diverse ecosystems that and Ž3. what are sedimentary products preserved of
need conservation. Thus, continued scientific atten- an anastomosing river system on a geological time
tion to the origin and evolution of anastomosing scale?
river systems is needed. In the first part of this paper, I propose a geomor-
In spite of the accumulating number of studies on phological definition and classification of anastomos-
anastomosing rivers, major controversies still exist ing rivers. In the second part, processes and external
about their definition, the causes for their existence, controls of modern and subrecent anastomosing rivers
and the significance of anastomosing river facies are reviewed. This leads to a conceptual model for
models. Multiple incongruous criteria have been used the origin of anastomosing rivers. In the third part,
by different authors for definition of the anastomos- the sedimentary products of modern and ancient
ing river type. What also remains unclear is the anastomosing rivers are discussed and criteria useful
relationship between anastomosing rivers and dis- for the identification of ancient anastomosing river
tributary river deltas, and the relationship between systems are proposed.
anastomosing rivers and straight rivers Žsensu
Schumm and Khan, 1972.. Considering the causes of
river anastomosis, controversies concern the roles of 2. Definition and classification
climate, base level, and floodplain sedimentation
rates. Modern anastomosing rivers have been re-
ported from various climatic zones; however, certain 2.1. Introduction
controls of anastomosis Žespecially vegetation. seem
to differ with climate. Are morphologically similar Schumm Ž1968, p. 1580. may have been the first
anastomosing rivers in different climates a product to point out that the term ‘anastomosing’ should not
of convergence wor equifinality ŽSchumm, 1988., be used as a synonym for braiding: AThe terms
with different processes or causes producing similar braiding and anastomosing have been used synony-
results? Apart from climate, the influences of base mously for braided river channels in this country, but
level and floodplain sedimentation rates are unclear. elsewhere, particularly in Australia, anastomosing is
Rapid base-level rise and high floodplain sedimenta- a common term applied to multiple-channel systems
tion rates have been mentioned as prime causes for on alluvial plains Ž . . . .. The channels transport flood
anastomosis in modern examples from western waters and, because of the small sediment load
Canada and Colombia ŽSmith, 1983, 1986.. In con- moved through them, aggradation, if it is occurring,
trast, in central Australia anastomosing rivers exist is a slow process. As a result, these low-gradient
under relatively low floodplain sedimentation rates, suspended-load channels are quite stable Ž . . . ..B Al-
resulting from a relatively stable base level ŽRust, though Schumm’s description seems to be a clear
1981; Nanson et al., 1986; Gibling et al., 1998.. starting point, nowadays there is a bewildering array
Also, the universal applicability of anastomosing of definitions of the anastomosing river type.
river facies models is not generally agreed upon. Preferably, a definition of a river type should
Reviewing fluvial facies models, Hickin Ž1993, p. meet the following criteria: Ž1. It should be workable
210. even suggested that: A . . . the ‘anastomosing in the sense that an air-photo interpretation or a
river facies model’ is an entirely premature concept, quick field survey should be sufficient to enable its
the full development of which should await much application. Therefore, it should not include genetic,
more extensive field studies of modern riversB. hydraulic or sedimentological characteristics, but in-
In this paper, the state of our knowledge of stead it should be based on a limited number of
anastomosing rivers in alluvial settings is evaluated, visual characteristics concerning channel pattern and
focusing on their geomorphology as well as on their floodplain geomorphology. Ž2. It should be useful in
sedimentology. Both modern and interpreted ancient the sense that it distinguishes a group of rivers with a
examples are reviewed. The three main questions similar genesis and similar sedimentological and hy-
addressed in this paper are: Ž1. when to call a river draulic characteristics.
152 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196

2.2. Channel pattern mosing channels. The angle of bifurcation reflects a

difference between the mechanisms leading to braid-
Following work of the past two or three decades, ing Žformation of mid-channel bars. vs. anastomosis
it is now generally accepted that an anastomosing Žavulsion, see Section 3.2.. However, the differences
river is a multi-channel river which is fundamentally in bifurcation angles between braiding and anasto-
different in form and process from a braided river. mosing are too small to serve as a basis for a
Can this river type be defined on the basis of channel clear-cut definition wsee for example Yonechi and
pattern? Maung Ž1986, their fig. 5.x.
A channel pattern is considered as a two-dimen- Knighton and Nanson Ž1993. stated that islands
sional, planform configuration only, regardless of between anastomosing channels are large relative to
any other floodplain characteristics. In this respect, the size of the channels. Some workers included a
two properties are most relevant: Ža. channel sinuos- measure of the length or width of alluvial islands in
ity Ždistance along the channel divided by straight the definition of the term anastomosing. In this
line distance. and Žb. channel multiplicity. context, the term anabranching has been fre-
Rust Ž1978. defined anastomosing rivers as: quently used. Bridge Ž1993, p. 21. stated that
high-sinuosity Ž) 1.5., multi-channel rivers. He ap- A . . . anastomosing channel segments Ž . . . . are longer
plied sinuosity and the braiding parameter Žnumber than a curved channel segment around a single braid
of braids per mean meander wavelength. to distin- or point bar and their width-scale flow patterns
guish between straight, braided, meandering and behave substantially independent of the adjacent seg-
anastomosing ŽTable 1.. He admitted that his bound- ments.B He considered anabranching as a synonym
ary values were arbitrary, but stated that a classifica- of anastomosing. Brice et al. Ž1978. Žin Schumm,
tion should be quantitative. Rust Ž1981, his table 2. 1985. used the term anabranching for channels en-
listed mean sinuosities of anastomosing rivers rang- closing islands having a width larger than three times
ing from 1.51 to 1.75. Smith Ž1983., however, de- channel width at average discharge. Schumm Ž1985.
scribed anastomosing rivers in western Canada and regarded anabranching rivers as being essentially
found low sinuosities Ž1.16 and 1.4.. He concluded braided rivers with large exposed bars in relation to
that anastomosing rivers have variable sinuosities channel width, as opposed to anastomosing rivers
ŽSmith and Putnam, 1980.. So it seems that sinuosity being true multi-channel systems. Nanson and
of the individual branches is not suitable for distin- Knighton Ž1996., however, considered anastomosing
guishing anastomosing rivers Žsee also Table 2., a rivers as a subgroup of anabranching rivers. By
fact also recognized by Knighton and Nanson Ž1993.. them, anabranching rivers were not defined by the
Yonechi and Maung Ž1986. pointed out that size of the alluvial islands or any other channel
braided channels bifurcate at acute angles to the pattern characteristic, but by the stability of river-
main flow direction, while anastomosing channels banks Žcaused by vegetation or otherwise. and con-
bifurcate at a wider range of angles and even at tinued existence of the islands up to nearly bankfull
obtuse angles. Harwood and Brown Ž1993. also noted stage. Anastomosing rivers were earlier classified as
the relatively large angles of bifurcation of anasto- low-energy, anabranching rivers with cohesive banks
ŽSmith and Smith, 1980; Rust, 1981..
In summary, present terminology is confusing and
Table 1 there is no consensus on how to incorporate the size
Classification of channel patterns according to Rust Ž1978. of alluvial islands in a useful definition of anastomo-
Single-channel Multi-channel sis, although there is a tendency to consider anasto-
Žbraiding Žbraiding
mosing as a hierarchically higher order of channel
parameter -1. parameter -1.
pattern not comparable to straight, meandering and
Low-sinuosity straight braided braided Že.g. Schumm, 1968, 1985; Bridge, 1993..
Ž P -1.5.
High-sinuosity meandering anastomosing
This idea, however, is hard to translate into a defini-
Ž P -1.5. tion based on channel pattern only. Boundary values
for the size of islands tend to be quite arbitrary and
 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196 153

Table 2
Quantitative data on anastomosing rivers
River, climatic setting, Qma Q bf Qmax Sch P wrd Sed. rate
and data source Žm3 rs. Žm3rs. Žm3 rs. Žcmrkm. Ž–. Ž–. Žmmryear.
Temperate humid
Mistaya1 34 390 15 0.6
Alexandra2 66 60 1.51 13 1.8
North Saskatchewan3 164 100 1.62 16 1.8
Upper Columbia4 108 275 770 9.6 1.16 16 1.7
Lower Saskatchewan5 648 1400 3000 12.2 1.4 a 1.5 b
Ovens6 225 89 1.49 17

Tropical humid
Magdalena7 7400 8800 10.0 28 3.8
Japura8 11,130 1.1
˜ 9
Solimoes 36,815 1.5 2.3
Magela Creek 10 40 1580 50 1.1 10 1.5

Arid and semi-arid

Red Creek 11 11 1.92 - 10
Upper Darling 12 681 200 5 2.3 8 0.02
Cooper Creek 13 40–100 5800–25,000 17.5 1.69 10 0.04
Fitzroy 14 30,000 25 1.1 c

Middle Niger 15 1330 9700 2.0 1.2 0.14

Okavango16 335 19 1.5 F 20

Subarctic
Attawapiskat 17 508 3115 52 - 1.5 70 d

Qma s mean annual discharge; Q bf s bankfull discharge; Qmax s maximum recorded discharge; Sch s channel slope; P s sinuosity;
wrd s widthrdepth ratio of channels; Sed. rate s average floodplain sedimentation rate.
a
P varies between ; 1.1 for recent avulsion channels and 1.65 for old channels.
b
Short-term sedimentation rates within the recent avulsion belt are up to 33 mmryear.
c
wrd strongly variable: 6–233.
d
Incision.
1
Smith Ž1986., Smith and Smith Ž1980..
2
Smith Ž1986., Smith and Smith Ž1980., Rust Ž1981..
3
Smith Ž1973, 1986., Smith and Smith Ž1980., Rust Ž1981..
4
Smith Ž1983, 1986., Water Survey of Canada Ž1991a., Makaske Ž1998..
5
Smith Ž1983, 1986., Smith et al. Ž1989., Water Survey of Canada Ž1991b..
6
Schumm et al. Ž1996..
7
Smith Ž1986..
8
Baker Ž1978..
9
Baker Ž1978., Mertes Ž1994..
10
Nanson et al. Ž1993., Knighton and Nanson Ž1993..
11
Schumann Ž1989..
12
Riley and Taylor Ž1978..
13
Rust Ž1981., Knighton and Nanson Ž1993..
14
Taylor Ž1999..
15
ORSTOM Ž1970., Gallais Ž1967., Direction Nationale de l’Hydraulique et de l’Energie, Bamako, Mali Žunpubl. data..
16
McCarthy et al. Ž1991, 1992., Smith et al. Ž1997..
17
King and Martini Ž1984..

not discriminative for a genetically, sedimentologi- 1993, his fig. 4.. Following Smith and Smith Ž1980,
cally and hydraulically different type of river, since their fig. 6. and Nanson and Knighton Ž1996., it can
large islands can be coalesced braid bars Že.g. Bridge, be concluded that it is the nature of the islands,
154 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196

rather than their size, which discriminates anasto- of his Kwilu-type rivers in the Zaire Basin likewise
mosing rivers from braided rivers. as: Amini alluvial plains: surrounded by levee de-
positsB. In various papers, extensive wetlands Žmainly
comprising the islands. and narrow prominent natu-
2.3. Floodplain geomorphology ral levees ŽFig. 2. were mentioned as characteristics
of anastomosing river floodplains in humid climates
The long recognized genetic association between ŽSmith and Smith, 1980; Smith and Putnam, 1980;
rivers and their floodplains led Nanson and Croke Smith, 1983, 1986.. Anastomosing systems in arid
Ž1992. to a classification of floodplains rather than and semi-arid climates ŽRust, 1981; Nanson et al.,
rivers. It seems reasonable to assume that anastomos- 1986; Schumann, 1989. lack these wetlands due to
ing rivers, if genetically, hydraulically and sedimen- aridity, but the islands to some extent do have the
tologically distinct, are characterized by a different saucer-like morphology that distinguishes them from
floodplain geomorphology too. Some authors, in dis- braid bars, although levees often are less pro-
cussing anastomosing river floodplain geomorphol- nounced. In some anastomosing river systems, levees
ogy, have also paid attention to the stabilizing effect are absent or very subtle phenomena and islands are
of vegetation. just near-level parts of the floodplain enclosed by
Smith Ž1976. suggested the stabilizing effect of multiple channel belts ŽTaylor, 1999..
vegetation is a major factor in the development of Although the rarity or absence of meander scrolls
anastomosing river morphology. Rust Ž1981., how- and oxbow lakes on anastomosing river floodplains
ever, described an anastomosing river in an arid was considered to be a fundamental characteristic by
environment on a scarcely vegetated floodplain and some workers ŽSmith, 1983, his Table 2; Nanson and
concluded that the role of vegetation is less impor- Croke, 1992., this is not generally agreed upon.
tant on arid floodplains. It is obvious that vegetation, Riley Ž1975, p. 3., outlining the differences between
being strongly climatically dependent, cannot serve distributary and braided channels in eastern Aus-
as a universal characteristic of anastomosing rivers. tralia, noted: ACutoffs, point bars and meander scrolls
The morphology of alluvial islands in an anasto- can all be found in association with the individual
mosing reach of the Solimoes ˜ River ŽAmazon Basin. distributariesB. Later, these distributary channels
was described by Sternberg Ž1959, p. 400. and Baker were called anastomosing channels by Taylor and
Ž1978, p. 225. as ‘saucer-like’ due to the bounding Woodyer Ž1978.. They regarded the topography of
natural levees. Savat Ž1975, p. 168. described islands the floodplain referred to by Riley, however, as relic.

Fig. 2. Topographic profile across the floodplain of the anastomosing Alexandra River, Alberta, Canada Žafter Smith and Smith, 1980..
 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196 155

Brizga and Finlayson Ž1990. described a recent and hence in some cases ‘floodbasins’ may rather be
Ž1952. channel avulsion of the Thomson River in near-level overbank areas.
Australia, leading to coexistence of two meandering
channels which both show point bars. The new 2.4. Proposed definition and classification
channel has a narrow meander belt only. It is unclear
whether anastomosis is a temporary situation in this In view of the considerations in the preceding
case. Elsewhere in Australia, Bowler et al. Ž1978. sections, I propose the following definition of anas-
documented a wide meander belt with scroll patterns tomosing rivers based on channel pattern and flood-
for what they called ‘anabranches’ of the Darling plain geomorphology: an anastomosing riÕer is com-
River, which in turn was typified by a narrow mean- posed of two or more interconnected channels that
der belt. Nowadays, the anabranches transport flood- enclose floodbasins. Below, I will elucidate the three
waters only. Schumm et al. Ž1996. described anasto- basic points of the related classification of river
mosing channels of the Ovens and King Rivers in types shown in Fig. 3.
south-eastern Australia as becoming more sinuous Firstly, anastomosing rivers are classified as rivers
with time. In the same region, Erskine Žpersonal having multiple coexistent channel belts, while
communication in Nanson and Knighton, 1996. also braided rivers are regarded as rivers with a single
identified laterally migrating anastomosing channels channel belt, but multiple thalwegs. Although the
of the Murray River. Outside Australia, Baker Ž1978. term ‘channel belt’ originally referred to the zone in
mentioned prominent scroll topography on anasto- which a meandering channel has been active wmore
mosing river floodplains of the Solimoes˜ River. commonly termed ‘meander belt’ Že.g. Fisk, 1947.x,
It can be concluded that, although meander belts here its meaning is widened to involve the zone of
with ridge-and-swale topography generally are not activity of a straight, meandering or braided channel,
characteristic features of anastomosing river flood- including bars, abandoned channel segments,
plains, they may be present. Consequently, it seems crevasse splays and levees, and which is bounded
that meandering and anastomosing are not mutually laterally by floodbasins or the floodplain margin.
exclusive characteristics of a river, which harmo- Generally, islands in an anastomosing river are larger
nizes with Schumm’s Ž1968, 1985. idea to classify than islands or bars in a braided river, which geneti-
anastomosis as a higher order pattern consisting of cally are in-channel features. Geomorphologically,
multiple channels which can either be straight, mean- saucer-shaped islands Žfloodbasins. characterize
dering or braided. Most authors agree that flood- anastomosing rivers, while convex-up islands Žbars.
basins constitute a major element of the geomorphol- characterize braided rivers. Meandering and straight
ogy of anastomosing river floodplains, including the rivers are classified as rivers with both a single
islands between the anastomosing channels. In con- channel belt and a single thalweg, since the flow in
trast, islands within braided channels typically have a their channel is not split by in-channel bars. To
convex-up morphology. distinguish between rivers with single and multiple
In this paper, the term ‘floodbasin’ stands for a thalwegs, the braid-channel ratio Ž B . of Friend and
flat, shallow, poorly drained floodplain depression Sinha Ž1993. is applied, which is defined as the sum
bounded by natural levees of active or abandoned of the mid-channel lengths of all the segments of
channels or other uplands ŽFig. 2.. Floodbasins may primary channels in a reach divided by the mid-
be very extensive and in wet climates they are channel length of the widest channel through the
typically swampy for most of the year. Therefore, reach. Here, a channel is taken to be braided if its
the term ‘backswamp’ was used as a synonym, braid-channel ratio is ) 1.5, although this value is
particularly in the Lower Mississippi Valley Že.g. quite arbitrary and scattered braids already occur in
Saucier, 1994, pp. 126–127.. To stress the geomor- channels with braid-channel ratios between 1.0 and
phology of these areas, rather than the climatically 1.5. The braid-channel ratio is preferable to the
dependent hydrological situation, the term floodbasin braiding parameter Ž B.P.. of Rust Ž1978., since the
is preferred here. It is acknowledged, however, that latter, being loosely defined as the number of braids
not all fluvial systems do form pronounced levees, per mean meander wave-length, requires identifica-
156 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196

Fig. 3. Proposed classification of alluvial river types based on channel pattern and floodplain geomorphology. Anastomosing rivers Žbelow.
are classified as a composite form of which the individual channel belts may have braided, meandering or straight channels Žabove..

tion of meanders for rivers that may not show any similar, i.e. absence of lateral erosion and deposition.
tendency towards meandering. Meandering in this sense means that a reach of the
Secondly, straight and meandering rivers are de- channel is substantially longer than the correspond-
fined with respect to the channel belt. Straight in this ing reach of the channel belt. To describe these
sense does not necessarily mean literally straight, but phenomena, Brice Ž1964. introduced the sinuosity
it means that the course of the channel and the index Ž Pind ., defined as the ratio of the channel
course of the axis of the channel belt are very length to the length of the meander-belt axis. A
 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196 157

sinuosity index of 1.3 is proposed as a boundary reflects fundamental differences in genesis between
value that separates straight from meandering chan- these two categories. Straight, meandering and
nels. For a given channel reach, the sinuosity index braided rivers all basically result from in-channel
will generally be lower than the sinuosity Ž P s processes such as lateral erosion and accretion or
distance along the channel divided by straight line mid-channel bar formation. Anastomosing rivers re-
distance., for which 1.5 is usually accepted as a sult from extra-channel processes, namely floodplain
boundary value between meandering and straight formation by overbank deposition and avulsion, i.e.,
ŽLeopold and Wolman, 1957; Rust, 1978.. the partial diversion of flow from an existing channel
Thirdly, anastomosing rivers are classified as a onto the floodplain Žsee Section 3.2.. The identifica-
composite form of which the individual channel belts tion of floodbasins, which is inherent to the classifi-
may have braided, meandering or straight channels. cation is genetically significant since these areas
This is in accordance with the classification of provide avulsion routes, and their state of preserva-
Schumm Ž1968, 1985.. An anastomosing river com- tion provides information on the lateral activity of
posed of multiple braided channel belts seems to be the channels. In addition, their identification is sedi-
an exceptional form, with rare examples mentioned mentologically significant, as they indicate major
by Smith and Smith Ž1980, their Fig. 6. and Taylor lateral differentiation in sedimentary facies from
Ž1999.. Meandering channels are not common fea- channel to floodbasin.
tures of anastomosing rivers, but they may occur. As will be illustrated in the next sections, anasto-
Most anastomosing rivers consist of straight chan- mosing rivers can occur in various geological set-
nels Že.g. Smith and Smith, 1980; Rust, 1981; Smith, tings, including deltaic and coastal plains. Deltaic
1983.. However, there is no fundamental reason for fluvial systems that are principally distributive may
an a priori exclusion of meandering and braided also be partly anastomosing, when radial distribu-
channels in anastomosing rivers by definition. For taries are interconnected or have bypasses that rejoin
simplicity, in this classification straight, meandering downstream. It is hypothesized that such deltaic
and braided are presented as mutually exclusive fluvial systems are not fundamentally different from
properties of a channel belt. In fact, this is only inland anastomosing rivers in forms, processes and
correct for straight and meandering, being defined by sediments Žsee also Makaske, 1998.. Another cate-
sinuosity index only. In theory, a braided channel gory of radial multi-channel systems that may also
belt may at the same time also be straight or mean- be partly anastomosing includes the low-sinuosity
dering. However, in this classification a channel belt meandering alluvial fans described by Stanistreet and
with multiple thalwegs Ž B ) 1.5. is simply classified McCarthy Ž1993.. However, most alluvial fans are
as braided irrespective of its sinuosity index. characterized by non-anastomosing, strictly radial
An important advantage of this classification is patterns.
that it can be applied at the local scale of individual The focus of this paper is on anastomosing rivers
channel reaches, without regard to the rest of the and therefore only three basic Žsingle channel-belt.
system. Strictly speaking, when considering isolated river types are discerned. This is a simplification, as
reaches of a channel, the term anastomosing has no nowadays more river types are recognized in a con-
meaning. This is important especially when interpret- tinuum from straight Žas defined above. to braided.
ing fluvial sediments, as exposure is often limited to One of these is the wandering river, which is men-
individual channel deposits and rarely will the recon- tioned here because the term anastomosing was fre-
struction of a multi-channel palaeoriver be possible quently applied to it Že.g. Mollard, 1973; Church,
Žsee also Section 4.5.. 1983, p. 180.. Wandering rivers, as described by
The classification is also meaningful with regard Church Ž1983. and Carson Ž1984b., occupy an inter-
to the genesis and sedimentology of rivers. The mediate position between meandering and braided
hierarchical difference in the classification between Že.g. Nanson and Croke, 1992; Hickin, 1993.. Be-
braided, meandering, and straight rivers with single cause of the presence of relatively stable vegetated
channel belts, on the one hand, and anastomosing islands in these rivers, they resemble anastomosing
rivers with multiple channel belts on the other hand, rivers in pattern. However, in well-documented wan-
158 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196

