River Meanders, Tributary Junctions, and Antecedent Morphology

Abstract

Tributaries to meandering rivers rarely join the river on the interior of bends. The limited drainage area on bend interiors explains why tributaries seldom form there, but not why existing tributaries are redirected as meanders develop. Other relevant factors include flow dynamics at junctions, runoff partitioning on inner vs. outer bends, and tributary deflection as the main channel migrates laterally. This study investigated whether the lack of confluences on bend interiors applies to lower coastal plain rivers in South and North Carolina, USA, where the factors above are not necessarily active, and if so how tributaries at sites of developing meanders are redirected. Of the 121 confluences examined using GIS data supplemented with field observations, none occurred on meander bend interiors. A total of 17 cases of potentially deflected tributaries were identified. Of these, 11 had sufficient evidence for a confident interpretation of how redirection occurred. In all 11 cases, pre-bend river paleochannels were involved in redirecting the tributaries away from the bend interior. This is explained by a model showing that the local slope gradient and mean depth advantages of the paleochannels provide velocity, stream power, and shear stress advantages over extension of the tributary channel into the bend interior. The results illustrate the importance of local hydraulic selection, and the influence of antecedent morphology on river hydrology and geomorphology.

1. Introduction

Tributaries to meandering rivers very rarely join the trunk stream on the inside of meanders on the main channel. Nearly all confluences occur on the outer portion of meander bends or on straight reaches. This has been noted and commented on since the early 20th century [1,2] but rarely analyzed. The purpose of this study is to investigate whether this pattern holds for the lowermost reaches of rivers in the South and North Carolina coastal plain and to explore the mechanisms of tributary deflection on evolving meanders in the region.

Hydro- and morphodynamics at meander bends inhibit tributaries on bend interiors and may promote tributary connections on bend exteriors, while limited drainage area and flow partitioning limit tributary formation and development within meander bend interiors. These phenomena adequately explain why tributaries rarely develop connections on existing bend interiors but cannot explain situations where confluences occur on initially straight reaches that subsequently develop into trunk stream bends that migrate away from the tributary. The latter requires that tributaries are deflected away from the interior of the growing bend. However, mechanisms of tributary deflection at developing meanders characterized by mobile channel margin and point bars are not applicable to the study area, where mobile bars are rare and point bars are fully vegetated. A lack of confluences at meander bend interiors and tributary deflection in the study area setting would therefore imply some other cause at work. Identifying such a mechanism could shed light on the poorly studied geomorphology and hydrology of low-gradient rivers with strong backwater effects. If tributary deflection (as well as a lack of tributary development) at trunk stream bends occurs in fluvial systems in different environmental settings and with multiple mechanisms, this suggests equifinality. This provides important contra-indications useful for interpreting fluvial system evolution and changes, as the rare exceptions—tributaries that do connect at bend interiors—indicate local or idiosyncratic controls at work.

These points are discussed in more detail below.

Why do tributary streams rarely join the main stream on meander bend interiors? The hydrology and fluvial geomorphology of meandering rivers is such that the flow velocity on the outside of the bend is higher, while the flow on the inside is slower. This typically results in erosion on outer bends (often characterized as cutbanks) and deposition on the inner bend in the form of point bars. Tributaries are likely to enter the main stream where the flow is more stable and less prone to deposition. At inner-bend point bar sites, blockages and accretion are prone to impede any tributary inflow. Further, tributaries often have steeper slopes than the main stream. Confluences with a steeper gradient that allows more efficient water flow occur outside of bends or on straighter sections of the river. However, the trend has rarely been critically examined. Studies of meander bend dynamics do not address the relationship of tributaries to evolving bends, as shown in a recent review [3]. A book-length treatment of theoretical morphodynamics of river meanders mentions tributaries and confluences only one time each, and not in the same passage [4].