dering rivers multiple channels and thalwegs exist tings include montane, foreland and intracratonic
within one channel belt occupying a narrow flood- basins as well as coastal environments. The spatial
plain confined by valley walls. If floodplains were scales of anastomosing rivers also vary widely ŽTa-
wider, multiple wandering river channel belts could ble 2.. However, it cannot be concluded that the
in principle form an anastomosing complex. formation of anastomosing rivers is completely inde-
It must be stressed that the proposed classification pendent of climate and geological setting.
applies to alluvial rivers only. Non-alluvial rivers The geomorphological processes in anastomosing
may to some extent resemble the classified alluvial rivers are of two kinds: Ž1. the processes creating the
rivers; for example, the steep-gradient Ž Sch f 12 pattern of multiple channel belts, i.e. the processes
mrkm. ‘anastomosing’ streams described by Miller associated with avulsion, and Ž2. the processes deter-
Ž1991.. However, these degrading streams lack lat- mining the morphology of the individual channel
eral floodbasins and, moreover, they are fundamen- belts of the system. In many different ways, climate
tally different in process since avulsion sites are and geology are important external controls of these
determined by resistant bedrock strata. processes.
Nanson and Knighton Ž1996. recently defined
anabranching rivers as: Asystems of multiple chan- 3.2. AÕulsion
nels characterized by vegetated or otherwise stable
alluvial islands that divide flows at discharges up to Allen Ž1965, p. 119. defined avulsion as: Athe
nearly bankfullB. They distinguished six types of sudden abandonment of a part or the whole of a
anabranching rivers on the basis of stream power, meander belt by a stream for some new course at a
sediment texture and river morphology. These types lower level on the floodplainB. Sometimes this is
included several types of wandering rivers as well as referred to as first-order avulsion, with second and
anastomosing rivers. The anastomosing rivers as de- third-order avulsion relating to reoccupation of old
fined in this paper, comprise their types 1, 2 and 3 channels and initiation of new braids within a braided
anabranching rivers. These are: cohesive sediment channel, respectively Že.g. Nanson and Knighton,
anabranching rivers Žtype 1., sand-dominated, 1996, p. 219.. In this paper I will use the term
island-forming anabranching rivers Žtype 2., and avulsion in the sense of Allen Ž1965., i.e., relating to
mixed-load laterally active anabranching rivers Žtype the process of formation of a new channel belt.
3.. Despite differences in sediment and channel mor- However, Allen unjustly restricted avulsion to mean-
phology, all three of these river types have flood- dering rivers only. It is also recognized now that
basins on their islands. ‘sudden abandonment’ of the existing channel belt
A limitation of the present classification is the not necessarily follows the formation of a new chan-
stage-dependency of the definition of braided and nel belt. Therefore, avulsion is better defined as: the
anastomosing rivers. At high stages braided rivers diversion of flow from an existing channel onto the
with submerged bars may seem single-thread chan- floodplain, eventually resulting in a new channel
nels, while at low stages anastomosing rivers may belt. As a product of avulsion, anastomosing rivers
carry water in a single main channel only, thereby essentially form in two ways: Ž1. by formation of
appearing as single-channel rivers. For proper classi- bypasses, while bypassed older channel-belt seg-
fication, rivers should be observed at various stages. ments remain active for some period; and Ž2. by
splitting of the diverted avulsive flow, leading to
contemporaneous scour of multiple channels on the
3. The origin of anastomosing rivers floodplain. The latter type of anastomosis may tem-
porarily exist as a by-product of an avulsion within a
3.1. Introduction larger-scale anastomosing system that was formed by
successive avulsive formation of new bypasses. A
Anastomosing rivers occur in a variety of envi- hierarchy may thus exist.
ronments. Climatic conditions vary from subarctic or Avulsion can take place in a variety of alluvial
temperate to tropical humid or arid. Geological set- environments. It has been reported from braided
 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196 159

rivers on alluvial fans ŽGole and Chitale, 1966; the main channelB. Thus, the avulsion process is
Knight, 1975; Wells and Dorr, 1987; Gohain and caused by the formation of alluvial ridges, as the
Parkash, 1990., braided river floodplains ŽColeman, deposition rate close to the channel is generally
1969; Nordseth, 1973; Carson, 1984b; Bristow, 1999; much higher than it is in the adjacent floodbasin.
Ethridge et al., 1999., meandering river floodplains Mackey and Bridge Ž1995. expressed this relation-
ŽFisk, 1944; Russell, 1954; Mike, 1975; Brizga and ship mathematically as follows:
Finlayson, 1990; Neller et al., 1992; Mack and
r z s aeyb Ž z c r z m .
Leeder, 1998; Smith et al., 1998., and deltaic plains
ŽFisk, 1944; Kruit, 1955; Elliot, 1974; Berendsen, in which: r z s deposition rate Žmryear., at distance
¨
1982; Tornqvist, 1994; Van Gelder et al., 1994; z Žm. from the edge of the channel belt; a s
Makaske, 1998.. maximum net deposition rate Žmryear. at the edge
Subsequent to avulsions, old channels may be of the channel belt; z crz m s dimensionless distance
abandoned at variable rates. Extremely rapid was the from the channel belt Ž – .; b s coefficient that de-
shift to a new course of the Yellow River in 1855, scribes the rapidity at which the rate of deposition
¨
which came about in 1 day ŽQian, 1990.. Tornqvist decreases with distance from the channel belt.
Ž1993b, pp. 155–157. described examples of rela- Mackey and Bridge Ž1995. argued that realistic
tively rapid Žperiod of coexistence of old and new values of b range from 0.35 to 1.4. Considerably
channel - 200 years. as well as more gradual Žperiod higher values of b ranging between 3 and 7 were
of coexistence 500–1000 years. avulsion in the ¨
calculated by Tornqvist et al. Ž1996. based on bore-
Rhine–Meuse delta in the Netherlands. It seems that hole data from the Rhine–Meuse and Mississippi
especially in anastomosing rivers, the abandonment deltaic plains covering long Ž1500 years. time spans.
of old channels may be a very slow process and in ¨
Similar values of b were obtained by Tornqvist et al.
some cases new avulsion channels obviously remain Ž1996. from single-flood data from the Rhine–Meuse
as secondary channels and never take over the entire delta. These high values of b suggest a relatively
discharge of the old channels, Eventually, these new rapid formation of natural levees. Higher compaction
secondary channels may be abandoned earlier than rates of silty, clayey and peaty floodbasin deposits
the older trunk channels. Formation and subsequent with respect to gravelly and sandy channel-belt de-
abandonment of such a secondary channel may be posits, further enhance the formation of alluvial
referred to as a ‘failed’ avulsion ŽMakaske, 1998, pp. ridges. Based on theoretical considerations, Bridge
220–224; Guccione et al., 1999.. In principle, fre- and Leeder Ž1979. suggested that avulsion frequency
quent avulsions andror slow abandonment of old increases with aggradation rate. This was confirmed
channels lead to continuing coexistence of younger by field data on the Holocene development of dis-
deepening channels and old vertically infilling chan- tributaries in the Rhine–Meuse delta, showing that
nels, composing an anastomosing system. the avulsion frequency was highest in the period of
The role of avulsion in the formation of anasto- high sedimentation rate due to rapid sea-level rise.
mosing systems is recognized by many investigators The number of avulsions decreased when the rate of
ŽRiley and Taylor, 1978; Woodyer et al., 1979; sea-level rise decreased and the aggradation rates
McIntosh, 1983; Smith, 1983; Jacobberger, 1987; ¨
slowed down ŽTornqvist, 1994.. An example of ex-
Smith et al., 1989, 1997; Schumann, 1989; Mc- ceptionally high avulsion frequency Žavulsion every
Carthy et al., 1992.. Here, the following questions twelfth year on average. associated with very rapid
are addressed: Ža. what kind of floodplain facilitates vertical aggradation was described for the modern
avulsions, Žb. what initiates the avulsion process, and Yellow River delta ŽVan Gelder et al., 1994.. Exper-
Žc. what is the evolution of the avulsion channel imental evidence for a rise of avulsion frequency
after initiation of the process? with increasing sedimentation rate was given by
Bridge and Leeder Ž1979. stated that: AAvulsion Bryant et al. Ž1995.. In case of alluvial fans and
is initiated if floodwaters travelling from an alluvial deltas, local high sedimentation rates near the fixed
ridge to the floodbasin, through crevasses or low entry point obviously favour the avulsion process
parts of the levees, have a gradient advantage over and cause an avulsion node, i.e., a small area where
160 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196

avulsion occurs much more frequently than else- Leeder, 1998.. Therefore, Knighton and Nanson
where. The resulting semi-circular fan or delta with a Ž1993. argued that a flow regime characterized by
convex-up cross profile is highly conductive to ran- concentrated floods of relatively high magnitude is
dom avulsions, i.e., avulsions not occurring at the conductive to anastomosis. Another possible trigger
avulsion node but scattered over the floodplain. This is a sudden tectonic event resulting in bank collapse
is well documented for the Kosi River alluvial fan in or breaching ŽAlexander and Leeder, 1987.. More
northern India ŽGole and Chitale, 1966; Wells and common triggers are obstructions, blocking dis-
Dorr, 1987; Gohain and Parkash, 1990; Mackey and charge in the main channel and forcing the water to
Bridge, 1995, pp. 20–23.. seek a new course, starting as a crevasse through the
Another boundary condition that controls the natural levee ŽSmith, 1983.. Channel obstructions
avulsion process is the overall gradient of the flood- mentioned in the literature include: beaver dams
plain. At a given aggradation rate, favourable condi- ŽRutten, 1967., log jams ŽSmith, 1983; Harwood and
tions for avulsion Ži.e. gradient advantage across the Brown, 1993., and ice jams ŽKing and Martini, 1984,
natural levee compared to the main channel. are Ethridge et al., 1999. in cold to temperate climates,
attained more easily on a low-gradient floodplain and dunes ŽMcIntosh, 1983; Jacobberger, 1988b;
than on a high-gradient floodplain because of the Makaske, 1998, pp. 163–164. in arid climates. Ex-
low channel gradients involved. The overall flood- amples of weak spots in the natural levee where
plain gradient can be strongly influenced by tecton- crevasses preferentially form, are beaver ŽSmith,
ics. Tilting of the floodplain can give rise to frequent 1983. or hippopotamus trails ŽMcCarthy et al., 1992..
avulsion in a preferential direction. Alexander and Not all crevasses develop into mature avulsion chan-
Leeder Ž1987. called this ‘topographically triggered nels. In fact, the majority will become plugged up
avulsion’. This effect was also quantified in simula- again ŽSmith, 1983.. Slingerland and Smith Ž1998.
tion models for alluvial stratigraphy ŽBridge and presented a quantitative conceptual model showing
Leeder, 1979; Bridge and Mackey, 1993; Mackey that whether a crevasse will heal or lead to an
and Bridge, 1995.. Field examples are documented avulsion depends upon sediment grain size, the ini-
for the Brahmaputra River ŽColeman, 1969., the tial depth of the crevasse and the ratio of crevasse, to
rivers on the Hungarian Plain ŽMike, 1975. and main channel bed slopes.
rivers in the upper Amazon foreland basins ŽDumont, Although the usual perception is that avulsions
1994.. Tricart Ž1965. and Gallais Ž1967. suggested are basically driven by development of floodplain
rapid local subsidence as a controlling mechanism relief through channel-belt aggradation, there is evi-
for a sequence of eastward avulsions of the Bani and dence for an additional mechanism. In studies on
Niger Rivers Žcentral Mali.. Makaske Ž1998, pp. arid and semi-arid anastomosing river floodplains,
168–169. elaborated on the relationship between the the loss of channel-flow capacity caused by in-chan-
half-graben tectonic framework and the avulsion his- nel fluvial deposition has been suggested as a driving
tory of this region. mechanism of avulsions. On the Cooper Creek and
Once the conditions on the floodplain are Upper Darling River floodplains, flood regimes are
favourable for avulsion, a trigger is needed to initiate typically flashy, and bankfull discharge is exceeded
the process. Jones and Schumm Ž1999. recently pub- far less than once every year ŽRiley and Taylor,
lished a useful overview of avulsion triggers and also 1978; Taylor and Woodyer, 1978; Woodyer et al.,
introduced the concept of the avulsion threshold, i.e., 1979; Rust, 1981; Schumann, 1989; Gibling et al.,
a state of extreme channel instability resulting in 1998.. Due to this, floodplain sedimentation rates are
avulsion. As explained above, aggradation may move relatively low ŽTable 2., and most sediment load is
a channel towards the avulsion threshold. The closer deposited within the channels during base-flow con-
a channel is to the avulsion threshold, the smaller the ditions. The in-channel deposits may be fine-grained
event needed to trigger the avulsion. Triggers may and frequently have a bench morphology ŽRiley and
determine the time as well as the location of avul- Taylor, 1978; Taylor and Woodyer, 1978; Woodyer
sion. A large flood usually determines the time of an et al., 1979; Schumann, 1989; Gibling et al., 1998..
avulsion ŽBrizga and Finlayson, 1990; Mack and Consequently, the channel looses capacity to accom-
 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196 161

modate the next flood and becomes liable to avul- starts as a small lobate crevasse splay that gradually
sion. Loss of channel-flow capacity as a driving develops into a large elongate crevasse splay com-
mechanism of avulsions may not be exclusively tied plex with a highly irregular planform. On the incipi-
to slowly aggrading floodplains in arid climates. ent crevasse splay unstable multi-channel patterns
Makaske Ž1998, pp. 93–96. suggested a similar exist. Along with splay progradation into the wet-
mechanism for the more rapidly aggrading flood- lands, the channels on the older part of the splay
plain of the anastomosing upper Columbia River in complex, start building natural levees and become
temperate humid western Canada ŽTable 2., where more stable. During progradation, channelized flow
measurements and calculations of sediment transport may bifurcate at channel-mouth bars, resulting in the
indicate substantial storage of coarse sediment load formation of separate splay lobes with their own
on the beds of laterally stable channels. If bed aggra- channels, prograding into the wetlands ŽFig. 4.. Of-
dation locally outpaces levee aggradation, flow is ten the developing channels rejoin further down-
forced out of the channels onto the floodplain where stream surrounding small floodbasin areas ŽSmith et
new channels are cut. This avulsion-mechanism need ´
al., 1989, pp. 7–8; Smith and Perez-Arlucea, 1994,
not be in conflict with the notion that avulsions are their fig. 4.. This type of anastomosis is considered a
driven by the evolution of floodplain relief, but stage in the avulsion cycle and therefore is relatively
rather complements it. Another avulsion-mechanism short-lived. The final stage is a single channel, deeply
was presented by Schumm et al. Ž1996., who ex- scoured through the inactive crevasse splay complex
plained avulsions of the anastomosing Ovens and and taking over all discharge from the temporary
King Rivers as a response to loss of hydraulic effi- multi-channel system. A similar avulsion-cycle was
ciency as channels became more sinuous with time. described by Van Gelder et al. Ž1994. for the Yellow
In any case, the chances for avulsion increase as River delta.
aggradation continues, but where and when will The above-described avulsion process, which in-
avulsion ultimately take place? During floods, the volves crevasse splay formation during very high
outer bends seem to be the most favourable spots for sediment supply to the floodbasin Žsee also Ethridge
avulsion. Smith et al. Ž1998. mention three reasons et al., 1999., is valid for rapidly aggrading systems
for this: Ž1. water-surface superelevation and higher only. Alternatively, in slowly aggrading systems,
velocities in the outer bend lead to higher erosive avulsion is a predominantly erosive process. Rust
power at potential avulsion sites; Ž2. inertia of flow Ž1981., for example, observed the absence of crevasse
in the outer bend directs overbank flow at a high splays and the presence of only poorly developed,
angle away from the channel flow towards the flood- discontinuous natural levees in the slowly aggrading
basins; and Ž3. levees in the outer bend tend to be Cooper Creek anastomosing system. He therefore
narrow due to lateral erosion and have a steep flood- argued that avulsion here takes place without prior
basinward slope. Additionally, Smith et al. Ž1998. crevasse splay formation. Schumann Ž1989. de-
showed with a numerical flow model how a chute scribed the formation of avulsion channels in the arid
cut-off could have triggered the 1870s avulsion of Red Creek system after only minor crevasse splay
the Saskatchewan River by raising the water-surface formation. In contrast to the process described by
elevation in the cut-off bend. Nevertheless, predict- Smith et al. Ž1989., channelization of flow in the
ing the exact time of avulsion is impossible, due to Red Creek system starts downstream, where the
the random nature of extreme floods and the various avulsion flow rejoins the main channel, and gradu-
triggers. Bridge and Leeder Ž1979. therefore argued ally travels upstream to the point where the avulsion
that it should be treated as a stochastic process. was initiated. Gibling et al. Ž1998. and McCarthy et
During the process of avulsion as described by al. Ž1992. proposed a similar mechanism for the
Smith et al. Ž1989. for the lower Saskatchewan formation of avulsion channels at Cooper Creek and
River, anastomosis developed within the avulsion in the Okavango Delta, respectively. An exceptional
belt Ži.e., the zone of the floodplain affected by an feature in the Okavango Delta, is that water from the
avulsion., resulting from contemporaneous scour of main channel percolates to the backswamps through
multiple channels. In their model, the avulsion belt permeable banks that are composed solely of vegeta-
162 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196

Fig. 4. Anastomosing channel pattern within a developing avulsion belt. From the initial massive crevasse splay complex shown on the left,
narrow lobes with channels prograded to the right, eventually rejoining and enclosing parts of the floodbasin. The trunk channel on the left
may be part of a larger-scale anastomosing system not shown here Žmodified from Makaske, 1998..

tion, as well as spilling over the banks. A slightly avulsion to be frequent. Nine major avulsions since
different mode of channel formation was described 4000 and 5400 BP, respectively, could be recognized
by Taylor Ž1999. for the anastomosing Fitzroy River. in floodplain-wide cross-sections of these systems
In this system, high-magnitude overbank flows are ŽMakaske, 1998, pp. 70–76; Morozova and Smith,
able to scour new channels on the floodplain that 1999.. Both systems are characterized by high flood-
initially lack upstream and downstream connection plain aggradation rates ŽTable 2.. The scarce field
with existing channels, but that gradually may extend evidence suggests that among different anastomosing
to become part of the anastomosing system. Aslan rivers, avulsion frequencies decrease with decreasing
and Blum Ž1999. inferred two different styles of floodplain aggradation rates, in accordance with re-
avulsion for Texas Gulf Coastal Plain rivers in re- sults of simulation studies ŽBryant et al., 1995. and
sponse to different floodplain aggradation rates. Dur- ¨
studies of fluvial deltaic sequences ŽTornqvist, 1994;
ing periods of low aggradation, avulsion took place Stouthamer and Berendsen, 1997.. In the slower
by reoccupation of abandoned channels on the flood- aggrading upper Inland Niger delta ŽTable 2., for
plain, with very little avulsion-related sediment example, only three major avulsions occurred during
deposition. In contrast, during periods of high aggra- the Holocene ŽMakaske, 1998, pp. 159–170.. Also
dation, avulsion came about by diversion into flood- in the extremely slowly aggrading anastomosing sys-
basins leading to metres-thick successions of avul- tems of the Channel Country wincluding Cooper Creek
sion deposits. ŽTable 2.x avulsion seems to be a rare process.
Channel chronologies for some anastomosing sys- Channel chronologies are absent for this arid system,
tems with good radiocarbon control, the upper due to poor conditions for radiometric dating, but air
Columbia and lower Saskatchewan Rivers, prove photos show no signs of avulsions over an area of
 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196 163