Callaway in 1902 proposed that tributary confluences on the outer bank of meander bends are caused by deposition of the tributary’s sediment on the river bank across from the tributary mouth. This deposition deflects trunk stream flow to the opposite bank, where the confluence occurs, inducing erosion and bend development [1]. Davis [2] pointed out that deposition of tributary sediments in the main channel usually occurs at the mouth of the tributary, not the opposite bank. He also suggested that migration of bends downvalley can capture tributaries at the outer bend, where tributary channel lengths are shortened (and thus slopes are increased) by truncation.

Flint’s (1980) study of fluvial tributary arrangements showed that on meander bend interiors there is limited drainage area, and therefore limited runoff generation, inhibiting tributary formation [5]. Abrahams (1984) confirmed the trend and also examined junction angles and found that the proportion of large tributaries on the concave side of bends is affected by junction angle adjustments that deflect a main stream toward a large tributary, thereby creating a bend with the large tributary on its concave side (outer bend) [6]. This increases the proportion of larger tributaries on the outside of bends.

Studies of flow dynamics at tributary confluences have shown that confluent bends (where tributary junctions occur at or near the apex of trunk stream meander bends) represent hydrodynamically stable or quasi-stable morphologies. For instance, ref. [7] analyzed flow structures at a confluent bend. Field measurements of velocity and bed topography revealed complex hydrodynamics that respond to changes in momentum–flux ratios, while channel morphology remains relatively stable. Because the planform of the confluent bend is consistent over time, they suggested that this represents a quasi-stable channel morphology. The fieldwork in [7] is based on a single site, but the results generally conform to laboratory and numerical modeling results and have been confirmed in other field studies (e.g., [8,9,10]). For tributaries on outer bends, the typical erosion of the main channel reduces the length of the tributary channel, thereby increasing its slope. This not only helps maintain efficient flow, but can also promote incision, reinforcing the outer-bend tributary location [11]. There exists a large body of hydrodynamic research on fluvial confluences, including confluent bends, and overall these support the idea that outer-bend confluences are stable configurations, but do not explain the absence of inner-bend confluences beyond implying that such junctions are likely to be less hydrodynamically stable.

The existing explanations do not, however, explain how tributary confluences on straight reaches of an actively meandering river that begins developing a meander extending away from the tributary mouth are deflected or displaced. This is not a major issue if the meander growth is initiated as a mobile channel margin bar, and meander growth includes the development and extension of a mobile, often sandy, point bar. In these cases, the tributary flow is deflected by the bar toward the downriver end of the developing bend. This explanation does not apply to actively migrating incised or entrenched meanders. In the only study of this phenomenon to date, it was shown that runoff partitioning dynamics on bend interiors inhibit the development of tributaries, and result in flow efficiency advantages for existing tributaries to redirect to the downstream end of the extending bend [11]. While the setting of the study area streams is quite different from the fluviokarst systems examined in [11], the bend interiors in the study area are characterized by conditions that also inhibit channel formation.

The bar deflection of tributary mouths may also not apply to situations where active point bars are minimally mobile, due to vegetation cover, often in combination with cohesive fine-grained sediments and soils. This is the case in the lower coastal plain river reaches in the study area, which have been described as “vegetation-bound” with respect to lateral migration [12]. The apparent inapplicability of existing explanations to the displacement or deflection of tributary mouths initially on straight reaches that evolve into meander bend interiors is the primary reason for this study. A second motivation is the poorly understood hydrology of the low-gradient, backwater-affected lower reaches of coastal plain rivers such as those of the study area. Such lowland rivers have specific scientific, management, and engineering issues that often do not conform to the general understanding of alluvial river hydrology and geomorphology [13]. Common deviations from alluvial rivers in general include pronounced backwater effects, reversing flows, very low sediment inputs from upstream, more frequent overbank flows, and more frequent and complex patterns of channel–floodplain–wetland connectivity (e.g., [14,15,16,17,18,19,20]). Indeed, the inner- vs. outer-bend tendencies of tributary junctions are sometimes less pronounced in meandering tidal channels [21,22]. On the other hand, while meander dynamics of tidal channels were once thought to differ fundamentally from those of fluvial systems, recent work shows the similarity of planform dynamics of fluvial and tidal systems [23]. Of relevance in the context of meander bends and lateral channel migration in fluvial–estuarine transition zones is the role of antecedent Holocene morphology in the routing and partitioning of inputs from both upstream and downstream [24,25,26].

The development of meander bend interiors at tributary mouths may be broadly analogous to the tributary damming by trunk stream development examined by [27]. The Mooi River (South Africa) is a much different fluvial system than those in the study area, but this study does address a case of a tributary junction at a meander bend interior, and the effects of trunk stream floodplain development. In that case, floodplain extension on the trunk stream raised the base level of the tributary stream, resulting in more diffuse tributary flow and wetland formation. Rather than deflection, the tributary was disconnected from the trunk stream, terminating in a pond.

1.1. Tributary Deflections

Tributary stream alignments that approach bend interiors but shift course abruptly toward the downstream end of the river bend suggest deflection. Tributaries generally join the trunk stream at acute angles relative to the downstream direction, to the extent that flow directions and channel network organization can be inferred from algorithms based on this principle [28,29]. Most junction angles are acute, consistent with empirical observations and theoretical models of branching flow networks [30,31]. Obtuse angles occur “only in extreme circumstances”, regardless of the relative size of the main stream and tributary channels [32]. Thus, downstream tributary deflection is consistent with maintenance of an acute angle. Deviations from acute angles are often utilized as clues of local controls or unusual circumstances that provide evidence of geomorphological controls or of phenomena such as stream capture [32,33,34]. The apparently rare cases of tributary confluences on meander bend interiors could conceivably play a similar role.

A study on the Mekong River found an unusually high number of obtuse junctions (30 percent) [32]. Meander extension of the incoming tributary and deflection across bedrock shoulders were found to be the dominant causes of obtuse tributary junctions. They noted, however, that deflection can also occur as natural levees and other features develop on the main river floodplain. A study of reconfigurations at channel confluences of Russian rivers included one example of an inner-bend confluence that was reoriented by mutual adjustments of both channels rather than downstream deflection of the smaller channel [33], though in that case there was a minimal disparity in the size of the two streams.

The strong dominance of the main stream ensures that a bar deposit at the tributary mouth, or a channel margin bar growing across it, causes the tributary to be deflected downstream. With tributary junction angles acute to start with, this displacement may result in a tributary flow almost parallel to the trunk stream flow and the channel margin bar [34].

If a tributary is not deflected at a developing river bend, it would have to extend its lower reach as the main channel migrates laterally, or experience termination in a floodplain lake or wetland as found by [27]. Possible extension is explored below.

1.2. Tributary Extension

For the case of a tributary junction at the site of lateral migration or bend development away from the tributary mouth, to maintain its location relative to the main channel the tributary channel must extend or lengthen as the trunk channel migrates. At a given point upstream of the initial tributary mouth at elevation H_D the elevation is H_u and distance (channel length) to the pre-migration mouth is L₁, with L₂ denoting the length of an extended channel keeping pace with trunk stream lateral migration. The relative slopes are

$${\displaystyle \frac{S_1}{S_2}}={\displaystyle \frac{(H_u-H_D)/L_1}{(H_u-H_D)/L_2}}$$

(1)

It is reasonable to assume that the downstream, tributary mouth H_D and upstream elevations are equal before and after, so

$${\displaystyle \frac{S_1}{S_2}}\approx {\displaystyle \frac{L_1}{L_2}}$$

(2)

Because L₂ > L₁, if the tributary channel extends, S₂ < S₁. The relative mean or characteristic velocities of the before-and-after tributary channels according to the Chézy equation are

$${\displaystyle \frac{V_1}{V_2}}={\displaystyle \frac{C_1\sqrt{R_1{S}_1}}{C_2\sqrt{R_2{S}_2}}}$$

(3)

C is the Chézy coefficient, and R is hydraulic radius (cross-sectional area by wetted perimeter, approximately equal to the mean depth in most cases). Assuming C₁ $\approx$ C₂,

$${\displaystyle \frac{V_1}{V_2}}={\left({\displaystyle \frac{R_1}{R_2}}\right)}^{0.5}{\left({\displaystyle \frac{S_1}{S_2}}\right)}^{0.5}$$