8000 km2 since the 1940s. The trees lining the linked to a low floodplain gradient, which therefore
anastomosing channels are hundreds of years old, is a major underlying cause of anastomosis. The
indicating stability over this period ŽG.C. Nanson, effect of aggradation, however, is ambiguous, since
personal communication, 2000.. it obviously favours frequent avulsion, while to some
Anastomosing rivers systems with a low avulsion extent it also favours the rapid abandonment of old
frequency, seem to maintain a multi-channel state by channels.
very slow abandonment of old channel segments
after avulsive formation of bypasses. Over long in-
3.3. Lateral channel stability
teravulsion periods, these anastomosing rivers are
virtually static multi-channel systems. Little is known
about the factors determining the rate of channel The lateral stability of the individual anastomos-
abandonment after avulsive switching. Given the ing channels has been pointed out in many studies
characteristics of the Channel Country anastomosing Že.g. Smith, 1976, 1983, 1986; Smith and Smith,
systems ŽGibling et al., 1998., where avulsion fre- 1980; Rust, 1981; Nanson et al., 1986., so there
quency is believed to be very low, the following appears to be a bias towards straight channels in
factors are considered important: Ž1. a low sediment anastomosing rivers. What is the reason for this?
load causing slow siltation of channels; Ž2. tough, Lateral stability of river channels is generally
strongly consolidated floodplain muds in combina- described as a function of slope, discharge and sedi-
tion with low stream power, allowing only slow ment composition of the banks. A basis for this idea
enlargement of channel-flow capacity of new chan- was provided by the classic paper of Leopold and
nels by bed scour and bank erosion; Ž3. little gradi- Wolman Ž1957.. Plotting channel slope against bank-
ent advantage of avulsion channels over older chan- full discharge, they found that braided and meander-
nels on the low-gradient floodplain due to poor ing channels could be separated by the line:
development of alluvial ridges; and Ž4. limited influ- S s 0.012 Qy0 .44
bf
ence of vegetation due to aridity. The latter contrasts
with plugging of old channels by woody debris jams in which: S s channel gradient Ž – .; Q bf s bankfull
and vegetation succession causing organic channel- discharge Žm3rs..
fills in temperate humid systems. Straight channels were also plotted but were not
In summary, frequent avulsion andror slow aban- discriminated by the above relationship in their dia-
donment of old channels leads to continuing exis- gram ŽFig. 5.. Schumm and Khan Ž1972. observed a
tence of contemporaneously active channels on the straight–meandering–braided sequence of channel
floodplain. During the avulsion process, splitting and patterns in a series of experiments with increasing
rejoining of the diverted flow may also cause tempo- valley gradient at constant discharge ŽFig. 6.. In the
rary anastomosis within the avulsion belt. Frequent same series of flume experiments, they showed that
avulsion is most importantly caused by: Ž1. rapid an increase in suspended sediment load has a stabi-
channel-belt aggradation; Ž2. rapid in-channel depo- lizing effect on the river banks. Later, however, it
sition leading to shallowing; and Ž3. frequent occur- appeared that the duration of their experiments was
rence of avulsion triggers. Common triggers are too short to reach equilibrium channel patterns ŽCar-
hydrological floods and obstructions temporarily re- son, 1984a, p. 331..
ducing channel-flow capacity. Slow abandonment of Discharge and slope can be combined into the
old channel segments is expected to be mainly deter- single parameter of stream power. Stream power is
mined by: Ž1. a low sediment load; Ž2. tough, strongly often used as a tool to investigate the lateral stability
consolidated floodplain muds in combination with of river channels Že.g. Chang, 1979; Ferguson, 1981;
low stream power of newly created channels; Ž3. Richards, 1982; Keller and Brookes, 1984; Brown,
little gradient advantage of avulsion channels over 1987; Nanson and Croke, 1992; Van den Berg, 1995;
older channels; and Ž4. limited influence of vegeta- Lecce, 1997; Makaske, 1998.. One can distinguish
tion. In general, most factors causing frequent avul- between gross stream power and specific stream
sion and slow abandonment of old channels are power. Gross stream power is defined as the rate at
164 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196

Fig. 5. Values of slope and bankfull discharge for various natural channels and a line separating fields of braided and meandering channels
Žafter Knighton, 1984; modified from Leopold and Wolman, 1957..

which potential energy of water flowing downhill is sinuous channels tend to have low specific stream
supplied to a unit wet channel length: power Ž1–60 Wrm2 ., while actively shifting low-
sinuosity channels have high stream power Ž) 100
V s g gQS
Wrm2 .. Actively meandering channels have inter-
in which: V s gross stream power ŽWrm.; g s mediate stream power values Ž5–350 Wrm2 .. These
density of water Žkgrm3 .; g s acceleration of grav- ranges of specific stream power, however, show
ity Žmrs 2 .; Q s discharge Žm3rs.; S s channel gra- wide overlaps.
dient Ž – .. Specific stream power Ž v . is gross stream Although certain ranges of slope–discharge com-
power divided by channel width Ž w .. binations or stream power seem to be associated with
Ferguson Ž1981., working with data from 95 certain channel patterns, it has been pointed out by
British rivers, concluded that inactive straight or various workers that these ranges characterize rather

Fig. 6. Relation between the slope of the surface on which a channel was formed Žvalley slope. and thalweg sinuosity Žafter Schumm and
Khan, 1972..
 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196 165

than predict those channel patterns ŽCarson, 1984a; older channels, since they necessarily support a part
Ferguson, 1987; Van den Berg, 1995.. This is partly of the peak discharge of the system. For the single-
due to the use of bankfull discharge and channel channel Blue River in Wisconsin, Lecce Ž1997. re-
slope, which are dependent on channel pattern. ported downstream decrease in gradient, channel ca-
Moreover, the channel width, necessary to calculate pacity and mean stream power, resulting in an in-
specific stream power, is dependent on the channel crease in magnitude and frequency of overbank
morphology. flooding. Knighton Ž1987. stated that: AAlluvial
Most important, however, is that stream power rivers with erodible boundaries flow in self-formed
represents a useful measure of the energy available channels which, when subject to relatively uniform
to modify the channel only if it is related to the governing conditions, are expected to show a consis-
sediment to be moved Že.g. Schumm and Khan, tency of form, or average geometry, adjusted to
1972; Carson, 1984a; Ferguson, 1987; Van den Berg, transmit the imposed water and sediment dischargesB.
1995.. Excess stream power above the threshold for One can seriously doubt whether this is true for
sediment entrainment is believed to relate to channel anastomosing rivers. Knighton and Nanson Ž1993.
behaviour ŽCarson, 1984a.. Therefore, Nanson and regard the formation of anabranches as a response to
Croke Ž1992. classified river floodplains on the basis the inability of the main channels to cope with high
of specific stream power in the channels and the magnitude discharges, because Athe resistivity of the
texture of the sediment on the floodplain. They banks constrains the size of the bankfull cross-sec-
described anastomosing river floodplains as having tional areaB. Some humid climate anastomosing sys-
gravel and sand in channels with a specific stream tems show prolonged annual flooding. For example,
power - 10 Wrm2 and abundant silt and clay on a mean of 45 days for the Columbia ŽLocking, 1983,
the floodplains. These floodplains were classified as p. 35., 50 days for the Magdalena ŽSmith, 1986. and
low-energy cohesive floodplains. even 100 days for the Solimoes˜ River ŽMertes, 1994.
Stream power considerations can be applied to floodplains were recorded. In semi-arid south-east
channel reaches, but not to river systems with multi- Australia, Rutherfurd Ž1994. described anastomosing
ple channel belts. Knighton and Nanson Ž1993. plot- reaches of the Murray River, where spilling over-
ted bankfull discharge against channel slope Žas in bank occurs more frequently than in the reaches
Fig. 5. for anastomosing, braided and meandering which have a single-channel morphology. Other
rivers. In their diagram, most anastomosing rivers semi-arid and arid climate anastomosing rivers expe-
plot within a zone below braided and meandering rience significant long-term in-channel accretion and
rivers, indicating low stream power. However, they therefore seem to be unable to transmit imposed
treated the anastomosing rivers as single hydraulic sediment load. ŽRiley and Taylor, 1978; Taylor and
entities. For explaining channel processes, individual Woodyer, 1978; Woodyer et al., 1979; Rust, 1981;
channel reaches of the anastomosing network should Schumann, 1989; Gibling et al., 1998.. Sedimentary
be considered because each channel adapts to its structures indicate that these in-channel deposits suf-
discharge independently of the other channels of the fer virtually no erosion during subsequent floods
system. Distribution of discharge over various chan- Že.g. Riley and Taylor, 1978, p. 100.. In the upper
nels also means reduction of stream power. The Columbia River a large fraction of the coarse bed
resulting relatively low stream power per channel material supplied from upstream is deposited in the
partly explains why channels in anastomosing sys- anastomosing reach ŽLocking, 1983; Makaske, 1998..
tems tend to be quite stable laterally Že.g. Harwood A part of this material may be trapped in large
and Brown, 1993, p. 747.. crevasse splays. Nevertheless, calculations and mea-
On low-gradient floodplains where mean stream surements of sediment transport suggest that supply
power is low, insufficient energy is available to of bedload greatly exceeds transport capacity within
enlarge channels, which therefore usually have little the anastomosing reach ŽMakaske, 1998, his Table
capacity to accommodate flood flows. Poor adjust- 3.10..
ment to flood flows for young channels in anasto- When considering the stable channel morphology
mosing networks results in slow abandonment of of anastomosing systems, time is an often neglected
166 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196

factor. Formation of wide alluvial ridges by active In summary, low stream power in combination
meandering of channels simply needs time. If chan- with cohesive bank material explains the lateral sta-
nels are subject to frequent avulsive switching, for bility of individual anastomosing channels. Low
instance due to rapid channel-belt aggradation, then stream power is mainly caused by a low floodplain
meanders may not fully develop because time of gradient that also favours avulsion ŽSection 3.2.,
channel-belt occupation is too short, although aver- although splitting of flow among different channels
age hydraulic conditions may favour slow lateral is an additional factor. The resistance against erosion
erosion and deposition. This means that, given con- of the banks is related to low stream power, since
stant rates of lateral migration, young channels seem overbank deposition of clay takes place predomi-
laterally more stable than old channels. This proba- nantly on low-gradient floodplains.
bly played a role in the longitudinal facies architec-
tural change of a palaeochannel described by
¨
Tornqvist et al. Ž1993.. They found evidence for 3.4. Climatic and base-leÕel changes
rapid downstream decrease in lateral migration of a
subrecent channel, together with a downstream de- Is anastomosis a state of transition, or is it an
creasing period of activity. Smith et al. Ž1989. sug- equilibrium form? In some cases, anastomosis is
gested that under rapid aggradation, avulsions inter- obviously transitional. For example, when anastomo-
rupt evolution of wide laterally accreted alluvial sis occurs within the avulsion belt as a result of
ridges in the anastomosing lower Saskatchewan splitting of the avulsive flow and contemporaneous
River. Likewise, Kolb Ž1963. and Saucier Ž1994, pp. scour of channels, as described for the lower
123–124. explained differences in size of Holocene Saskatchewan River by Smith et al. Ž1989.. They
Mississippi River meander belts by differences in considered a single channel belt as the normal state
time of occupation. Thus, in some cases apparent towards which the lower Saskatchewan River will
lateral stability of anastomosing channels may be return once the avulsion is completed. On a larger
caused by relatively short periods of activity due to scale, rivers that usually have a single channel belt,
frequent avulsion. may be anastomosed for relatively short periods
Climatic influence on lateral channel stability oc- during avulsive switching ŽFig. 7A., when the new
curs through the role of vegetation or duricrusts. In channel belt gradually takes over discharge from the
humid climates, vegetation and organic deposits have older channel belt Že.g. Brizga and Finlayson, 1990..
a stabilizing effect on river banks Že.g. Smith, 1976; The question here is whether all rivers naturally
Hickin, 1984; Harwood and Brown, 1993; Huang strive for a single channel belt, and if so, which
and Nanson, 1997.. Cairncross et al. Ž1988. and conditions prevent them from reaching that state,
Stanistreet et al. Ž1993. also described channel stabi- thereby causing long-lived anastomosis.
lization by extensive peat growth in a semi-arid Some investigators regard anastomosis as an ex-
anastomosing system. In arid environments, duricrust pression of instability induced by climatic change
formation can be an important cause of channel ŽGarner, 1959, 1967; Baker, 1978.. Climatic change
confinement ŽTricart, 1959; Friend et al., 1979; Gib- from aridity to humidity will subject channels of the
ling and Rust, 1990.. drainage network to floods they cannot accommo-

Fig. 7. Diagrams showing theoretical channel chronologies for fluvial systems with anastomosis. Horizontal bars represent periods of
activity of individual channels in a floodplain-wide cross section. Time scale only indicates order of magnitude. Note that anastomosing
systems may range from short-lived to long-lived, and from very dynamic to semi-static, as a function of avulsion frequency and the time
needed for total abandonment of older channels. ŽA. Fluvial system with a low avulsion frequency and for most of the time a single active
channel belt. Brief periods of anastomosis occur during avulsive switching to a new channel belt. ŽB. Fluvial system in which relatively
frequent avulsions and slow abandonment of older channels after avulsion lead to continuous anastomosis with two coexisting channels. ŽC.
Fluvial system in which highly frequent avulsions and slow abandonment of older channels lead to continuous anastomosis with four
coexisting channels. ŽD. Fluvial system in which infrequent avulsions and extremely slow abandonment of older channels lead to continuous
anastomosis with four coexisting channels.
 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196 167

date. Bowler et al. Ž1978. believed that avulsion of Holocene conditions. In the arid Inland Niger Delta,
the Darling River was associated with hydrological degradation of the channel network due to in-channel
changes from late glacial arid to more humid fluvial and aeolian deposition is believed to have
168 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196

been an important cause of avulsions at the onset of Whether a single channel belt is the norm for
periods with increasing discharge ŽGallais, 1967, p. alluvial rivers, with anastomosis as an expression of
55; McIntosh, 1983; Makaske, 1998, pp. 170–171.. instability remains open due to a poor hydraulic
Little is known about the response time of river understanding of anastomosis. However, it is a fact
systems to altered hydrological regimes. that some anastomosing rivers are long-lived. Anas-
An important indirect climatic influence is the rise tomosis of the upper Columbia River has persisted
in sea level in late Pleistocene and Holocene times. since nearly 3000 BP at least ŽMakaske, 1998, p.
Such a rapid base-level rise can be considered as an 75., while Morozova and Smith Ž1999, their Fig. 8.
instability imposed upon a fluvial system. Smith and suggest coexistence of channels in the lower
Smith Ž1980. suggested: AIn fluvial systems adjoin- Saskatchewan River since approximately 4000 BP.
ing marine basins, rapid sea-level rise might provide Knighton and Nanson Ž1993, p. 622. even suppose
a downstream control for upstream alluviation and that Cooper Creek has maintained as an anastomos-
possible development of channel anastomosisB. The ing system for over 50,000 years; however, coexis-
idea of downstream control ŽMackin, 1948. was first tence of channels over this period could not be
applied to anastomosing rivers by Smith Ž1973.. proven. Nanson and Huang Ž1999. stated that
Rapidly aggrading alluvial fans, deposited by tribu- anabranching Žincluding anastomosing. rivers de-
taries entering an alluvial valley, provided local ele- velop in order to maintain or enhance water and
vations of base level and caused upstream anastomo- sediment throughput across extensive low-gradient
sis. He stated that: AUnder such conditions grade floodplains. Supporting this statement with an analy-
should decrease with aggradationB ŽSmith, 1973, p. sis of basic hydraulic relationships, they considered
203.. The combination of rapid aggradation and anastomosing rivers as hydraulically efficient forms,
reduction of gradient favours avulsion frequency, more or less in equilibrium with imposed water and
while a low gradient also favours lateral channel sediment discharges. Makaske Ž1998, p. 93. ques-
stability. Rapid sea-level rise is considered a main tioned the universal applicability of this concept and
factor for the coexistence of many laterally stable demonstrated the inefficiency of the anastomosing
delta distributaries in the mid-Holocene Rhine– upper Columbia River and its inadequacy to trans-
¨
Meuse delta ŽTornqvist ¨
et al., 1993; Tornqvist, 1993a, port the coarse bedload fraction.
1994; Makaske, 1998, p. 229.. Evidence from other If the concept of anastomosis as a state of disequi-
near-coastal areas is very limited. Another mecha- librium is right, long-lived anastomosis points to
nism of base-level rise, important in some continen- continuous or repeated disturbance by allogenic con-
tal arid settings, is the formation of large dune fields trols. A type of continuous disturbance can be rapid
blocking the course of a river. Jacobberger Ž1988a, base-level rise, while repeated disturbance is most
p. 356. mentioned this as a prime cause for anasto- likely associated with the effects of climatic changes
mosis in the Inland Niger Delta. on catchment and floodplain. Continuous or repeated
Tectonic or isostatic movements provide another disturbance causes relatively high avulsion fre-
cause for Žrelative. base-level rise. Local uplift of the quency, leading to continuous coexistence of chan-
riverbed reduces the river gradient upstream of the nels ŽFig. 7B–C.. On the other hand, in a number of
uplift, thereby inducing anastomosis. Evidence to anastomosing river systems, avulsions are not caused
support this mechanism is given by Burnett and by external disturbance of the fluvial system, but
Schumm Ž1983. and Ouchi Ž1985.. On a larger scale, rather seem autogenic Že.g. Schumann, 1989;
rapidly subsiding foreland basins are considered ideal Schumm et al., 1996; Gibling et al., 1998.. Little is
settings for anastomosing river systems by Smith and known quantitatively about avulsion frequencies in
Putnam Ž1980. and Smith Ž1986.. Alternatively, these systems, but at least for the Channel Country
Bakker et al. Ž1989. described Late Pleistocene anas- anastomosing systems they seem to be low Žsee
tomosing channel patterns in a subsiding, intramon- Section 3.2.. Since these systems probably evolve
tane strike-slip basin, whereas McCarthy Ž1993. extremely slowly under more or less stable external
found intracratonic half grabens to be important set- conditions, they can hardly be taken to represent a
tings for extensive modern anastomosing systems. state of disequilibrium ŽFig. 7D. and seem funda-
 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196 169

mentally different in process from anastomosing sys- strongly with an increase in avulsion frequency.
tems caused by external disturbance. At the moment, Channel chronologies for anastomosing rivers pub-
it may be most appropriate to characterize long-lived lished by Makaske Ž1998, his fig. 3.11. and Moro-
anastomosis in general as a state of dynamic equilib- zova and Smith Ž1999, their fig. 8. resemble Fig.
rium, with avulsions maintaining a multi-channel 7B–C. Holocene records of avulsion frequency and
system, while older channels are slowly abandoned. periods of activity of channels for the Rhine–Meuse
¨
delta ŽTornqvist, 1994; Stouthamer and Berendsen,
3.5. A conceptual genetic model for long-liÕed anas- unpubl. data. show no systematic changes in periods
tomosis of activity, along with marked changes in avulsion
frequency and the number of coexisting channels,
With respect to genesis and lifetime, four basic suggesting that the assumption of more or less con-
types of anastomosing rivers have been identified in stant channel lifetimes in Fig. 7A–C is realistic.
the preceding sections: Ž1. short-lived anastomosis Thus, avulsion frequency seems to be one of the
within an avulsion belt; Ž2. short-lived, transitional prime controlling factors of anastomosis. The next
anastomosis during a single avulsion, due to tempo- question is in which alluvial setting avulsion fre-
rary coexistence of an older trunk channel and a new quency has its optimum. Below, this question and
avulsion channel; Ž3. long-lived, highly dynamic the relationship between channel morphology Žde-
anastomosing systems, constantly rejuvenated by fre- termined by lateral channel stability. and long-lived
quent avulsions caused by allogenic controls; and Ž4. anastomosis Ždetermined by avulsion frequency. will
long-lived, semi-static anastomosing systems, which be explored in a simple conceptual model.
seem in equilibrium with stable external conditions. A spatial distribution of alluvial river types which
Since these four types are morphologically quite reflects the influence of a downstream reduction in
similar, anastomosing rivers can be considered an floodplain gradient and a downstream increase in
example of equifinality, i.e., different combinations aggradation rate Žassociated with downstream con-
of processes or causes produce a similar form. Con- trol. is shown in Fig. 8. The thickness of the bars
sequently, it is hard to define one universal genetic represents the probability of occurrence of a certain
model for anastomosing rivers. river type as a function of its position on a scale
Fig. 7A–C shows how a rise in avulsion fre- from proximal to distal. Because of a reduction of
quency results in long-lived anastomosis. An as- gradient downstream, lateral channel stability in-
sumption is that channel lifetimes do not decrease creases as well as avulsion frequency. This leads to a

Fig. 8. Spatial distribution of alluvial river types from proximal to distal.