(4)

The (hypothetical) extended tributary channel is therefore a less efficient flow pathway than the pre-migration path. Now we turn to comparing the extended tributary channel to an alternative deflected pathway, denoted with the subscript d.

$${\displaystyle \frac{V_d}{V_2}}={\left({\displaystyle \frac{R_d}{R_2}}\right)}^{0.5}{\left({\displaystyle \frac{S_d}{S_2}}\right)}^{0.5}$$

(5)

One measure of sediment transport and erosive capacity is stream power per unit weight of water y = VS, so by plugging Equation (5) into the unit stream power equation we obtain

$${\displaystyle \frac{{\psi }_d}{{\psi }_2}}={\left({\displaystyle \frac{R_d}{R_2}}\right)}^{0.5}{\left({\displaystyle \frac{S_d}{S_2}}\right)}^{1.5}$$

(6)

The mean boundary shear stress is τ = ρgRS, where ρ is water density and g is the gravity constant. Since ρg is constant,

$${\displaystyle \frac{{\tau }_d}{{\tau }_2}}=\left({\displaystyle \frac{R_d}{R_2}}\right)\left({\displaystyle \frac{S_d}{S_2}}\right)$$

(7)

The velocity advantage of a deflected channel relative to an extended channel, and any advantage in erosive capability is therefore a function of hydraulic radius or mean depth and slope gradient relative to an extended channel. This will be used to assess the potential advantages of deflected tributary channels vs. (hypothetical) extended channels.

2. Study Area

The study area includes coastal plain rivers of the Neuse and Cape Fear River systems of North Carolina and the Winyah Bay drainage area of South Carolina (Figure 1). The rivers were selected to represent larger regional rivers rising in the mountains and Piedmont as well as smaller systems mainly or entirely within the coastal plain. The study rivers also include “blackwater” systems whose drainage basins are mostly forested, with extensive riparian swamps, and confined to the coastal plain (so named because organic acids combined with low suspended sediment concentrations give the water a dark color). “Brownwater” rivers with higher turbidity and mainly originating inland of the coastal plain are also represented. The climate is humid subtropical with precipitation occurring throughout the year and a mean annual precipitation of 1200-to-1500 mm yr⁻¹. The regional geomorphic setting is a low-gradient, low-relief coastal plain consisting of Pleistocene and Holocene coastal terrace surfaces composed of sand, marine clay, and fluvial sediments.

The study rivers include the Black River and its tributary, Black Mingo Creek, SC, and the Great Pee Dee and Waccamaw Rivers, SC. Though the Waccamaw and Great Pee Dee Rivers are connected, their confluence is within the estuarine portion of the system, and they are separate watersheds in their fluvial reaches. The Cape Fear River and its tributary, the Northeast Cape Fear River, NC, were included, along with the lower Neuse and Trent Rivers, NC. The Pee Dee, Cape Fear, and Neuse Rivers are brownwater rivers that rise in the North Carolina Piedmont province; the others are blackwater systems confined to the coastal plain. All the studied reaches are subject to backwater effects due to astronomical or wind tides and storm surges. Drainage areas range from 626 km² (Black Mingo Creek) to 23,730 km² (Cape Fear) and 47,060 km² (Pee Dee). These rivers were selected because they have been studied in a previous work [35], where the environmental setting is described in detail.

The studied reaches are low-gradient (channel slopes of 1.0 × 10⁻⁵ to 7.0 × 10⁻⁵). The watersheds have mixed land uses, but within the studied reaches the river corridors are mainly forested. Urbanization is minimal along the studied reaches, though cities and towns are present further upstream and at the estuaries at the lower end.