170 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196

high probability of an anastomosing system com- determine the morphology of the river system at the
posed of straight channels near the mouth. Moreover, floodplain scale and the channel-belt scale, respec-
conditions of base-level rise may cause a high aggra- tively. At the floodplain scale, the avulsion fre-
dation rate downstream, and enhance the chance quency determines whether or not long-lived anasto-
for anastomosis. Upstream, conditions are less mosis will occur. Optimum conditions for a high
favourable for anastomosis. Here, locally high avul- avulsion frequency causing anastomosis obviously
sion frequencies on alluvial fans do not cause anasto- exist on a low-gradient floodplain subject to rapid
mosing systems. base-level rise forcing rapid aggradation, while am-
The relationships among the factors determining ple avulsion triggers exist, such as log and ice jams
the morphology of a fluvial system on an alluvial and a strongly peaked flow regime. At the scale of
plain are shown in Fig. 9. Two groups of processes individual channel belts, lateral channel stability de-

Fig. 9. Conceptual model showing various factors determining the morphology of a fluvial system on an alluvial plain.
 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196 171

termines the place of a channel belt in the braided– tant features of these systems. Nanson and Huang
meandering–straight sequence. In addition to stream Ž1999. proposed a hydraulic explanation for this type
power, lateral channel stability is also strongly influ- of anastomosing river. For the geological record,
enced by the sediment composition of the banks, however, the importance of long-lived, frequently
which is influenced by the type of sediment supplied avulsing anastomosing rivers is paramount, since
by the catchment. Another important factor promot- these systems tend to be rapidly aggrading and pro-
ing bank stability is the binding force of vegetation. duce thick sequences of anastomosing river deposits.
Since the occurrence of anastomosis determines the
discharge per channel and thereby channel stream
power, floodplain-scale processes are linked to chan- 4. The sedimentary facies of anastomosing river
nel-belt-scale processes. In addition, floodplain gra- systems
dient is an important factor at both levels.
Concerning the factors ‘floodplain gradient’ and
4.1. Introduction
‘aggradation rate’, Table 2 shows essentially two
groups of anastomosing rivers: Ž1. the temperate and
tropical humid climate anastomosing rivers which all The deposits of anastomosing rivers can be ex-
have high floodplain aggradation rates, while some pected to be relatively abundant in the stratigraphic
of the gradients are relatively high as well; Ž2. the record due to high aggradation rates in many anasto-
arid, semi-arid and subarctic anastomosing rivers, mosing river systems. Identification of ancient anas-
which show low floodplain aggradation rates or inci- tomosis gives important clues for interpretation of
sion as well as relatively low gradients. It should be the large-scale palaeogeographical setting ŽGalloway
stated that the number of modern anastomosing rivers and Hobday, 1983, p. 51.. In this part of this paper, I
studied is much too low to allow a generalization of will deal with the question how we can recognize the
the suggested relationships between climate, flood- deposits of anastomosing rivers in the stratigraphic
plain aggradation rate, and channel gradient. Never- record. This requires analysis of their alluvial archi-
theless, the reported modern anastomosing rivers tecture as well as their sedimentary facies.
either have: Ž1. low gradients; Ž2. high floodplain The term ‘fluvial architecture’, nowadays more
aggradation rates; or Ž3. both. In the latter case, commonly termed ‘alluvial architecture’, was intro-
conditions for anastomosis according to the pre- duced by J.R.L. Allen A . . . to encompass the geome-
sented conceptual model are optimal. In the first two try and internal arrangement of channel and over-
cases, conditions for anastomosis seem to be less bank deposits in a fluvial sequenceB ŽMiall, 1996, p.
favourable. The role of factors now grouped as 34.. The term ‘facies’ is used in a purely descriptive
‘triggers’ is thought to be especially relevant for sense in this paper, following the definition of Read-
those anastomosing rivers. At present, these factors ing Ž1996, p. 19.: AA rock facies is a body of rock
are still poorly documented. Additionally, in a num- with specified characteristics. Ž . . . . If fossils are
ber of anastomosing river systems, in-channel ag- absent or of little consequence and emphasis is on
gradation Žnot necessarily associated with rapid the physical and chemical characteristics of the rock,
base-level rise. may be much more important than then the term lithofacies is appropriateB.
floodplain aggradation as a driving force of avul-
sions. However, very few data exist on rates of 4.2. The sedimentary enÕironments and lithofacies of
in-channel aggradation. modern anastomosing riÕer systems
Thus, it is clear that the presented model does not
equally well explain the origin of all types of long- Above, different genetic types of anastomosing
lived anastomosing rivers. Long-lived, semi-static rivers have been identified ŽSections 3.4 and 3.5..
anastomosing systems ŽFig. 7D. that exist under These different types partly generate different de-
stable external conditions, do not seem to depend on posits, with respect to lithofacies as well as alluvial
a high avulsion frequency. Very slow abandonment architecture. Additionally, differences in climatic set-
of older channels and low sediment load are impor- ting provide another fundamental control on the
172 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196

character of anastomosing river deposits. Just as it is gin, although typically more rich in sand, were de-
impossible to present one genetic model for anasto- scribed by Elliott Ž1974. from deltaic interdistribu-
mosing rivers, it is also impossible to present one tary bays. Close to the avulsion channels, natural
universal facies model for anastomosing rivers. The levee deposits rest on top of avulsion-belt deposits.
types of anastomosing river settings with associated Avulsion channels may form a temporary anasto-
lithofacies that will be discussed below by no means mosing network within the avulsion belt ŽFig. 4..
give a complete picture of the full range of anasto- Infilling of these channels with sandy bedload results
mosing river sedimentary environments. They only in stringers of sandy channel deposits, which are
represent what already has been discovered, after inset in the predominantly silty body of avulsion-belt
investigation of a limited number of modern exam- deposits ŽFig. 10.. During progradation of the avul-
ples. Especially with respect to sedimentary struc- sion belt, deposition of a thick fine-grained sequence
tures, information is still scarce; most studies pre- precedes scour of avulsion channels. Many of the
dominantly present textural facies. anastomosing avulsion channels are abandoned be-
fore they have scoured through the base of the thick
avulsion deposits. Hence, their sandy channel-fills
4.2.1. Short-liÕed anastomosis within an aÕulsion are fully surrounded by finer avulsion-belt deposits.
belt This is an important diagnostic feature of this type of
The deposits associated with anastomosing chan- anastomosing river deposits. Nevertheless, bigger
nels within an avulsion belt were only described avulsion channels may scour into the pre-avulsion
from the lower Saskatchewan River ŽSmith et al., deposits.
´
1989; Smith and Perez-Arlucea, ´
1994; Perez-Arlucea Sand bodies within the avulsion-belt deposits show
and Smith, 1999.. At present, it is still unclear to a great variety in shapes and sizes. Channel-fills tend
what extent the occurrence of this type of anastomos- to be ribbon-shaped with widthrthickness ratios -
ing river sedimentary environment is tied to the 10. Their thickness is typically 1–3 m, although
specific setting of the lower Saskatchewan River: a some exceed 7 m. Laterally extensive wup to nearly 1
wide, rapidly aggrading floodplain in a temperate km wide ŽN.D. Smith, personal communication,
humid climate. Sedimentation rates of this system 2000.x, sometimes discontinuous sand sheets repre-
are strongly variable in time and space ranging from sent crevasse splays and channel mouth bars. Typical
0.87–1.67 mmryear for subrecent peat and organic- ´
thicknesses are 0.5 to 2 m ŽPerez-Arlucea and Smith,
rich deposits away from active channels ŽMorozova 1999..
and Smith, 1999, their table 1. to 33 mmryear, The avulsion of the lower Saskatchewan River in
locally in the recent avulsion belt ŽSmith et al., 1989, the 1870s has affected over 500 km2 of wetlands up
p. 20.. to today ŽSmith et al., 1998., with avulsion-belt
The avulsion-belt deposits of the lower deposits up to 3.5 m in thickness. This suggests that
Saskatchewan River show wide textural variation, avulsion-belt deposits may be a volumetrically im-
reflecting proximal–distal relationships, location of portant element of the floodplain deposits of long-
avulsion channels, sediment supply, and the stage of term aggrading, avulsive river systems.
development of the avulsion. Sandy crevasse splay
deposits Žsee Section 4.2.2 for description. are com-
mon depositional units near the trunk channel and 4.2.2. Long-liÕed, rapidly aggrading anastomosing
avulsion channels. Away from channels and in more riÕers in a humid setting
distal parts of the avulsion belt, the lens-shaped body Despite differences in scale and geological set-
of avulsion-belt deposits is predominantly built up of ting, humid climate, rapidly aggrading anastomosing
silt, underlain and often capped by organic-rich de- rivers show remarkable similarity in sedimentary
posits. Vertically, these silty deposits, which are environments and lithofacies. A widely used lithofa-
volumetrically dominant over sandy avulsion-belt cies model for this type of anastomosing rivers is
deposits, are characterized by a coarsening-upward mainly based on a few well-documented examples in
sequence. Coarsening-upward units of a similar ori- western Canada. A major difference with the model
 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196 173

Fig. 10. Hypothetical section across a former avulsion belt where a new trunk channel has established, based on field evidence from the
lower Saskatchewan River. The cross-section shows avulsion deposits sandwiched between older alluvium and a thin veneer of
post-avulsion materials derived from a new alluvial ridge. Avulsion deposits are predominantly silty and clayey sediments enclosing
variously shaped and sized sand bodies formed by splays Žthinnest., anastomosing channels Žthicker. and the final new alluvial ridge
Žthickest. Žafter Smith et al., 1989..

discussed above is that anastomosis in this model is cross-valley alluvial fans. An exception is the lower
not the product of a single avulsion, but a result of Saskatchewan River, which is rather a medium-scale
repeated avulsions driven by continuous rapid flood- river with a wide floodplain. Besides well-developed
plain aggradation. Other important elements of this anastomosis within the recent avulsion belt Žsee Sec-
model are abundant bank vegetation and erosion-re- tion 4.2.1., the lower Saskatchewan River also dis-
sistant peat layers, promoting lateral channel stabil- plays a larger-scale anastomosing pattern, caused by
ity. repeated avulsion during the middle and late
The investigated rivers in western Canada include Holocene, resulting in long-term coexistence of mul-
the Alexandra, North Saskatchewan, Mistaya, upper tiple channel belts ŽMorozova and Smith, 1999, their
Columbia and lower Saskatchewan Rivers, which Fig. 8.. Initially, downstream increasing isostatic
show average floodplain aggradation rates between rebound was believed to be a major cause of
0.6 and 1.8 mmryear. These are mainly relatively Holocene floodplain aggradation of the Saskatchewan
small-scale rivers in a proximal, montane setting River Že.g. Smith, 1983.; however, Morozova and
with abundant sediment supply to a narrow confined Smith Ž1999. argued that since 5400 BP differences
floodplain, where local base level is controlled by in isostatic rebound have been insignificant.
174 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196

In the first anastomosing river facies model ŽSmith the Columbia River ŽD.G. Smith, personal communi-
and Smith, 1980., it is shown how rapid vertical cation, 1998.x and relatively narrow. Widthrthick-
accretion in combination with restricted lateral ness ratios of around 7 ŽMakaske, 1998, p. 72. and
movement of channels produces a network of thick 11 ŽSmith, 1986, p. 186. were reported for recently
and narrow sand bodies in the subsurface ŽFig. 11.. formed channel sand bodies in the upper Columbia
The sand bodies are embedded in Žsometimes or- River. In lateral and vertical direction, channel de-
ganic. floodbasin fines, and the lateral lithological posits change abruptly into finer grained overbank
boundaries are rather abrupt. It is now realized that deposits. The channel-fill deposits range from gravel
the vertical scale of this diagram is exaggerated far ŽSmith and Smith, 1980. to silt ŽSmith, 1975., but
too much to give a realistic picture ŽD.G. Smith, predominantly they consist of sand. Sets of planar
personal communication, 1998.. tabular cross beds result from deposition by migrat-
In later publications, the core of the model re- ing sand waves. In addition, multi-storied fining-up-
mained unchanged, although it became recognized ward textural sequences were identified, which were
that crevasse splay deposits and avulsions were cru- interpreted as flood cycle deposits during channel
cial elements ŽSmith, 1983; Smith et al., 1989.. The aggradation. The above-mentioned characteristics
role of avulsion was a significant addition to the typify the fills of commonly occurring, vertically
model since this prevents the channel deposits from infilling, laterally stable anastomosing channels.
reaching unrealistic thicknesses, as was suggested in Nevertheless, small-scale Ž1 m thick. sets of inclined
Fig. 11. An improved textural facies diagram was heterolithic stratification ŽIHS., produced by narrow
recently published by Makaske Ž1998. ŽFig. 12.. laterally accreting point bars, were found locally in
On the basis of the research in western Canada, the channel-fills of the upper Columbia River ŽSmith,
six anastomosing river sedimentary environments and 1983; Thomas et al., 1987, p. 138.. In the lower
associated lithofacies were identified Že.g. Smith, Saskatchewan River, lateral point bar accretion is
1983.. The lake, marsh and mire environments and probably much more widespread since bigger mature
facies are laterally extensive and show little variabil- channels of this anastomosing system show a ten-
ity. These facies were estimated to make up 60% to dency towards meandering. This will result in chan-
90% ŽSmith, 1983. of anastomosing river deposits. nel sand bodies with widthrthickness ratios much
The channel-related environments and facies are lat- greater than the values mentioned above. Although
erally more restricted and internally more complex. not being typical, and therefore not included in the
Aggradation in anastomosing channels can result ‘classic’ facies model, it should be realized that such
in channel-fills that are remarkably thick w5–12 m in meandering channel elements may well be part of

Fig. 11. Textural facies model of a rapidly aggrading anastomosing river system, in a temperate humid, montane setting Žfrom Smith and
Smith, 1980; qSEPM, reprinted by permission..
 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196 175

Fig. 12. Textural facies model of the upper Columbia River ŽBritish Columbia, Canada., a rapidly aggrading anastomosing system in a
temperate humid, montane setting. Area shown measures ; 2 km in width. Vertical scale strongly exaggerated; thickness shown ; 10 m
Žfrom Makaske, 1998..

anastomosing river deposits, since anastomosis does nated fine sand and silt, which fine and taper out
not exclude meandering Žor even braiding. of indi- laterally in floodbasin deposits. These sediment bod-
vidual channels within the anastomosing network ies can be up to 4 m thick ŽSmith, 1983, p. 163. and
Žsee also Section 2.3.. Makaske Ž1998, pp. 219–226. 2 km wide ŽN.D. Smith, pers. commun., 2000. in the
interpreted subrecent anastomosing channel-fills on lower Saskatchewan River. Usually, they are topped
the humid coastal plain of the Rhine–Meuse delta to by floodbasin deposits such as lake or marsh de-
represent various modes of vertical and lateral accre- posits, or peat. Directly underneath levees, however,
tion, including lateral concave-bank bench accretion avulsion-belt deposits ŽFig. 10. may be present, re-
as described by Hickin Ž1979, 1986. and Page and ferring to the origin of the associated channel.
Nanson Ž1982.. CreÕasse splays are deposited in the floodbasin,
The natural leÕees, which flank the channels, are adjacent to spots where the natural levee is dissected
wedge-shaped in cross-section. They consist of lami- by a small channel drawing flow from the main
176 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196

channel. Being deposited by miniature fluvial sys- biased dataset, as they are all located in temperate
tems, their structure, texture and geometry tends to humid continental settings and represent small to
be very complex. Their planform is generally lobate medium-scale anastomosing rivers.
but can be highly irregular. In cross-section, crevasse Research in the tropical humid large-scale anasto-
splays are lens-shaped and up to 3 m thick. They mosing system of the Magdalena River in Colombia
show a coarsening-upward sequence at the base and ŽSmith, 1986., showed that ‘the model from western
a thin fining-upward sequence at the top. Their Canada’ has a more universal applicability. A re-
texture ranges from coarse silt to coarse sand with markable feature revealed by deep boreholes Žup to
granules. Cross-lamination as well as high-angle 55 m. was the vertically uniform trend in mean grain
cross-beds are the internal structures of the sandy size of buried Magdalena channel-fills, which was
deposits. Usually, crevasse splay deposits are sand- attributed to dominant vertical accretion of the chan-
wiched between floodbasin muds or peat in a con- nel deposits. Lateral accretion would have resulted in
formable fashion; however, crevasse channels may a fining-upward trend. The common occurrence of
incise through the splay deposits into the subsurface, organic litter is another typical feature observed in
while taking over a substantial portion of the flow of the Magdalena channel deposits. Mud interbeds
the trunk channel. This process may herald formation seemed to be absent in the boreholes, but mud balls
of a new channel belt in the anastomosing river were found in grab samples from the channel beds.
system. In the lower Saskatchewan River this in- The overbank deposits of the Magdalena generally
volves the evolution of the crevasse splay into a contain only thin accumulations of organic material,
highly complex avulsion belt Žsee Section 4.2.1.. in contrast to the overbank deposits of the lower
However, in most cases the feeding crevasse channel Saskatchewan River. This was attributed to the high
becomes plugged up again after some time and the clastic input associated with the foreland basin set-
crevasse splay is abandoned and invaded by vegeta- ting of the system. The Magdalena River floodplain
tion. has a very high average long-term aggradation rate
Marshes, lakes and mires are wetland environ- of 3.8 mmryear ŽTable 2. over 7500 radiocarbon
ments with a low bottom elevation relative to the years. Rapid aggradation is believed to be more or
natural levees and crevasse splays. In lakes and less equal to tectonic downwarping of the foreland
marshes, suspended load is deposited, while mires basin.
generally are too remote from the channels to receive
significant amounts of clastics. Marsh deposits con-
sist of organic and clastic mud. They may be lami- 4.2.3. Long-liÕed, slowly aggrading anastomosing
nated, but usually they are bioturbated. Lake deposits riÕers in an arid setting
consist of laminated clay and silt, although bioturba- Not all anastomosing river deposits comfortably
tion may also have destroyed any structure. Mires fit into the models described above. Especially some
are extensive and contain peat up to 3.5 m in long-lived, slowly aggrading anastomosing rivers in
thickness in the Saskatchewan marshes. However, arid climates show major differences in sedimentary
development of peat strongly varies among the in- environments and lithofacies. Although a number of
vestigated humid climate anastomosing rivers, such anastomosing rivers has been identified, com-
predominantly depending on influx of clastics in prehensive descriptions of sedimentary environments
relation to floodplain width. In proximal montane and lithofacies are presently only available for Cooper
settings with abundant supply of clastics to a narrow Creek in central Australia and the upper Inland Niger
floodplain, for example in the upper Columbia River, Delta in central Mali. Geologically, both systems
mires are rare and short-lived, resulting in few thin have a roughly comparable distal, intracratonic set-
peat beds ŽFig. 12.. ting. Nevertheless, differences in tectonic control
The cited examples from western Canada were result in differences in sedimentation rates: the tec-
the first detailed descriptions of anastomosing river tonically stable Cooper Creek floodplain aggrades at
sedimentology and therefore have set the standard. 0.04 mmryear, while the slowly subsiding Inland
However, these anastomosing systems represent a Niger Delta aggrades at 0.14 mmryear ŽTable 2..
 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196 177

Work of Rust Ž1981. on the arid Cooper Creek admixed siltrsand fraction in the clayey floodbasin
anastomosing system revealed the following impor- deposits. Levees and crevasse splays are relatively
tant differences with ‘the model from western fine-grained in this system. In Fig. 13 the sedimen-
Canada’: Ž1. a lower proportion of channel sands tary environments and textural facies of the upper
relative to overbank fines, due to lower channel Inland Niger Delta are summarized. This diagram is
density; Ž2. less organics in the overbank sediments not intended to be a universal facies model for
due to much lower plant growth; Ž3. absence of long-lived, slowly aggrading, arid anastomosing
crevasse splay deposits and fewer levee deposits, rivers, but is shown here rather as an example of an
since levees are subtle and discontinuous. Rust Ž1981. alternative anastomosing river facies model.
described the channel-sands as isolated sand stringers With very few modern examples investigated,
encased in a predominantly muddy sequence. Inter- generalizing about the sedimentary environments and
nally, the sand bodies typically showed planar lithofacies of long-lived, slowly aggrading, arid anas-
cross-stratification resulting from deposition as alter- tomosing rivers is unjustified. Another more funda-
nating side bars, underlain by trough cross-stratified mental problem is that Quaternary climatic oscilla-
sand and capped by horizontally and ripple-laminated tions in combination with slow aggradation rates
sand. The mud was commonly structureless due to prevented formation of thick sequences produced by
bioturbation and alternate shrinkage and swelling. such anastomosing rivers under equilibrium climate
Duricrusts and evaporite horizons occurred in addi- conditions. Therefore, in modern and subrecent envi-
tion to abundant desiccation cracks. ronments there is no opportunity to check our as-
Rust’s model was extended by Gibling et al. sumptions of what these sequences will look like.
Ž1998., who provided many details about the Cooper
Creek sediments. They showed that the Cooper Creek
channel bodies essentially comprise mud-rich chan- 4.2.4. Other long-liÕed anastomosing riÕers
nel-fills that are sandier towards their bases. They It is clear from the above that the sedimentology
reported lateral and vertical accretion of the channel of long-lived, slowly aggrading, arid anastomosing
deposits, often in the form of accretionary benches rivers still needs attention, since it is unclear how
with inclined heterolithic stratification wIHS as de- representative the examples studied to date are. Be-
fined by Thomas et al. Ž1987.x at higher levels. The sides that, some anastomosing river settings have
dimensions of the muddy channel bodies can be up never or barely been studied sedimentologically. In
to 7–10 m thick and ) 100 m wide. Widthrthick- particular, we lack information about cold-climate
ness ratios of these bodies were estimated at ) 15 anastomosing rivers. Plan views from, for example,
and possibly 100 or more, based on limited data. the Ob and Mackenzie Rivers, suggest that extensive
Notably heterolithic anastomosing channel-fills were anastomosing river systems exist in the subarctic
also described by Taylor and Woodyer Ž1978. and zone.
Woodyer et al. Ž1979. from semi-arid eastern Aus- Currently, the only sedimentologically studied
tralia. The heterolithic benches can be found at subarctic anastomosing river is the Attawapiskat
different levels in the channel. The lower benches River in central Canada ŽKing and Martini, 1984..
are characterized by cross-bedded fine sands with Since this river is incising estuarine clays and Pleis-
mud interbeds, while the upper benches consist of tocene tills that are uplifted due to istostatic rebound,
interlaminated sands and muds. only thin accumulations of anastomosing river de-
Makaske Ž1998. reported thick Ž4–5 m. vertically posits occur in this setting. Channel deposits in main
accreted clayey upper parts of channel-fills in the channels principally consist of thin boulder and peb-
semi-arid upper Inland Niger Delta. Sandy deposits ble lags over shoals eroded in underlying till, and
in the lower part of the channel-fills were formed junction bars Ždownstream from islands. mainly con-
during an earlier phase of predominantly lateral ac- sisting of sorted coarse to medium-sized sand, show-
cretion. Aeolian dune deposits are abundant in this ing well-developed fining upward. In secondary
anastomosing system, in the channels as well as on channels, only thin silt and clay drapes erosively
the floodplain. Aeolian sediments also occur as an overlie the hard substratum. Levees mainly consist of
178 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196

Fig. 13. Textural facies model of the upper Inland Niger Delta, a slowly aggrading anastomosing river system in a semi-arid climatic setting.
Area shown measures ; 40 km in width. Channel belts are oversized for clarity. Vertical scale strongly exaggerated; thickness shown ; 10
m. Remarkable features are the abundant aeolian lithofacies originating from formation of different types and generations of dunes, and the
relatively fine-grained channel deposits filling up oversized channels Žfrom Makaske, 1998..

laminated silt, while marshes show regular alterna- bles in the overbank sediments. Remarkably, crevasse
tion of silt with organic-rich layers. Further away splays are absent in this system, although frequent
from the channels, mires exist with up to 0.5 m thick ice jams in the lower reaches cause flooding and
grassy peat. Peat may also fill up abandoned sec- overbank sedimentation.
ondary channels. Important climatic indicators are Since the preservation potential of the deposits
the considerable amounts of ice-rafted sand and peb- described by King and Martini Ž1984. is extremely
 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196 179

low, from a sedimentological point of view it would 4.3. The lithofacies of interpreted ancient anastomos-
be more interesting to investigate long-term aggrad- ing riÕer systems
ing, subarctic anastomosing rivers in subsiding set-
tings. However, these may be scarce due to post-gla- Knowledge on the sedimentology of modern anas-
cial isostatic rebound in most subarctic regions. Other tomosing systems was first applied to ancient fluvial
gaps in our sedimentogical knowledge concern deposits by Le Blanc Smith and Eriksson Ž1979.,
rapidly aggrading arid anastomosing rivers and Smith and Putnam Ž1980. and Putnam and Oliver
slowly aggrading humid anastomosing rivers, but Ž1980.. At that time, still relatively little was known
these types of anastomosing rivers are probably less in detail about anastomosing river lithofacies, and
common than the types of anastomosing rivers stud- consequently the validity of these early interpreta-
ied hitherto. tions has been questioned later. Especially the work

Table 3
Alluvial–architectural characteristics of interpreted subrecent and ancient anastomosing river systems
Study location and data source Proportion Laterally wrt b of channel
of overbank connected channel sandstone
deposits Žin %. sandstone bodiesa bodies
Clifton Fm. ŽNew Brunswick, Canada.1 70 unknown F 73
Ft. Union Fm. ŽWyoming, U.S.A.. 2 45 yes 67
Ft. Union Fm. ŽWyoming, U.S.A.. 3 50–90 unknown 10
Cumberland Group ŽNova Scotia, Canada. 4 70 unknown 3–38
TatrotrPinjor Fm. ŽIndia. 5 65 unknown G 7–26
Wasatch Fm. ŽWyoming, U.S.A.. 6 80 yes 105–165
Raton Fm. ŽColoradorNew Mexico, U.S.A..7 75 unknown 2–120
Vrijheid Fm. ŽSouth Africa. 8 no data yes 50–500
St. Mary R. Fm. ŽAlberta, Canada.9 85 inferred 8–27
Pitalito Basin ŽColombia.10 no data yes no data
Cutler Fm. ŽNew Mexico, U.S.A..11 75 yes F 40
Dakota Fm. ŽUtah, U.S.A..12 60 unknown 7–20
Betuwe Fm. ŽThe Netherlands.13 90 yes 5–G 67
BrazeaurBelly R. Fm. ŽAlberta, Canada.14 no data yes 10
WillwoodrFt. Union Fm. ŽWyoming, U.S.A.15 70 yes - 10
Riverine Plain ŽNew South Wales, Australia.16 no data yes 70–1000
a
Planform pattern shows bifurcation andror convergence of sandstone bodies.
b
Widthrthickness ratio.
1
Rust and Legun Ž1983..
2
Flores and Hanley Ž1984..
3
Johnson and Pierce Ž1990..
4
Rust et al. Ž1984..
5
Kumar and Tandon Ž1985..
6
Warwick and Flores Ž1987..
7
Flores and Pillmore Ž1987..
8
Cairncross et al. Ž1988..
9
Nadon Ž1988, 1994., Currie et al. Ž1991..
10
Bakker et al. Ž1989..
11
Eberth and Miall Ž1991..
12
Kirschbaum and McCabe Ž1992..
13
¨
Tornqvist ¨
et al. Ž1993., Tornqvist Ž1993a., Makaske Ž1998..
14
Putnam Ž1993..
15
Kraus Ž1996., Kraus and Gwinn Ž1997., Kraus and Wells Ž1999..
16
Page and Nanson Ž1996..
180 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196

on the Mannville Formation ŽSmith and Putnam, Internally they either show fining upward Že.g. Flo-
1980; Putnam and Oliver, 1980; Putnam, 1982, res and Pillmore, 1987; Cairncross et al., 1988;
1983., which heavily relied on interpretation of geo- Kirschbaum and McCabe, 1992. or they are rela-
physical subsurface data, has given rise to alternative tively homogeneous in texture Že.g. Rust et al., 1984;
interpretations ŽWightman et al., 1987; see also Mi- ¨
Johnson and Pierce, 1990; Tornqvist et al., 1993;
all, 1996, p. 279.. Later, more convincing outcrop Nadon, 1994.. Grain size ranges from coarse to fine
studies were published ŽRust and Legun, 1983; Flo- sand. Conglomerate sometimes constitutes a lag near
res and Hanley, 1984; Rust et al., 1984., and at the erosional base of those bodies Že.g Flores and
present there is a number of well-described ancient Hanley, 1984; Warwick and Flores, 1987; Eberth
fluvial sequences that are believed to be of anasto- and Miall, 1991.. These lags may contain fragments
mosing river origin ŽTable 3.. The lithofacies and of mudstone, petrified wood, or coal. In some cases,
facies geometries of these ancient systems will be the ribbon bodies are less sandy and have a notably
dealt with together in this section. Besides ‘true’ heterolithic composition of interbedded sandstones,
ancient examples, a few subrecent anastomosing river siltstones and mudstones ŽKirschbaum and McCabe,
systems are also included in this review Že.g. Bakker 1992.. Ribbons may consist of multiple stories, sepa-
¨
et al., 1989; Tornqvist, 1993a.. rated by erosion surfaces ŽKraus, 1996.. The domi-
nant sedimentary structure is trough cross-bedding
4.3.1. Sandstone facies with subordinate planar cross-bedding, while near
Two geometric types of sandstone facies are dis- the top of the bodies ripple or climbing-ripple lami-
tinguished: ribbon bodies and tabular bodies. nation is usually present. The cross-bedded sets may
The ribbon bodies are characterized by a sharp, be up to 1.2 m thick. Lateral accretion stratification
scoured, often concave base and a gradational or is not a consistent feature of these sandstone bodies,
sharp flat top ŽFig. 14.. Their thickness is usually a but occasional examples are reported ŽRust and Le-
few metres, but may be up to tens of metres Že.g. gun, 1983; Rust et al., 1984; Warwick and Flores,
Warwick and Flores, 1987.. Normally, they have a 1987; Kirschbaum and McCabe, 1992; Nadon, 1994;
width in the order of tens to hundreds of metres. Makaske, 1998, pp. 201–202.. Some of these exam-

Fig. 14. Ribbon sandstone body in the St. Mary River Formation ŽAlberta, Canada.. The ribbon has a sharp concave base and a flat top, and
is approximately 50 m wide. Widthrthickness ratio is around 12. This sandstone body is interpreted as channel-fill deposits of an ancient
anastomosing river system ŽNadon, 1994..
 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196 181

ples can be classified as inclined heterolithic stratifi- deposits may be heterolithic and display mudrock
cation, due to the presence of mudstone drapes. partings ŽJohnson and Pierce, 1990.. Frequently, the
The tabular bodies typically have flat, non-erosive tabular bodies are attached like ‘wings’ ŽFig. 15. to
bases and either sharp or gradational lower and the top of ribbon sandstone bodies ŽEberth and Miall,
upper contacts. Their thickness is usually limited to a 1991; Kirschbaum and McCabe, 1992; Nadon, 1994;
few metres and often they are less than 1 m thick. Kraus, 1996.. Kraus Ž1996, p. 361. proposed two
Laterally, however, they can be hundreds to thou- origins for the wings of relatively small ribbon sand-
sands of metres wide before grading into finer- stone bodies: Ž1. deposition by sheet floods from a
grained deposits. Documented vertical textural trends main channel, resulting in lobate sandy crevasse
include: Ž1. upward fining ŽNadon, 1994.; Ž2. up- splay deposits, which are later scoured by developing
ward coarsening ŽFlores and Pillmore, 1987; War- crevasse splay channels Žrepresented by the ribbons.;
wick and Flores, 1987.; Ž3. upward coarsening fol- Ž2. overbank deposition from crevasse splay chan-
lowed by upward fining ŽRust and Legun, 1983; Rust nels during the final infilling of the channel. Wings
et al., 1984.; and Ž4. no vertical trend at all ŽJohnson originating from the latter process may be considered
and Pierce, 1990.. The sand grain size may be as levee deposits. Kraus, also reported ‘tiers’, i.e.,
coarse, but it is usually medium to fine, grading to laterally extensive sheet-like sandstone bodies that
silt. The dominant structures are ripples, climbing- connect the tops of multiple stratigraphically equiva-
ripples and parallel laminations, with occasional lent ribbon sandstones. This suggests an origin as
small-scale trough cross-beds and wave-ripple lami- crevasse splay with multiple coexistent crevasse
nation Že.g. Nadon, 1994.. The tops of these bodies channels. Facies architecture and types and arrange-
are often heavily rooted and bioturbated and may ment of palaeosols strongly favoured interpretation
show traces of palaeosol formation Že.g. Johnson and of such complexes as avulsion-belt deposits ŽKraus,
Pierce, 1990.. 1996; Kraus and Gwinn, 1997; Kraus and Wells,
The ribbon sandstone bodies are commonly inter- 1999., analogous to the deposits found in parts of the
preted as channel-fills, while the tabular bodies are recent avulsion belt of the lower Saskatchewan River
taken as crevasse splay or levee deposits. The levee ŽSmith et al., 1989.. Nadon Ž1994, his Fig. 8. also

Fig. 15. Close-up of lateral wing attached to the right-hand margin of the ribbon sandstone body of Fig. 14. The wing is capped by a
sequence of weathered mudrock sheets and erosion-resistant tabular sandstone bodies Žpositive relief.. The upper part of the sandstone
ribbon gradually passes into the wing. The wing is interpreted as levee deposits, whereas the tabular sandstone bodies are believed to
represent crevasse splay deposits ŽNadon, 1994..
182 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196

described channel sandstone ribbons with lateral rent or wave-ripple laminated, and laminae of fine
sandstone wings. Although he interpreted the wings sand may locally be present. The colours range from
as levee deposits, the fact that the wings partly also grey and green to yellow, brown and red Že.g. Kumar
underlie the erosive base of the channel sand body, and Tandon, 1985.. Grey and green mudrock may
rather indicates a crevasse splay origin for the wings, contain organic material like leaves, wood fragments
with subsequent channel scour and infilling. In other and seeds. Additionally, dark coaly intervals are
cases, the top of the sandstone body gradually passes typical for this type of mudrock facies. In contrast,
into the wing Že.g. Fig. 15., suggesting a levee origin yellow, and red mudrocks usually lack organics, but
for the wing. can have calcrete nodules or horizons Že.g. Rust and
Although natural levee and crevasse splay de- Legun, 1983; Kumar and Tandon, 1985.. Desicca-
posits may be preserved as sandstone bodies, it tion cracks of 1 m deep were found in this type of
should be reminded that these deposits in modern mudrock facies ŽRust and Legun, 1983.. In general,
anastomosing river environments also have impor- root mottling and burrows are common character-
tant fine-grained components, which are part of the istics of mudrock facies and are often found in
´
mudrock facies ŽSmith and Perez-Arlucea, 1994.. association with incipient palaeosols, which can be
Another point is that the ribbon sandstone bodies traced laterally in the sheets over long distances.
may not represent all channel facies, as channels Based on pedogenic characteristics, Kraus and Aslan
sometimes fill in with mud rather than sand Že.g. Ž1993. separated mudrocks into two categories. They
Page and Nanson, 1996.. Thick muddy sequences distinguished mudrocks with cumulative palaeosols
may occur along the margins of channel sandstone and mudrocks with simple Žnoncumulative .
bodies due to concave-bank bench accretion associ- palaeosols. Cumulative palaeosols tend to be well-
ated with confined meandering of palaeochannels in developed and typify massive claystones, whereas
anastomosing systems ŽMakaske, 1998, pp. 225– simple palaeosols show much weaker profile devel-
226.. opment in more heterogeneous mudrocks consisting
of alternating claystone, mudstone and sandstone
4.3.2. Mudrock facies ŽKraus, 1996; Kraus and Wells, 1999..
In this paper, the term mudrock is used, following The mudrock facies is usually interpreted to rep-
Lundegard and Samuels Ž1980., for a sedimentary resent a number of depositional environments. Inter-
rock which consists for more than 50% of grains laminated wedges of siltstones and mudstones adja-
finer than sand. Depending on silt–clay content, cent to ribbon sandstone bodies are assumed to be
siltstones, mudstones, or claystones can be distin- levee deposits, while lenses of siltstone and mud-
guished. stone, sandwiched between claystone beds, are likely
Usually in ancient anastomosing systems, the mu- to represent crevasse splay deposits or avulsion-belt
drock facies strongly dominates over the sandstone deposits Žsee next paragraph.. More finely laminated
facies. Mudrock beds have a sheet-like appearance and possibly wave-rippled mudrock is usually taken
ŽFig. 15., and the sheets are bounded or split by as lacustrine deposits Že.g. Nadon, 1994.. Faunal or
sandstone or coal facies. The thickness of the sheets palaeobotanical evidence Že.g. Van der Woude, 1983,
is highly variable and may range up to 20 m ŽEberth 1984. can be necessary to give this interpretation a
and Miall, 1991.. Alternatively, mudrock may be firmer basis, as bioturbation may have destroyed the
present as plugs overlying ribbon sandstone bodies primary structures. Often, rather structureless and
Že.g. Rust and Legun, 1983.. Eberth and Miall Ž1991. bioturbated Žperhaps mottled. mudrock is assumed to
also reported U-shaped mixed-fill units, comprising be a marsh deposit Žcf. Smith, 1983..
abundant laminated mudrock over relatively thin Interpretation of palaeosols in mudrock may pro-
basal Žpebbly. sandstone. vide relative indications of sedimentation rates. The
The mudrocks usually consist of interbedded silt- cumulative palaeosols distinguished by Kraus and
stone, mudstone, and claystone. The individual beds Aslan Ž1993. were interpreted to result from concur-
are either massive or horizontally laminated. The rent pedogenesis and deposition, indicating that de-
siltstone portion of the mudrock facies may be cur- position rate was slow enough to incorporate thin
 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196 183

increments of freshly deposited material into the of an arid climate. In contrast, dark grey and green
developing soil profile. On the contrary, simple colours and abundant organic matter are suggestive
palaeosols were believed to relate to rapid sedimen- of a humid climate. Caution must be used, however,
tation, preventing cumulative soil development. In in making these palaeoclimatic interpretations, as
this case, each soil profile represents a single deposi- these properties may show considerable local varia-
tional episode, which was followed by a period of tion, and they may partly represent the conditions of
pedogenesis. Based on these pedological interpreta- early diagenesis rather than the conditions of deposi-
tions and the arrangement of the sand bodies with tion. Miall Ž1996, pp. 437–442. gives an overview
respect to the palaeosols, simple palaeosol units could of the problems associated with palaeoclimatic inter-
be attributed to avulsion-belt andror levee deposi- pretation of overbank deposits.
tion ŽKraus, 1996; Kraus and Gwinn, 1997; Kraus
and Wells, 1999.. In this model, cumulative palaeosol 4.3.3. Coal facies
units represent overbank deposits formed far from Excluding the coal incorporated in abandoned
major channels and avulsion belts wincluding for channel plugs, the coal facies usually occurs in
example the marsh deposits as described by Smith seams, with blanket geometries. The seams com-
Ž1983.x. An example of subrecent fine-grained avul- monly are laterally extensive and can be traced for
sion-belt deposits associated with anastomosing rib- up to kilometres. The coal seams are usually thin,
bon sand bodies in a deltaic setting was presented by and intercalated mudrock beds frequently occur. In
Makaske Ž1998, pp. 220–223.. In this case, the lens general, thickness ranges up to 2 m. In systems
of predominantly clayey avulsion-belt deposits is where the mudrock facies is virtually absent Že.g.
sandwiched between peat layers. Likewise, a neigh- Cairncross et al., 1988., the coal facies strongly
bouring subrecent anastomosing system with com- dominates the anastomosing river lithofacies. In such
parable deposits and facies architecture earlier cases, the seam thickness may be up to 8 m ŽLe
¨
described by Tornqvist et al. Ž1993. can now be Blanc Smith and Eriksson, 1979.. Coal blankets are
interpreted as an abandoned avulsion belt. most frequently encased in mudrock, but scoured
Mudrock plugs are commonly interpreted as aban- upper or lateral contacts with ribbon sandstone bod-
doned channel deposits based on geometry, because ies are also common.
lithologically they can be identical to other mudrock The coal facies is interpreted to represent the mire
facies. Abandoned channels, however, may also al- depositional environment. Coal is not present in all
¨
most completely fill up with peat Že.g. Tornqvist et ancient anastomosing river systems. Absence or rar-
al., 1993. and end up as coal plugs. Mudrock plugs ity of coal has been attributed to high clastic influx
need not always to be underlain by a sandstone body. ŽNadon, 1994. or arid climatic setting ŽRust and
Some anastomosing systems comprise channels that Legun, 1983; Kumar and Tandon, 1985; Eberth and
carry little sandy bedload Že.g. Page and Nanson, Miall, 1991.. The thickness of coal seams is strongly
1996; Gibling et al., 1998.. When surrounded by influenced by compaction. They may represent sig-
overbank mud, mud-filled channels may be hard to nificant quantities of peat. For example, the esti-
recognize in the stratigraphic record. Ancient exam- mated proportions of pre-compacted and post-com-
ples of predominantly mud-filled channels are the pacted coal and carbonaceous mudstones differ from
U-shaped mixed fills described by Eberth and Miall 40% to 11%, respectively, in sections studied by
Ž1991.. These were interpreted to represent crevasse Kirschbaum and McCabe Ž1992.. McCabe Ž1984, p.
channels that tapped the upper levels of flow of main 20. cited peat-to-coal compaction ratios between
channels, resulting in channel infilling by suspended 1.4:1 and 30:1.
load under generally low-energy conditions.
The mudrock facies contains important palaeocli- 4.4. The alluÕial architecture of interpreted ancient
matic information indicated by its colours, organic anastomosing riÕer systems
content and palaeosols. Red and yellow colours,
calcrete Žnodules., desiccation cracks, and lack of In this section, three aspects of anastomosing
organic matter are commonly considered indicative river alluvial architecture will be discussed: Ž1. the
184 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196