3. Methods

Lengths of the lower reaches of the study rivers ranging from 39 to 183 km were examined, covering the fluvial–estuarine transition zone as defined in [35]. The exception is the Waccamaw River, where the entire 226 km length was examined. These are all freshwater reaches, though salinities of up to about 2 psu may occur near the heads of estuaries during low-flow periods or storm surges. Geographical Information System data obtained from the U.S. Geological Survey via the National Map (https://apps.nationalmap.gov/viewer/, accessed on 14 April 2025) were examined, including topographic maps, digital elevation models (DEMs), and color aerial images. The topographic map scale is 1:24,000, and the DEM is from the 3DEP Elevation model, with a pixel size of 1/3 arc-second (about 10 m). The aerial images (U.S. National Aerial Imagery program, supplemented by images from Google Earth^TM (Google Earth Pro 7.3.6.9345) are at variable horizonal resolutions ranging from 0.3 to 30 m. Each tributary that was not confined to the river floodplain and valley bottom was examined: that is, only those tributaries whose headwaters were in the uplands outside the valley floor on the main stream were included. Soil maps were obtained via the U.S. Department of Agriculture SOILWEB system, based on soil maps at a 1:24,000 scale. (https://casoilresource.lawr.ucdavis.edu/gmap/, accessed 4 December 2024).

For each tributary confluence, its occurrence on the left or right bank was recorded. The planform of the receiving stream was recorded as a bend if a curve was present where the amplitude/wavelength ratio was >0.25. Otherwise, the planform was labeled as straight. The wavelength is the straight-line distance from the upstream to the downstream end of the curve, and the amplitude is the distance from the wavelength baseline to the outer bank at the apex. Though the details vary from study to study, these morphometric measurements are typical in studies of meander bends (e.g., 3, 4, 13, 23), though the 0.25 ratio was established for this study to include developing as well as strongly developed bends. The location of the confluence was recorded as outside the bend if it occurred at or near the bend apex on the outer, convex portion of the bend, and inside the bend if the confluence was along the wavelength baseline on the concave, inner portion of the bend. If the junction was on the outer bend but on the upper or downstream portion rather than near the apex, it was recorded as a straight/outer bend. As data collection proceeded with no inner-bend confluences observed, another category was added—straight/inner bend—for confluences on the concave side of minor bends that did not meet the amplitude/wavelength criterion. In addition, field observations via kayak were made on Black Mingo Creek and the Waccamaw, Northeast Cape Fear, Neuse, and Trent Rivers. The main purpose was to determine whether unmapped tributary confluences could be observed at bend interiors.

Tributaries were identified that approached a meander bend interior within the floodplain and valley floor of the trunk stream but abruptly turned downstream, away from the bend interior (Figure 2). These were considered as potentially deflected tributaries, and were evaluated for evidence of deflection, based on evidence in the GIS data or in the field. They were also assessed with respect to the ratios in Equations (5)–(7). While precise numerical values for the hydraulic radius and slope could not be produced from the GIS data, the relative values for the apparently deflected channels vs. the hypothetically extended tributary could be, allowing a determination of whether the ratios are >1.

4. Results

4.1. Tributary Junctions

Table 1. Tributary junctions on the study rivers. Length indicates the length of the channel examined upstream from the estuary. L, R designates the left or right bank of the main stream. OB = confluence on the outer bend, S = confluence on the straight reach, IB = confluence on the bend interior. “N Field” indicates the number of bends observed in the field, including those without tributary confluences.

River	Length (km)	L	R	OB	S	IB	S/OB	S/IB	N Field
Black River, SC	183	10	12	10	7	0	4	1	3
Black Mingo Creek, SC	39	2	4	1	5	0	0	0	13
Great Pee Dee River, SC	174	13	6	7	11	0	1	0	3
Waccamaw River, SC	226	11	12	14	5	0	4	0	77
Cape Fear River, NC	121	6	14	6	10	0	2	2	3
NE Cape Fear River, NC	144	5	7	7	3	0	2	0	13
Neuse River, NC	52	4	5	2	5	0	2	0	22
Trent River, NC	50	4	6	3	5	0	1	1	16

shows that low-gradient coastal plain streams in the study area influenced by coastal backwater effects exhibit the same pattern as other fluvial systems: that is, tributaries join the main stream on straight reaches of the trunk stream, or on outer meander bends. Of the 121 confluences examined, 50 junctions (41 percent) were on outer bends of the main stream, and 51 (42 percent) were on straight reaches. An additional 16 (13 percent) were in the straight/outer bank category. No confluences occurred on bend interiors, though four fell into the straight/inner-bend category.