proportion of overbank deposits relative to channel of channels, however, is not considered in these
deposits, Ž2. the lateral connectedness of the channel models.
sandstone bodies and Ž3. the geometry Žwidthrthick- The ancient river systems ŽTable 3. all show a
ness ratio. of the channel sandstone bodies. These high proportion of overbank deposits, which is in
properties are of considerable economic significance. agreement with data from the modern record. Where
In Table 3, data about these aspects for sixteen modern and ancient environments are compared, cor-
interpreted ancient and subrecent anastomosing river rection should be made for compaction of muds and
systems are summarized. organics. Compaction differences may explain why
the proportion of overbank deposits in the Holocene
Betuwe Formation ŽThe Netherlands. is higher than
4.4.1. Proportion of oÕerbank deposits in the ancient examples of Table 3.
Subsurface data from modern anastomosing rivers
suggest that large amounts of overbank deposits are 4.4.2. Lateral connectedness of channel sandstone
preserved ŽSmith and Smith, 1980; Smith, 1983, bodies
1986.. The proportion of overbank deposits pre- If preserved in the stratigraphic record, sandy
served in the stratigraphic record is controlled by channel-fills of anastomosing rivers will mainly oc-
several interrelated variables, most importantly: Ž1. cur as networks encased in overbank muds. For most
lateral mobility of the fluvial channels; Ž2. channel examples in Table 3, lateral connectedness of chan-
density; and Ž3. floodplain aggradation rate. Since nel sandstone bodies was reported. However, not
many modern anastomosing rivers have straight, lat- every network of sandstone bodies in the strati-
erally stable channels, overbank deposits tend to graphic record is of anastomosing river origin, since
escape lateral erosion. Rust Ž1981. believed that contacts between sandstone bodies can be erosive
channel density in arid anastomosing river systems is and do not necessarily originate from confluence or
significantly lower than in humid anastomosing river bifurcation of contemporaneously active channels
systems. In this respect he quoted values of 3% of Žsee also Schumm et al., 1996..
the floodplain occupied by active channels on the The degree of connectedness of the individual
arid Cooper Creek floodplain as opposed to 20% for sandstone bodies is an important variable in reservoir
the humid Alexandra River. These values, however, characterization. Essential is planform information
are strongly determined by large differences in basin on the sandstone bodies. In some cases, planform
width and probably are not representative for arid exposures are available in which channel sandstone
and humid anastomosing rivers in general. Two-di- body connectedness can be readily observed Že.g.
mensional Žcross-valley. simulation studies Že.g. Eberth and Miall, 1991.. In other cases, tight core
Bridge and Leeder, 1979; Bridge and Mackey, 1993. control is needed, which may be aided in subrecent
suggested a positive correlation between aggradation systems by geomorphological evidence for the sub-
rate and preservation of overbank deposits, which ¨
surface patterns ŽBakker et al., 1989; Tornqvist,
could be especially relevant for anastomosing river 1993a; Berendsen et al., 1994.. In the deeper subsur-
systems, since a number of them show high aggrada- face, pressure data can be used to test fluid connec-
tion rates ŽTable 2.. The validity of this correlation tion between reservoir channel sand bodies Že.g.
however, was questioned by Bryant et al. Ž1995.. Putnam, 1993; Wightman et al., 1987.. Sometimes,
They found experimental evidence for a rise in avul- connectedness is rather inferred from the observation
sion frequency with sedimentation rate, which de- of more channel sandstone bodies at exactly the
creases the proportion of overbank deposits by in- same stratigraphic level in cross-sections. Nadon
creasing the number of channel sandstone bodies. Ž1994. and Kraus and Wells Ž1999., for example,
Mackey and Bridge Ž1995. and Heller and Paola observed sandstone ‘wings’, attached to the top of
Ž1996. used three-dimensional simulation models to channel bodies, of which the beds could be traced
further explore the complex relationship between from one sandstone body to an adjacent sandstone
floodplain aggradation rate and the proportion of body. Kraus and Wells Ž1999., however, also had
overbank deposits preserved. Long-term coexistence planform data suggesting connectedness.
 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196 185

It is hard to prove whether several connected ees laterally connected to channel sandstone bodies,
sandstone bodies represent contemporaneously active were misidentified as channel-fill deposits, leading to
channels. In the case of organic-rich Holocene to overestimation of the width of the channel body; Ž2.
Late Pleistocene deposits, 14 C-dating of the period of the widthrdepth ratio of the palaeochannels was
activity of the individual branches may provide con- greater than those of modern anastomosing channels;
¨
clusive evidence Že.g. Tornqvist, 1993b, 1994.. or Ž3. the sandstone ribbons were partly formed by
However, in most cases, coexistence of channels can lateral accretion.
only be assumed on the basis of stratigraphic argu- The first explanation particularly applies to situa-
ments Že.g. Currie et al., 1991; Bakker et al., 1989; tions where reconstructions are based on geophysical
Flores and Hanley, 1984.. or limited borehole data. In these cases, wide spacing
of logs may also lead to apparent lateral amalgama-
4.4.3. Widthr thickness ratio of channel sandstone tion of separate channel sandstone bodies Že.g.
bodies Wightman et al., 1987.. The second and the third
The importance of the shape of channel deposits explanation may in fact both be applicable at the
as an indicator of former lateral channel stability was same time, since actively meandering channels tend
stressed by Collinson Ž1978.. Friend et al. Ž1979. to have higher widthrdepth ratios than laterally sta-
and Friend Ž1983. used the widthrthickness ratio of ble channels. Page and Nanson Ž1996. reported later-
channel-fill bodies to infer former channel be- ally extensive sandstone sheets wwrt up to 1000
haviour. Bodies with widthrthickness ratios - 15 ŽTable 3.x forming the base of Late Quaternary
were called ribbons and thought to reflect fixed channel-fill sequences, which are believed to repre-
channels subject to switching by avulsion. Bodies sent, at least partly, an anastomosing system. The
with widthrthickness ratios ) 15 were called sheets sheets are relatively thin, since the upper parts of the
and were viewed as a product of lateral migration of channel-fills consist of mud. The width of the sheets
a channel. Nadon Ž1994. proposed a widthrthick- is obviously due to lateral channel migration, as
ness ratio of 30 as an upper limit of ribbons. Al- testified by meander scars and scroll-patterned flood-
though both the values of 15 and 30 are arbitrary, the plains visible on air photographs. Lateral accretion
concept of reconstructing channel behaviour from units were also recognized in some of the ancient
the geometry of the channel-fill bodies is useful. channel sandstone bodies described by Rust and
Naturally, it first should be ascertained that the sand Legun Ž1983., Rust et al. Ž1984., Nadon Ž1994. and
body really is the product of channelized flow Že.g. Kirschbaum and McCabe Ž1992.. The on average
by longitudinal dimension.. Sandstone sheets may, relatively great widthrthickness ratio of ancient
for example, also represent crevasse splays deposited anastomosing channel sandstone bodies ŽTable 3.
by sheetfloods. renders it unlikely that the ancient channels were
The widthrthickness ratio of a channel-fill body generally keeping pace with strong vertical flood-
is determined by: Ž1. the widthrdepth ratio of the plain aggradation Žsecond picture of Fig. 16., as was
original channel; Ž2. a lateral migration component; suggested for modern anastomosing channels by
and Ž3. a vertical aggradation component ŽFig. 16.. Smith and Smith Ž1980. Žsee also Fig. 11.. In that
From Table 3, it appears that the widthrthickness case, one would expect much lower widthrthickness
ratios of the ancient and subrecent channel sandstone ratios for most of the examples in Table 3.
bodies vary widely and that many of them are not
ribbons in the terminology of Friend Ž1983. Ži.e.
wrt - 15.. Comparison with the widthrdepth ratios 4.5. EÕaluation
of modern anastomosing channels ŽTable 2. shows
that on average the widthrthickness ratio of the Only a limited number of ancient stratigraphic
channel sandstone bodies is significantly greater than examples has been interpreted as anastomosing in
the widthrdepth ratios of modern anastomosing origin. Since anastomosis typically occurs in the
channels. There are three possible explanations: Ž1. lower reaches of rivers in aggrading basinal settings
sandstone sheets, representing crevasse splays or lev- where preservation potential tends to be high, one
186 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196

Fig. 16. Cross-sections showing channel geometry, lateral migration and vertical aggradation determining the widthrthickness ratio of a
sandstone body.

would expect ancient anastomosing river systems to sedimentological information about other kinds of
be relatively abundant in the stratigraphic record. I modern anastomosing rivers becomes available, addi-
feel that probably many examples were not identified tional case studies are still urgently needed for com-
as ancient anastomosing river systems, mainly be- parison with the stratigraphic record.
cause of: Ž1. the restricted sedimentological database Anastomosing river systems are relatively large-
from modern anastomosing environments; Ž2. the scale geomorphological phenomena. Applying the
relatively large scale of anastomosing rivers as a term ‘anastomosing’ to a certain facies association in
floodplain-wide fluvial form; and Ž3. the difficulty of an ancient rock unit implies that the palaeogeomor-
proving coexistence of multiple channels in ancient phology of the river system can be inferred. Prob-
examples. lems arising from the application of geomorphologi-
It must be realized that most of what we know cal classifications to ancient fluvial sediments were
about the sedimentary facies of anastomosing rivers discussed Žamong others. by Friend Ž1983. and Miall
is based on a limited number of facies descriptions Ž1985.. As exposure is limited and especially plan-
of modern anastomosing rivers. These modern rivers form information on ancient river systems is gener-
represent a dataset that is still strongly biased to- ally scarce, conclusions on palaeoriver morphology
wards small to medium-scale anastomosing rivers in usually can only be drawn at the level of individual
humid climatic settings. Although gradually more river branches.
 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196 187

The problem of limited chronological control of three ‘fluvial associations’ in which river deposits
ancient systems, and therefore the inability of prov- occur: Ž1. the alluvial fan association, Ž2. the flood-
ing coexisting of channels, will remain in the near plain association and Ž3. the coastal plain–delta as-
future. Presently, late Quaternary systems, within the sociation. Their coastal plain–delta association in
range of 14 C-dating Ž50,000 years., offer by far the fact represents the anastomosing river environment.
best opportunities to prove former anastomosing sys- An important step forward would be to implement
tems, and to study their long-term development and anastomosis Žand systematic variations in the number
sedimentary sequences, within a detailed timeframe of coexisting channels. in simulation models for
Že.g. Makaske, 1998; Morozova and Smith, 1999.. fluvial architecture Že.g. Mackey and Bridge, 1995.
Since preserved floodplain organics are required, and sequence stratigraphic models for nonmarine
humid fluvial systems are much more suitable than strata Že.g. Wright and Mariott, 1993; Shanley and
arid systems. In the latter environment, thermolumi- McCabe, 1993..
nescence dating ŽTL. of fluvial sediments may pro- At present, there is no consensus on the question
vide an alternative Že.g. Page and Nanson, 1996; whether the anastomosing fluvial environment in-
Page et al., 1996.. An advantage is that TL has the cludes the fluviodeltaic environment. There is a ten-
ability to date to 300,000 years; however, the level dency to consider the anastomosing river environ-
of precision of TL dates is generally too low to ment as a continental counterpart of the fluviodeltaic
prove coexistence of palaeochannels. environment, but some workers apply the term anas-
Besides these identification issues, another impor- tomosing to subrecent and modern channels in deltaic
tant problem is that the development of an anasto- environments ŽPeng, 1990; King and Martini, 1984.,
mosing rivers facies model was hindered by the lack especially in the Rhine–Meuse delta in the Nether-
of a sound conceptual framework with regard to the ¨
lands ŽTornqvist, 1993a; Weerts and Bierkens, 1993;
definition and classification of anastomosing rivers. Bosch and Kok, 1994; Weerts, 1996; De Groot and
In current definitions, anastomosis Žin the sense of De Gans, 1996; Makaske, 1998.. In my view, coexis-
multiple channels. and lateral stability of channels tence of channels in deltaic environments is not
are often inseparably coupled. As shown earlier in fundamentally different in processes and products
this paper, these two properties are driven by differ- from anastomosis in continental settings. In both
ent process complexes Že.g. Fig. 9.. As a result, not situations, multiple channels result from avulsions,
all laterally stable channels are parts of multi-chan- with the only difference that avulsions in near-coastal
nel systems Že.g. Nanson and Croke, 1992, p. 477., settings may be caused or triggered by marine pro-
and on the other hand, not every anastomosing, cesses Že.g. silting up of river mouths by longshore
multi-channel system needs to consist of laterally currents, river flooding due to set up of sea level
stable channels Žsee Section 2.3.. However, in com- caused by storm andror spring tide.. Although flu-
prehensive studies on fluvial styles and facies mod- viodeltaic lithofacies may differ in detail from conti-
els Že.g. Miall, 1985, 1992., the single-channel nental anastomosing river lithofacies we.g. muddrapes
straight river has been neglected. As a result, ribbon in fluviodeltaic channel deposits Žincluding IHS.
sandstones are often automatically taken to represent recording tidal or marine influence ŽSmith, 1987;
anastomosing rivers. Thomas et al. 1987.x, studies of Holocene flu-
A ‘straight river facies model’ needs to be defined viodeltaic sequences demonstrate that alluvial archi-
as a logical extension of the suite of braided and tecture can be very similar to that of inland anasto-
meandering river facies models in the low-energy ¨
mosing counterparts Že.g. Tornqvist et al., 1993;
domain. These models, although including overbank Makaske, 1998..
deposits, are mainly based on the lithofacies of Application of concepts derived from anastomos-
channel sediments and can be applied at the channel ing rivers to fluviodeltaic environments may lead to
scale. In addition, the anastomosing river model refinement of sequence stratigraphic models for non-
should be considered as a model for a large-scale marine strata Že.g. Wright and Mariott, 1993; Shan-
fluvial depositional system. Reineck and Singh Ž1980, ley and McCabe, 1993.. In sequence stratigraphic
pp. 305–306. for example, distinguish in this respect studies, sand body widthrthickness ratio is often
188 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196

interpreted to be inversely related to floodplain sedi- more rapidly subsiding settings, for example foreland
mentation rate Že.g. Shanley and McCabe, 1991; basins, at least TST-deposits have a much higher
Mjøs and Prestholm, 1993; Aitken and Flint, 1995.. chance of preservation.
A similar relationship was found by Tornqvist ¨ Although sequence stratigraphy primarily focuses
Ž1993a,b. for the Holocene sequence of the fluvial on the effect of eustatic sea-level changes on near-
Rhine–Meuse delta. Floodplain sedimentation rates coastal depositional systems, important systematic
in near-coastal fluvial settings are principally driven changes in base level Žrelated to climatic cycles. may
by relative sea-level rise. A rapid rise in base level also occur in intracratonic nonmarine settings. Lake-
Žsea level. causes a prolonged reduction of channel levels in semi-arid or arid environments rapidly rise
gradients, favouring low-energy, laterally stable and fall in response to arid-humid climatic cycles
channels that will be preserved as sand bodies with a and lake-feeding fluvial systems may respond by
low widthrthickness ratio. It is generally accepted changing morphology. In the Inland Delta of the
that fluvial systems respond to changes in external Niger River in central Mali, late Quaternary arid
conditions Žincluding base-level changes. by changes periods resulted in degradation of the channel net-
in channel morphology Že.g. Schumm, 1993.. How- work, by in-channel aeolian sedimentation and dis-
ever, evidence from the Rhine–Meuse delta suggests section of the levees by gullies. During transition
that near-coastal fluvial systems may also respond to from arid to humid the fluvial network could not
base-level rise by changing the number of coexistent accommodate the rising discharge that resulted in
¨
channels ŽTornqvist, 1994; Makaske, 1998, pp. 226– avulsions and coexistence of multiple channels. At
229.. Starting from a positive relationship between the same time, local base level rose rapidly due to
base-level rise, aggradation rate and avulsion fre- formation of large lakes downstream, partly because
quency, it is reasonable to suppose that rapid sea-level large cross-valley longitudinal dunes, erected during
rise makes the number of coexistent delta distribu- the preceding arid period, obstructed river flow ŽTri-
taries increase by frequent avulsion ŽFig. 9.. Since cart, 1959, 1965; Gallais, 1967; Makaske, 1998.. In
the same amount of water has to be distributed over such arid intracratonic settings, sedimentation rates
an increasing number of channels, stream power for tend to be slow, while climatic changes are relatively
individual channels will inevitably decrease, result- frequent Žor at least were during the Late Quater-
ing in an increase in lateral channel stability and nary.. This inhibits thick sequences of anastomosing
eventually a decrease in sand body widthrthickness river deposits formed under constant climatic condi-
ratio. tions. Nevertheless, anastomosing river deposits are
In sequence stratigraphic terms, it can be stated likely to make up an important part of the cyclic
that multiple coexistent, laterally stable channels may basin-fill.
typify transgressive systems tracts ŽTSTs.; i.e. near-
shore depositional systems associated with rapidly
rising sea level. Some authors, however, argue that
initially rapid sea-level rise does not create much 5. Conclusions
accommodation space in upper nonmarine parts of
deltas ŽShanley and McCabe, 1993.. Only when the In presently popular definitions of anastomosing
coastline has shifted far inland, relative sea-level rise rivers, lateral stability of channels is presumed along
is ‘felt’ in this area. This may be near sea-level with their multi-channel character. However, the
highstand, rather than during rapid transgression. In coupling of channel-scale properties and multiple-
this case, multiple coexistent channels can be ex- channel character in the same definition seems con-
pected to typify the early highstand systems tract ceptually wrong. Besides, it is unpractical in sedi-
ŽHST. instead of the TST. Another matter is the mentological applications. Therefore, I propose the
preservation potential of HST and TST deposits. If following redefinitions: Ž1. at the channel-belt scale,
basin subsidence is slow, for example along passive the term ‘straight’ should be used to describe later-
continental margins, a large part of the TST-deposits ally stable channels having a sinuosity index lower
may be eroded during the next sea-level lowstand. In than 1.3; Ž2. at a larger scale, rivers should be called
 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196 189

anastomosing if a riÕer is composed of two or more lived, highly dynamic anastomosing systems, con-
interconnected channels that enclose floodbasins. In stantly rejuvenated by frequent avulsions caused by
this paper, I started from these definitions. The con- allogenic controls; and Ž4. long-lived, semi-static
clusions reached concerning the genesis of anasto- anastomosing systems, which seem in equilibrium
mosing rivers are listed below. with stable external conditions. Since these four
Ž1. Anastomosing rivers are formed by avulsions, types are morphologically quite similar, anastomos-
i.e., flow diversions that cause the formation of new ing rivers can be considered an example of equifinal-
channels on the floodplain. As a product of avulsion, ity, i.e., different combinations of processes or causes
anastomosing rivers essentially form in two ways: produce a similar form.
Ža. by formation of bypasses, while bypassed older Since anastomosis typically occurs in the lower
channel-belt segments remain active for some pe- reaches of rivers in Žrapidly. aggrading basinal set-
riod; and Žb. by splitting of the diverted avulsive tings where preservation potential tends to be high,
flow, leading to contemporaneous scour of multiple ancient anastomosing river systems are expected to
channels on the floodplain. be relatively abundant in the stratigraphic record.
Ž2. Avulsions are primarily driven by aggradation Below, the main characteristics of anastomosing river
of the channel belt andror loss of channel capacity deposits will be given.
by in-channel deposition. Both processes are favoured Ž1. The alluvial architecture is characterized by a
by a low floodplain gradient. Of secondary impor- large proportion of overbank deposits Žprobably 50%
tance are a number of avulsion triggers such as to 90%. which encase laterally connected channel
extreme floods, log and ice jams, and in-channel sandstone bodies.
aeolian dunes. Although some of these triggers are Ž2. Laterally extensive, thick lenses of lithologi-
associated with a specific climate, the occurrence of cally heterogeneous, fine-grained avulsion deposits
anastomosis is not. can be an important element of the overbank de-
Ž3. Long-lived anastomosis is caused by frequent posits of anastomosing rivers. These deposits may
avulsions andror slow abandonment of old channels. also fully surround anastomosing channel sand-
With respect to avulsion frequency, anastomosing stones. Typically, they also encase sheet-like crevasse
rivers range from highly dynamic to semi-static. In splay sandstones.
semi-static systems, anastomosis seems to result Ž3. The geometry of anastomosing channel sand-
mainly from extremely slow infilling of old channels stone bodies is frequently characterized by a ribbon
after infrequent formation of new channels. In more shape and a flat top.
dynamic systems, anastomosis rather results from the Ž4. Channel deposits predominantly consist of
frequent formation of new channels, while old chan- sandstone which may fine upward, but also can be
nels remain active for some time. quite homogeneous. In the latter case, the sandstone
Ž4. A rapid rise of base level is conductive to may pass abruptly into overlying mudrock.
anastomosis, but is not a necessary condition. Some Ž5. Crevasse splay deposits and thick natural levee
extensive anastomosing river systems apparently ex- deposits are common elements.
ist in equilibrium with stable external conditions. Ž6. Lacustrine deposits and coal commonly occur
Ž5. The often-observed high lateral stability of in association with anastomosing river overbank de-
individual channels in anastomosing rivers is caused posits Žin arid settings, lacustrine deposits and coal
by low stream power in combination with resistant may be absent..
banks. Both conditions are strongly related to a low It should be noted that none of these character-
floodplain gradient. istics is unique to anastomosing river deposits. Ap-
With respect to genesis and lifetime, four basic plication of the geomorphological term ‘anastomos-
types of anastomosing rivers can be identified: Ž1. ing’ to ancient fluvial systems requires careful and
short-lived anastomosis within an avulsion belt; Ž2. often large-scale palaeogeographical reconstruction.
short-lived, transitional anastomosis during a single In practice, subrecent anastomosing river environ-
avulsion, due to temporary coexistence of an older ments offer by far the best opportunities to realize
trunk channel and a new avulsion channel; Ž3. long- this. In ancient rocks, geomorphological interpreta-
190 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196

tion of fluvial facies will often be restricted to a Miall, A.D. ŽEd.., Fluvial Sedimentology. Canadian Society of
classification of individual palaeochannels as being Petroleum Geologists Memoir 5, Calgary, pp. 211–230.
Bakker, J.G.M., Kleinendorst, T.W., Geirnaert, W., 1989. Tec-
braided, meandering or straight, and sometimes even tonic and sedimentary history of a late Cenozoic intramontane
a less precise reconstruction will be the best attain- basin ŽThe Pitalito Basin Colombia.. Basin Research 2, 161–
able. In these cases, a purely descriptive geometric 187.
classification of ribbon and sheet sandstone bodies is Berendsen, H.J.A., 1982. De genese van het landschap in het
an alternative. zuiden van de provincie Utrecht, een fysisch-geografische
studie. Utrechtse Geografische Studies 25, Geografisch Insti-
tuut, Rijksuniversiteit Utrecht, Utrecht.
Berendsen, H.J.A., Faessen, E.L.J.H., Kempen, H.F.J., 1994. Zand
in banen, zanddiepte-attentiekaarten van het Gelders riv-
Acknowledgements ierengebied. Provincie Gelderland, Arnhem.
Bosch, J.H.A., Kok, H., 1994. Toelichtingen bij de geologische
kaart van Nederland schaal 1:50.000; blad Gorinchem West
Many individuals contributed to this review by Ž38 W.. Rijks Geologische Dienst, Haarlem.
commenting and taking time for discussion. Thanks Bowler, J.M., Stockton, E., Walker, M.J., 1978. Quaternary
are due to my former colleagues at the Faculty of stratigraphy of the Darling River near Tilpa, New South
Wales. Proceedings of the Royal Society of Victoria 90,
Geographical Sciences of the Utrecht University:
Melbourne, 79–88.
Henk Berendsen, Ward Koster, Torbjorn ¨ Tornqvist,
¨ Brice, J.C., 1964. Channel patterns and terraces of the Loup
Janrik Van den Berg and Henk Weerts. Likewise, Rivers in Nebraska. U.S. Geological Survey Professional Pa-
Derald Smith is sincerely thanked for his construc- per 422-D, Washington.
tive comments on early drafts of this paper. A dis- Bridge, J.S., 1993. The interaction between channel geometry,
water flow, sediment transport and deposition in braided rivers.
cussion with John Bridge on the classification of
In: Best, J.L., Bristow, C.S. ŽEds.., Braided Rivers. Geological
river channel patterns was stimulating. I also wish to Society Special Publication 75, London, pp. 13–71.
thank Gerald Nanson for discussion and supplying Bridge, J.S., Leeder, M.R., 1979. A simulation model of alluvial
extra information about Australian anastomosing stratigraphy. Sedimentology 26, 617–644.
rivers. The thoughtful comments of the reviewers Bridge, J.S., Mackey, S.D., 1993. Revised alluvial stratigraphy
Norman Smith and Andrew Miall enabled me to ´
model. In: Marzo, M., Puigdefabregas, C. ŽEds.., Alluvial
Sedimentation. Special Publication of the International Associ-
considerably improve the manuscript. ation of Sedimentologists 17, Blackwell, Oxford, pp. 319–336.
Bristow, C.S., 1999. Gradual avulsion, river metamorphosis and
reworking by underfit streams: a modern example from the
Brahmaputra River in Bangladesh and a possible ancient
References example in the Spanish Pyrenees. In: Smith, N.D., Rogers, J.
ŽEds.., Fluvial Sedimentology VI. Special Publication of the
International Association of Sedimentologists 28, Blackwell,
Aitken, J.F., Flint, S.S., 1995. The application of high-resolution Oxford, pp. 221–230.
sequence stratigraphy to fluvial systems: a case study from the Brizga, S.O., Finlayson, B.L., 1990. Channel avulsion and river
Upper Carboniferous Breathitt Group, eastern Kentucky, USA. metamorphosis: the case of the Thomson river, Victoria, Aus-
Sedimentology 42, 3–30. tralia. Earth Surface Processes and Landforms 15, 391–404.
Alexander, J., Leeder, M.R., 1987. Active tectonic control on Brown, A.G., 1987. Holocene floodplain sedimentation and chan-
alluvial architecture. In: Ethridge, F.G. ŽEd.., Recent Develop- nel response of the lower River Severn. United Kingdom.
ments in Fluvial Sedimentology. Society of Economic Paleon- ¨ Geomorphologie 32 Ž3., 293–310.
Zeitschrift fur
tologists and Mineralogists Special Publication 39, Tulsa, pp. Bryant, M., Falk, P., Paola, C., 1995. Experimental study of
243–252. avulsion frequency and rate of deposition. Geology 23 Ž4.,
Allen, J.R.L., 1965. A review of the origin and characteristics of 365–368.
recent alluvial sediments. Sedimentology 5, 89–191. Burnett, A.W., Schumm, S.A., 1983. Alluvial river response to
Aslan, A., Blum, M.D., 1999. Contrasting styles of Holocene neotectonic deformation in Louisiana and Mississippi. Science
avulsion, Texas Gulf Coastal Plain, USA. In: Smith, N.D., 222, 49–50.
Rogers, J. ŽEds.., Fluvial Sedimentology VI. Special Publica- Cairncross, B., Stanistreet, I.G., McCarthy, T.S., Ellery, W.N.,
tion of the International Association of Sedimentologists 28, Ellery, K., Grobicki, T.S.A., 1988. Palaeochannels Žstone-rolls.
Blackwell, Oxford, pp. 193–209. in coal seams: modern analogues from fluvial deposits of the
Baker, V.R., 1978. Adjustment of fluvial systems to climate and Okavango Delta, Botswana, southern Africa. Sedimentary Ge-
source terrain in tropical and subtropical environments. In: ology 57, 107–118.
 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196 191

Carson, M.A., 1984a. The meandering–braided river threshold: a Flores, R.M., Hanley, J.H., 1984. Anastomosed and associated
reappraisal. Journal of Hydrology 73, 315–334. coal-bearing fluvial deposits: Upper Tongue River Member,
Carson, M.A., 1984b. Observations on the meandering–braided Palaeocene Fort Union Formation, northern Powder River
river transition, The Canterbury Plains, New Zealand: part Basin, Wyoming, U.S.A. In: Rahmani, R.A., Flores, R.M.
two. New Zealand Geographer 40 Ž1., 89–99. ŽEds.., Sedimentology of Coal and Coal-bearing Sequences.
Chang, H.H., 1979. Minimum stream power and river channel Special Publications of the International Association of Sedi-
patterns. Journal of Hydrology 41, 303–327. mentologists 7, Blackwell, Oxford, pp. 85–103
Church, M., 1983. Pattern of instability in a wandering gravel bed Flores, R.M., Pillmore, C.L., 1987. Tectonic control on alluvial
channel. In: Collinson, J.D., Lewin, J. ŽEds.., Modern and paleoarchitecture of the Cretaceous and Tertiary Raton Basin,
Ancient Fluvial Systems. Special Publication of the Interna- Colorado and New Mexico. In: Ethridge, F.G., Flores, R.M.,
tional Association of Sedimentologists 6, Blackwell, Oxford, Harvey, M.D. ŽEds.., Recent Developments in Fluvial Sedi-
pp. 169–180. mentology. Society of Economic Paleontologists and Mineral-
Coleman, J.M., 1969. Brahmaputra River: channel processes and ogists Special Publication 39, Tulsa, 311–320.
sedimentation. Sedimentary Geology 3, 129–239. Friend, P.F., 1983. Towards the field classification of alluvial
Collinson, J.D., 1978. Vertical sequence and sand body shape in architecture or sequence. In: Collinson, J.D., Lewin, J. ŽEds..,
alluvial sequences. In: Miall, A.D. ŽEd.., Fluvial Sedimentol- Modern and Ancient Fluvial Systems. Special Publication of
ogy. Canadian Society of Petroleum Geologists Memoir 5, the International Association of Sedimentologists 6, Black-
Calgary, pp. 577–586. well, Oxford, pp. 345–354.
Currie, P.J., Nadon, G.C., Lockley, M.G., 1991. Dinosaur foot- Friend, P.F., Sinha, R., 1993. Braiding and meandering parame-
prints with skin impressions from the Cretaceous of Alberta ters. In: Best, J.L., Bristow, C.S. ŽEds.., Braided Rivers.
and Colorado. Canadian Journal of Earth Sciences 28, 102– Geological Society Special Publication 75, London, pp. 105–
115. 111.
De Groot, Th.A.M., De Gans, W., 1996. Facies variations and Friend, P.F., Slater, M.J., Williams, R.C., 1979. Vertical and
sea-level-rise response in the lowermost RhinerMeuse area Lateral Building of River Sandstone Bodies 136. Journal of
during the last 15000 years Žthe Netherlands.. In: Beets, D.J., the Geological Society of London, Ebro Basin, Spain, pp.
Fischer, M.M., De Gans, W. ŽEds.., Coastal Studies on the 39–46.
Holocene of the Netherlands. Mededelingen Rijks Geologische ´
Gallais, J., 1967. Le Delta interieur du Niger et ses bordures;
Dienst 57, Haarlem, pp. 229–250. ´
etude ´
morphologique. Memoires ´
et documents, nouvelle serie
Dumont, J.F., 1994. Neotectonics of the Amazon headwaters. In: 3, Centre National de la Recherche Scientifique, Paris.
Schumm, S.A., Winkley, B.R. ŽEds.., The variability of large Galloway, W.E., Hobday, D.K., 1983. Terrigenous Clastic Depo-
alluvial rivers. American Society of Civil Engineers, New sitional Systems; Applications to Petroleum, Coal and Ura-
York, pp. 103–113. nium exploration. Springer, New York.
Eberth, D.A., Miall, A.D., 1991. Stratigraphy, sedimentology and Garner, H.F., 1959. Interpretation of globally distributed anasto-
evolution of a vertebrate-bearing, braided to anastomosing mosing channel drainages. Geological Society of America
fluvial system, Cutler Formation ŽPermian-Pennsylvanian., Bulletin 70 Ž12., 1607.
north-central New Mexico. Sedimentary Geology 72, 225–252. Garner, H.F., 1967. Rivers in the making. Scientific American
Elliot, T., 1974. Interdistributary bay sequences and their genesis. 216, 84–94.
Sedimentology 21, 611–622. Gibling, M.R., Rust, B.R., 1990. Ribbon sandstones in the Penn-
Ethridge, F.G., Skelly, R.L., Bristow, C.S., 1999. Avulsion and sylvanian Waddens Cove Formation, Sydney Basin, Atlantic
crevassing in the sandy, braided Niobrara River: complex Canada: the influence of siliceous duricrusts on channel-body
response to base-level rise and aggradation. In: Smith, N.D., geometry. Sedimentology 37, 45–65.
Rogers, J., Fluvial Sedimentology VI. Special Publication of Gibling, M.R., Nanson, G.R., Maroulis, J.C., 1998. Anastomosing
the International Association of Sedimentologists 28, Black- river sedimentation in the Channel Country of central Aus-
well, Oxford, pp. 179–191. tralia. Sedimentology 45, 595–619.
Ferguson, R.I., 1981. Channel form and channel changes. In: Gohain, K., Parkash, G., 1990. Morphology of the Kosi megafan.
Lewin, J. ŽEd.., British Rivers. Allen and Unwin, London, pp. In: Rachocki, A.H., Church, M. ŽEds.., Alluvial Fans: a Field
90–211. Approach. Wiley, Chichester, pp. 151–178.
Ferguson, R.I., 1987. Hydraulic and sedimentary controls of chan- Gole, C.V., Chitale, S.V., 1966. Inland delta building activity of
nel pattern. In: Richards, K.S. ŽEd.., River Channels; Environ- Kosi River. Journal of the Hydraulics Division, Proceedings of
ment and Process. Institute of British Geographers Special the American Society of Civil Engineers 92, 111–126.
Publication 18, Blackwell, Oxford, pp. 129–155. Guccione, M.J., Burford, M.F., Kendall, J.D. ŽEds.., Pemiscot
Fisk, H.N., 1944. Geological investigation of the alluvial valley of Bayou, a large distributary of the Mississippi River and a
the lower Mississippi. Mississippi River Commission, U.S. possible failed avulsion. In: Smith, N.D., Rogers, J. ŽEds..,
Army Corps of Engineers, Vicksburg. Fluvial Sedimentology VI. Special Publication of the Interna-
Fisk, H.N., 1947. Fine-grained alluvial deposits and their effects tional Association of Sedimentologists 28, Blackwell, Oxford,
on Mississippi River activity. Mississippi River Commission, pp. 211–220.
U.S. Army Corps of Engineers, Vicksburg. Harwood, K., Brown, A.G., 1993. Fluvial processes in a forested
192 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196

anastomosing river: flood partitioning and changing flow pat- Knighton, A.D., Nanson, G.C., 1993. Anastomosis and the contin-
terns. Earth Surface Processes and Landforms 18, 741–748. uum of channel pattern. Earth Surface Processes and Land-
Heller, P.L., Paola, C., 1996. Downstream changes in alluvial forms 18, 613–625.
architecture: an exploration of controls on channel-stacking Kolb, C.R., 1963. Sediments forming the bed and banks of the
patterns. Journal of Sedimentary Research 66 Ž2., 297–306. lower Mississippi River. Sedimentology 2, 227–234.
Hickin, E.J., 1979. Concave-bank benches on the Squamish River, Kraus, M.J., 1996. Avulsion deposits in lower Eocene alluvial
British Columbia. Canadian Journal of Earth Sciences 16, rocks, Bighorn Basin, Wyoming. Journal of Sedimentary Re-
200–203. search 66 Ž2., 354–363.
Hickin, E.J., 1984. Vegetation and river channel dynamics. Cana- Kraus, M.J., Aslan, A., 1993. Eocene hydromorphic paleosols:
dian Geographer 28 Ž2., 111–126. significance for interpreting ancient floodplain processes. Jour-
Hickin, E.J., 1986. Concave-bank benches in the floodplains of nal of Sedimentary Petrology 63, 453–463.
Muskwa and Fort Nelson rivers, British Columbia. The Cana- Kraus, M.J., Gwinn, B., 1997. Facies and facies architecture of
dian Geographer 30 Ž2., 111–122. Paleogene floodplain deposits, Willwood Formation, Bighorn
Hickin, E.J., 1993. Fluvial facies models: a review of Canadian Basin, Wyoming, USA. Sedimentary Geology 114, 33–54.
research. Progress in Physical Geography 17 Ž2., 205–222. Kraus, M.J., Wells, T.M., 1999. Recognizing avulsion deposits in
Huang, H.Q., Nanson, G.C., 1997. Vegetation and channel varia- the ancient stratigraphical record. In: Smith, N.D., Rogers, J.
tion; a case study of four small streams in southeastern ŽEds.., Fluvial Sedimentology VI. Special Publication of the
Australia. Geomorphology 18, 237–249. International Association of Sedimentologists 28, Blackwell,
Jacobberger, P.A., 1987. Geomorphology of the upper Inland Oxford, pp. 251–268.
Niger Delta. Journal of Arid Environments 13, 95–112. Kruit, C., 1995. Sediments of the Rhone ˆ Delta; grain size and
Jacobberger, P.A., 1988a. Drought-related changes to geomorpho- microfauna. PhD thesis, Rijksuniversiteit Groningen, Gronin-
logic processes in central Mali. Geological Society of America gen.
Bulletin 100, 351–361. Kumar, R., Tandon, S.K., 1985. Sedimentology of Plio-Pleisto-
Jacobberger, P.A., 1988b. Mapping abandoned river channels in cene late orogenic deposits associated with intraplate subduc-
Mali through directional filtering of Thematic Mapper data. tion, the Upper Siwalik Subgroup of a part of Panjab sub-
Remote Sensing of Environment 26, 161–170. Himalaya. Sedimentary Geology 42, 105–158.
Johnson, E.A., Pierce, F.W., 1990. Variations in fluvial deposition Le Blanc Smith, G., Eriksson, K.A., 1979. A fluvioglacial and
on an alluvial plain: an example from the Tongue River glaciolacustrine deltaic depositional model for Permo-
Member of the Fort Union Formation ŽPaleocene., southeast- Carboniferous coals of the northeastern Karoo basin, South
ern Powder River Basin, Wyoming, U.S.A. Sedimentary Geol- Africa. Palaeogeography, Palaeoclimatology, Palaeoecology
ogy 69, 21–36. 27, 67–84.
Jones, L.S., Schumm, S.A., 1999. Causes of avulsion: an overview. Lecce, S.A., 1997. Spatial patterns of historical overbank sedi-
In: Smith, N.D., Rogers, J. ŽEds.., Fluvial Sedimentology VI. mentation and floodplain evolution, Blue River, Wisconsin.
Special Publication of the International Association of Sedi- Geomorphology 18, 265–277.
mentologists 28, Blackwell, Oxford, pp. 171–178. Leopold, L.B., Wolman, M.G., 1957. River channel patterns:
Keller, E.A., Brookes, A., 1984. Consideration of meandering in braided, meandering and straight. U.S. Geological Survey
channelization projects: selected observations and judgements. Professional Paper 282-B, Washington.
In: Elliot, C.M. ŽEds.., River Meandering. Proceedings of the Locking, T., Hydrology and sediment transport in an anastomos-
conference Rivers ’83, New Orleans, Louisiana, October 24– ing reach of the upper Colombia River, B.C. MSc thesis,
26, 1983. American Society of Civil Engineers, New York, Department of Geography, University of Calgary, Calgary.
pp. 384–397. Lundegard, P.D., Samuels, N.D., 1980. Field classification of
King, W.A., Martini, I.P., 1984. Morphology and recent sediments fine-grained sedimentary rocks. Journal of Sedimentary Petrol-
of the lower anastomosing reaches of the Attawapiskat River, ogy 50, 781–786.
James Bay, Ontario, Canada. Sedimentary Geology 37, 295– Mack, G.H., Leeder, M.R., 1998. Channel shifting of the Rio
320. Grande, southern Rio Grande rift: implications for alluvial
Kirschbaum, M.A., McCabe, P.J., 1992. Controls on the accumu- stratigraphic models. Sedimentary Geology 117, 207–219.
lation of coal and on the development of anastomosed fluvial Mackey, S.D., Bridge, J.S., 1995. Three-dimensional model of
systems in the Cretaceous Dakota Formation of southern Utah. alluvial stratigraphy: theory and application. Journal of Sedi-
Sedimentology 39, 581–598. mentary Research B 65 Ž1., 7–31.
Knight, M.J., 1975. Recent Crevassing of the Erap River, Papua Mackin, J.H., 1948. Concept of a graded river. Geological Society
New Guinea. Australian Geographical Studies 13, pp. 77–84. of America Bulletin 59, 463–511.
Knighton, A.D., 1984. Fluvial forms and processes. Arnold, Lon- Makaske, B., 1998. Anastomosing rivers; forms, processes and
don. sediments. Nederlandse Geografische Studies vol. 249. Konin-
Knighton, A.D., 1987. River channel adjustment; the downstream klijk Nederlands Aardrijkskundig GenootschaprFaculteit
dimension. In: Richards, K.S. ŽEd.., River Channels; Environ- Ruimtelijke Wetenschappen, Universiteit Utrecht, Utrecht.
ment and Process. Institute of British Geographers Special McCabe, P.J., 1984. Depositional environments of coal and coal-
Publication 18, Blackwell, Oxford, pp. 98–128. bearing strata. In: Rahmani, R.A., Flores, R.M. ŽEds.., Sedi-
 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196 193

mentology of Coal and Coal-bearing Sequences. Special Publi- efficiency leading to fluvial diversity. In: Miller, A., Gupta, A.
cations of the International Association of Sedimentologists 7, ŽEds.., Varieties of Fluvial Form. Wiley, New York, pp.
Blackwell, Oxford, pp. 13–42. 477–494.
McCarthy, T.S., 1993. The great inland deltas of Africa. Journal Nanson, G.C., Knighton, A.D., 1996. Anabranching rivers: their
of African Earth Sciences 17, 275–291. cause, character and classification. Earth Surface Processes
McCarthy, T.S., Stanistreet, I.G., Cairncross, B., 1991. The sedi- and Landforms 21, 217–239.
mentary dynamics of active fluvial channels on the Okavango Nanson, G.C., Rust, B.R., Taylor, G., 1986. Coexistent mud
fan, Botswana. Sedimentology 38, 471–487. braids and anastomosing channels in an arid-zone river: Cooper
McCarthy, T.S., Ellery, W.N., Stanistreet, J.G., 1992. Avulsion Creek, Central Australia. Geology 14, 175–178.
mechanisms on the Okavango fan, Botswana: the control of a Nanson, G.C., East, T.J., Roberts, R.G., 1993. Quaternary stratig-
fluvial system by vegetation. Sedimentology 39, 779–795. raphy, geochronology and evolution of the Magela Creek
McIntosh, R.J., 1983. Floodplain geomorphology and human oc- catchment in the monsoon tropics of northern Australia. Sedi-
cupation of the upper Inland Delta of the Niger. Geographical mentary Geology 83, 277–302.
Journal 149, 182–201. ¨¨
Neller, R.J., Salo, J.S., Rasanen, M.E., 1992. On the formation of
Mertes, L.A.K., 1994. Rates of flood-plain sedimentation on the blocked valley lakes by channel avulsion in upper Amazon
central Amazon River. Geology 22, 171–174. foreland basins. Zeitschrift fur¨ Geomorphologie 36 Ž4., 401–
Miall, A.D., 1985. Architectural-element analysis: a new method 411.
of facies analysis applied to fluvial deposits. Earth-Science Nordseth, K., 1973. Floodplain construction on a braided river;
Reviews 22, 261–308. the islands of Koppangoyene¨ on the river Glomma. Norsk
Miall, A.D., 1992. Alluvial deposits. In: Walker, R.G., James, geografisk Tidsskrift 27, 109–126.
N.P. ŽEds.., Facies Models; Response to Sea Level Change. ORSTOM, 1970. Monographie hydrologique du bassin du Niger;
Geological Association of Canada, St. John’s. pp. 119–142. ´
partie 1: le Niger superieur et le Bani. Office de la Recherche
Miall, A.D., 1996. The geology of fluvial deposits; sedimentary Scientifique et Technique Outre-Mer, Paris.
facies, basin analysis and petroleum geology. Springer, Berlin. Ouchi, S., 1985. Response of alluvial rivers to slow active tectonic
Mike, K., 1975. Utilization of the analysis of ancient river beds movement. Geological Society of America Bulletin 96, 504–
for the detection of Holocene crustal movements. Tectono- 515.
physics 29, 359–368. Page, K.J., Nanson, G.C., 1982. Concave-bank benches and asso-
Miller, J.R., 1991. Development of anastomosing channels in ciated floodplain formation. Earth Surface Processes and
south-central Indiana. Geomorphology 4, 221–229. Landforms 7, 529–543.
Mjøs, R., Prestholm, E., 1993. The geometry and organization of Page, K.J., Nanson, G.C., 1996. Stratigraphic architecture result-
fluviodeltaic channel sandstones in the Jurassic Saltwick For- ing from Late Quaternary evolution of the Riverine Plain,
mation, Yorkshire, England. Sedimentology 40, 919–935. south-eastern Australia. Sedimentology 43, 927–945.
Mollard, J.D., 1973. Airphoto interpretation of fluvial features. In: Page, K., Nanson, G.C., Price, D., 1996. Chronology of Mur-
Fluvial Processes and Sedimentation. Proceedings of hydrol- rumbidgee River palaeochannels on the Riverine Plain, south-
ogy symposium held at the University of Alberta, Edmonton, eastern Australia. Journal of Quaternary Science 11 Ž4., 311–
May 8–9, 1973. National Research Council of Canada, Ot- 326.
tawa, pp. 341–380. Peng, S., 1990. Fundamental characteristics of the anastomosing
Morozova, G.S., Smith, N.D., 1999. Holocene avulsion history of fluvial system on complex deltaic plain. Chinese Science
the lower Saskatchewan fluvial system, Cumberland Marshes, Bulletin 35 Ž10., 835–839.
Saskatchewan-Manitoba, Canada. In: Smith, N.D., Rogers, J. ´
Perez-Arlucea, M., Smith, N.D., 1999. Depositional patterns fol-
ŽEds.., Fluvial Sedimentology VI. Special Publication of the lowing the 1870s avulsion of the Saskatchewan River ŽCum-
International Association of Sedimentologists 28, Blackwell, berland Marshes, Saskatchewan, Canada.. Journal of Sedimen-
Oxford, pp. 231–249. tary Research 69 Ž1., 62–73.
Nadon, G.C., 1988. Tectonic controls on sedimentation within a Putnam, P.E., 1982. Aspects of petroleum geology of the Lloyd-
foreland basin: the Bearpaw, Blood Reserve and St. Mary minster heavy oil fields, Alberta and Saskatchewan. Bulletin
River Formations, southwestern Alberta. Canadian Society of of Canadian Petroleum Geology 30 Ž2., 81–111.
Petroleum Geologists’ field guide to Sequences, Stratigraphy, Putnam, P.E., 1983. Fluvial deposits and hydrocarbon accumula-
Sedimentology: Surface and Subsurface Technical Meeting, tions: examples from the Lloydminster area, Canada. In:
September 14–16, Alberta. Calgary. Canadian Society of Collinson, J.D., Lewin, J. ŽEds.., Modern and Ancient Fluvial
Petroleum Geologists, Calgary. Systems. Special Publication of the International Association
Nadon, G.C., 1994. The genesis and recognition of anastomosed of Sedimentologists 6, Blackwell, Oxford, pp. 517–532.
fluvial deposits: data from the St. Mary River Formation, Putnam, P.E., 1993. A multidisciplinary analysis of Belly River-
southwestern Alberta, Canada. Journal of Sedimentary Re- Brazeau ŽCampanian. fluvial channel reservoirs in west-central
search B 64 Ž4., 451–463. Alberta, Canada. Bulletin of Canadian Petroleum Geology 41
Nanson, G.C., Croke, J.C., 1992. A genetic classification of Ž2., 186–217.
floodplains. Geomorphology 4, 459–486. Putnam, P.E., Oliver, T.A., 1980. Stratigraphic traps in channel
Nanson, G.C., Huang, H.Q., 1999. Anabranching rivers: divided sandstones in the Upper Mannville ŽAlbian. of east-central
194 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196

Alberta. Bulletin of Canadian Petroleum Geology 28 Ž4., Schumm, S.A., 1985. Patterns of alluvial rivers. Annual Review
489–508. of Earth and Planetary Sciences 13, 5–27.
Qian, N., 1990. Fluvial processes in the lower Yellow River after Schumm, S.A., 1988. Variability of the fluvial system in space
levee breaching at Tongwaxiang in 1855. International Journal and time. In: Rosswall, T., Woodmansee, R.G., Risser, P.G.
of Sediment Research 5 Ž2., 1–13. ŽEds.., Scales and Global Change. Wiley, New York, pp.
Reading, H.G., 1996. Sedimentary Environments: Processes, Fa- 225–250.
cies and Stratigraphy. Blackwell, Oxford. Schumm, S.A., 1993. River response to baselevel change: implica-
Reineck, H.-E., Singh, I.B., 1980. Depositional Sedimentary Envi- tions for sequence stratigraphy. The Journal of Geology 101,
ronments. Springer, Berlin. 279–294.
Richards, K.S., 1982. Rivers; Form and Process in Alluvial Chan- Schumm, S.A., Khan, H.R., 1972. Experimental study of channel
nels. Methuen, London. patterns. Geological Society of America Bulletin 83, 1755–
Riley, S.J., 1975. Some differences between distributing and 1770.
braiding channels. Journal of Hydrology ŽNew Zealand. 14 Schumm, S.A., Erskine, W.D., Tilleard, J.W., 1996. Morphology,
Ž1., pp. 1–8. hydrology, and evolution of the anastomosing Ovens and King
Riley, S.J., Taylor, G., 1978. The geomorphology of the Upper Rivers, Victoria, Australia. Geological Society of America
Darling River System with special reference to the present Bulletin 108, 1212–1224.
fluvial system. Proceedings of the Royal Society of Victoria Shanley, K.W., McCabe, P.J., 1991. Predicting facies architecture
90, Melbourne. pp. 89–102. through sequence stratigraphy, an example from the Kaiparow-
Russell, R.J., 1954. Alluvial morphology of Anatolian rivers. its Plateau, Utah. Geology 19, 742–745.
Annals of the Association of American Geographers 44, 363– Shanley, K.W., McCabe, P.J., 1993. Alluvial architecture in a
391. sequence stratigraphic framework: a case history from the
Rust, B.R., 1978. A classification of alluvial channel systems. In: Upper Cretaceous of southern Utah, USA. In: Flint, S.S.,
Miall, A.D. ŽEd.., Fluvial Sedimentology. Canadian Society of Bryant, I.D. ŽEds.., The Geological Modelling of Hydrocarbon
Petroleum Geologists Memoir 5, Calgary, pp. 187–198. Reservoirs and Outcrop Analogues. Special Publication of the
Rust, B.R., 1981. Sedimentation in an arid-zone anastomosing International Association of Sedimentologists 15, Blackwell,
fluvial system: Cooper’s Creek, Central Australia. Journal of Oxford, pp. 21–56.
Sedimentary Petrology 51, 745–755. Slingerland, R., Smith, N.D., 1998. Necessary conditions for a
Rust, B.R., Legun, A.S., 1983. Modern anastomosing-fluvial de- meandering-river avulsion. Geology 26 Ž5., 435–438.
posits in arid Central Australia, and a Carboniferous analogue Smith, D.G., 1973. Aggradation of the Alexandra-North
in New Brunswick, Canada. In: Collinson, J.D., Lewin, J. Saskatchewan River, Banff Park, Alberta. In: Morisawa, M.
ŽEds.., Modern and Ancient Fluvial Systems. Special Publica- ŽEd.., Fluvial Geomorphology. A proceedings volume of the
tion of the International Association of Sedimentologists 6, 4th annual geomorphology symposia series held at Bingham-
Blackwell, Oxford, pp. 385–392 ton, New York, September 27–28, 1973. State University of
Rust, B.R., Gibling, M.R., Legun, A.S., 1984. Coal Deposition in New York, Binghamton, pp. 201–219.
an Anastomosing-fluvial System: the Pennsylvanian Cumber- Smith, D.G., 1976. Effect of vegetation on lateral migration of
land Group south of Joggins, Nova Scotia, Canada. Special anastomosed channels of a glacier meltwater river. Geological
Publication of the International Association of Sedimentolo- Society of America Bulletin 87, 857–860.
gists 7, Blackwell, Oxford, pp. 105–120. Smith, D.G., 1983. Anastomosed fluvial deposits: modern exam-
Rutherfurd, I.D., 1994. Inherited controls on the form of a large, ples from Western Canada. In: Collinson, J., Lewin, J. ŽEds..,
low energy river: the Murray River, Australia. In: Schumm, Modern and Ancient Fluvial Systems. Special Publication of
S.A., Winkley, B.R. ŽEds.., The Variability of Large Alluvial the International Association of Sedimentologists 6, Black-
Rivers. American Society of Civil Engineers, New York, pp. well, Oxford, pp. 155–168.
177–197. Smith, D.G., 1986. Anastomosing river deposits, sedimentation
Rutten, M.G., 1967. Flat-bottomed glacial valleys, braided rivers rates and basin subsidence, Magdalena River, northwestern
and the beaver. Geologie en Mijnbouw 46, 356–360. Colombia, South America. Sedimentary Geology 46, 177–196.
Saucier, R.T., 1994. Geomorphology and Quaternary geologic Smith, D.G., 1987. Meandering river point bar lithofacies models:
history of the Lower Mississippi Valley. Mississippi River modern and ancient examples compared. In: Ethridge, F.G.,
Commission, U.S. Army Corps of Engineers, Vicksburg. Flores, R.M., Harvey, M.D., ŽEds.., Recent Developments in
Savat, J., 1975. Some morphological and hydraulic characteristics Fluvial Sedimentology. Society of Economic Paleontologists
of river-patterns in the Zaire Basin. Catena 2, 161–180. and Mineralogists Special Publication 39, Tulsa, pp. 83–91.
Schumann, R.R., 1989. Morphology of Red Creek, Wyoming, an Smith, D.G., Putnam, P.E., 1980. Anastomosed river deposits:
arid-region anastomosing channel system. Earth Surface Pro- modern and ancient examples in Alberta, Canada. Canadian
cesses and Landforms vol. 14, 277–288. Journal of Earth Sciences 17, 1396–1406.
Schumm, S.A., 1968. Speculations concerning paleohydraulic Smith, D.G., Smith, N.D., 1980. Sedimentation in anastomosed
controls on terrestrial sedimentation. Geological Society of river systems: examples from alluvial valleys near Banff,
America Bulletin 79, 1573–1588. Alberta. Journal of Sedimentary Petrology 50, 157–164.
 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196 195

Smith, N.D., 1975. Sedimentary environments and Late Quater- Longitudinal facies architectural changes of a Middle Holocene
nary history of a ‘low-energy’ mountain delta. Canadian Jour- anastomosing distributary system ŽRhine–Meuse delta, central
nal of Earth Sciences 12, 2004–2013. Netherlands.. Sedimentary Geology 85, 203–220.
´
Smith, N.D., Perez-Arlucea, M., 1994. Fine-grained splay deposi- ¨
Tornqvist, T.E., Bridge, J.S., Asselman, N.E.M., 1996. Quantita-
tion in the avulsion belt of the lower Saskatchewan River, tive analysis of overbank deposition in the Rhine–Meuse and
Canada. Journal of Sedimentary Research B 64, 159–168. Mississippi deltas: implications for 3-d modelling of alluvial
Smith, N.D., Cross, T.A., Dufficy, J.P., Clough, S.R., 1989. architecture. In: Programma en abstracts lezingen, 3 e Neder-
Anatomy of an avulsion. Sedimentology 36, 1–23. lands Aardwetenschappelijk Congres, 2–3 mei 1996, Veld-
¨
Smith, N.D., McCarthy, T.S., Ellery, W.N., Merry, C.L., Ruther, hoven. Koninklijk Nederlands Geologisch Mijnbouwkundig
H., 1997. Avulsion and anastomosis in the panhandle region of GenootschaprKoninklijke Nederlandse Akademie van Weten-
the Okavango Fan, Botswana. Geomorphology 20, 49–65. schappenrNederlandse Organisatie voor Wetenschappelijk
´
Smith, N.D., Slingerland, R.L., Perez-Arlucea, M., Morozova, Onderzoek, lezing 45.
G.S., 1998. The 1870s avulsion of the Saskatchewan River. ´
Tricart, J., 1959. Geomorphologie dynamique de la moyenne
Canadian Journal of Earth Sciences 35 Ž4., 453–466. ´ du Niger ŽSoudan.. Annales de Geographie
vallee ´ 68, 333–343.
Stanistreet, I.G., McCarthy, T.S., 1993. The Okavango Fan and Tricart, J., 1965. Rapport de la mission de reconnaissance
the classification of subaerial fan systems. Sedimentary Geol- ´
geomorphologique ´ moyenne du Niger ŽJanvier–
de la vallee
ogy 85, 115–133. ´
Avril 1957.. Memoire de l’Institut Fondamental d’Afrique
Stanistreet, I.G., Cairncross, B., McCarthy, T.S., 1993. Low sinu- Noire 72, Institut Fondamental d’Afrique Noire, Dakar.
osity and meandering bedload rivers of the Okavango Fan: Van den Berg, J.H., 1995. Prediction of alluvial channel pattern of
channel confinement by vegetated levees ´ without fine sedi- perennial rivers. Geomorphology 12, 259–279.
ment. Sedimentary Geology 85, 135–156. Van der Woude, J.D., 1983. Holocene paleoenvironmental evolu-
Sternberg, H.O’R., 1959. Radiocarbon dating as applied to a tion of a perimarine fluviatile area; geology and paleobotany
problem of Amazonian morphology. Comptes rendus du XVI- of the area surrounding the archeological excavation at the
IIe Congres` International de Geographie,
´ Rio de Janeiro, 1956, Hazendonk river dune ŽWestern Netherlands.. Analecta Prae-
Tome 2. Comite´ National du Bresil, ´ ´
Union Geographique historica Leidensia 16. Instituut voor Prehistorie, Rijksuniver-
Internationale, Rio de Janeiro: pp. 399–424. siteit Leiden, Leiden.
Stouthamer, E., Berendsen, H.J.A., 1997. Holocene avulsions in Van der Woude, J.D., 1984. The fluviolagoonal palaeoenviron-
the Rhine–Meuse delta, The Netherlands. Abstracts, 6th Inter- ment in the RhinerMeuse deltaic plain. Sedimentology 31,
national Conference on Fluvial Sedimentology, 22–26 395–400.
September 1997, Cape Town. University of Cape Town, Cape Van Gelder, A., Van den Berg, J.H., Cheng, G., Xue, C., 1994.
Town, 200. Overbank and channelfill deposits of the modern Yellow River
Taylor, C.F.H., 1999. The role of overbank flow in governing the delta. Sedimentary Geology 90, 293–305.
form of an anabranching river: the Fitzroy river, northwestern Warwick, P.D., Flores, R.M., 1987. Evolution of fluvial styles in
Australia. In: Smith, N.D., Rogers, J. ŽEds.., Fluvial Sedimen- the Eocene Wasatch Formation, Powder River Basin,
tology VI Special Publication of the International Association Wyoming. In: Ethridge, F.G., Flores, R.M., Harvey, M.D.
of Sedimentologists 28, Blackwell, Oxford, pp. 77–91. ŽEds.., Recent Developments in Fluvial Sedimentology. Soci-
Taylor, G., Woodyer, K.D., 1978. Bank deposition in suspended- ety of Economic Paleontologists and Mineralogists Special
load streams. In: Miall, A.D. ŽEd.., Fluvial Sedimentology. Publication 39, Tulsa, pp. 303–310.
Canadian Society of Petroleum Geologists Memoir 5, pp. Water Survey of Canada, 1991a. Historical streamflow summary,
257–275. British Columbia, 1990. Water Survey of Canada, Ottawa.
Thomas, R.G., Smith, D.G., Wood, J.M., Visser, J., Calverley- Water Survey of Canada, 1991b. Historical streamflow summary,
Range, E.A., Koster, E.H., 1987. Inclined heterolithic stratifi- Manitoba, 1990. Water Survey of Canada, Ottawa.
cation: terminology, description, interpretation and signifi- Weerts, H.J.T., 1996. Complex confining layers; architecture and
cance. Sedimentary Geology 53, 123–179. hydraulic properties of Holocene and Late Weichselian de-
¨
Tornqvist, T.E., 1993a. Holocene alternation of meandering and posits in the fluvial Rhine–Meuse delta, The Netherlands.
anastomosing fluvial systems in the Rhine–Meuse delta Nederlandse Geografische Studies 213, Koninklijk Nederlands
Žcentral Netherlands. controlled by sea-level rise and subsoil Aardrijkskundig GenootschaprFaculteit Ruimtelijke Weten-
erodibility. Journal of Sedimentary Petrology 63, 683–693. schappen, Universiteit Utrecht, Utrecht.
¨
Tornqvist, T.E., 1993b. Fluvial sedimentary geology and chronol- Weerts, H.J.T., Bierkens, M.F.P., 1993. Geostatistical analysis of
ogy af the Holocene Rhine–Meuse delta, The Netherlands. overbank deposits of anastomosing and meandering fluvial
Nederlandse Geografische Studies 166, Koninklijk Nederlands systems, Rhine–Meuse delta, central Netherlands. Sedimen-
Aardrijkskundig GenootschaprFaculteit Ruimtelijke Weten- tary Geology 85, 221–232.
schappen Universiteit Utrecht, Utrecht. Wells, N.A., Dorr, J.A., 1987. Shifting of the Kosi River, North-
¨
Tornqvist, T.E., 1994. Middle and late Holocene avulsion history ern India. Geology 15, 204–207.
of the River Rhine ŽRhine–Meuse delta Netherlands.. Geology Wightman, D.M., Pemberton, S.G., Singh, C., 1987. Depositional
22, 711–714. modelling of the upper Mannville ŽLower Cretaceous., east
¨
Tornqvist, T.E., Van Ree, M.H.M., Faessen, E.L.J.H., 1993. central Alberta: implications for the recognition of brackish
196 B. Makasker Earth-Science ReÕiews 53 (2001) 149–196

water deposits. In: Tillman, R.W., Weber, K.J. ŽEds.., Re- Wright, V.P., Marriott, S.B., 1993. The sequence stratigraphy of
sevoir sedimentology. Society of Economic Paleontologists fluvial depositional systems: the role of floodplain sediment
and Mineralogists Special Publication 40, Tulsa, pp. 189–220. storage. Sedimentary Geology 86, 203–210.
Woodyer, K.D., Taylor, G., Crook, K.A.W., 1979. Depositional Yonechi, F., Maung, W., 1986. Subdivision on the anastomosing
processes along a very low-gradient suspended-load stream: river channel with a proposal of the Irrawaddy type. The
the Barwon River, New South Wales. Sedimentary Geology Science Reports of the Tohoku University, 7th Series ŽGeogra-
22, 97–120. phy. 36 Ž2., 102–113.