The study rivers generally have low banks and are not incised. Based on field observations, the bank tops or natural levees are <1.5 m above normal water levels, as indicated by bank vegetation, mud drape deposits, and morphology. Exceptions are where the channels abut older alluvial terraces along the valley sides. In many cases, in the fluvial–estuarine transition zone there are no distinct banks; rather, there is a gradual transition from open water to scattered hydrophytic trees in standing water to regularly inundated swamp or marsh. Floodplain soils and sediments are mostly organic mucks and peats, or fine-grained mineral deposits ranging from clay to loam in texture.

Unvegetated point bars are uncommon, though their occurrence increases in the Pee Dee, Waccamaw, and Northeast Cape Fear Rivers in the upper portions of the studied reaches. However, evidence of meander extension was observed in some cases, particularly on outer-bend cutbanks, where erosional scarps and bank failure scars are evident, along with some exposed woody roots.

No junctions with larger tributaries on bend interiors were detected on maps or imagery or observed in the field, and any subchannels on the floodplain in such inner-bend settings showed evidence of reversing flow via backflooding at higher water levels, and shallow return flow to channels as high water recedes. Along reaches where distinct banks or natural levees are otherwise evident, bend interiors are often characterized by shallow water, gradual channel-to-floodplain slopes, and frequent flooding, consistent with ongoing or recent bend development and their status as vegetated point bars.

4.2. Deflected Tributaries

Seventeen cases were found of tributaries that were potentially deflected, one of which is shown in Figure 2. As described in Methods, the alignment of these streams approaches a bend interior, but shifts abruptly toward the downstream end of the bend. In two cases (Halfmoon Swamp and Sterritt Swamp confluences with the Waccamaw River), while the general geometry of the tributary path suggests possible deflection, no other clear evidence of tributary redirection was observed. The junction of Indian Creek and the Cape Fear River is disturbed by artificial drainage features, preventing interpretation.

At several sites, the tributary junction occupied a topographically complex area of low-elevation, low-relief, and multiple, sometimes overlapping, paleochannels on the floodplain. At these sites (Lester Creek/Black River, Gore Creek/Waccamaw River, Jumping Creek/Trent River), tributary deflection may well have occurred due to local depressions or swales not evident as distinct paleochannels, due to slope or depth advantages (see Equations (5)–(7)), but this cannot be proven at present.

At the other 11 sites, tributaries either avulsed into paleochannels of the main river or occupied the pre-bend channel as the bend developed. These include the Tilly Swamp and West Ash Swamp junctions with the Waccamaw River, Carvers Creek confluence with the Cape Fear River, and Beaver Dam Creek confluence with the Trent River. All 11 sites were examined in detail, 7 of which are described below. The Waccamaw sites were excluded because of the poor quality of the available DEM data, and Carvers Creek was excluded because the evidence of deflection is equivocal. At all 11 sites, the width of the deflected portion of the tributary channel indicated a greater bankfull cross-sectional area than the tributary upstream of the deflection, and at least local slope advantages where the tributary encountered former channels of the trunk stream. Therefore, the velocity, stream power, and shear stress relative to an extension of the tributary channel favored the paleochannel occupation.

The Bigham Branch/Great Pee Dee River confluence is shown in Figure 3. Roughly across the base of the river bend, a ridge representing the pre-bend channel path is visible. An embayment representing a portion of the former channel is at the downstream end of the bend where the tributary connects. The abrupt bend of Bigham Branch behind the paleochannel levee ridge suggests that the tributary was deflected downstream as the bend extended, but this was a case of Bigham Branch simply remaining in the river paleochannel. Halfmoon Creek is similar, but appears to have been deflected due to an earlier episode of lateral channel migration. The junction with the Neuse River (Figure 4) shows the lower creek occupying the pre-bend Neuse channel.

Core Creek joints the Neuse River via an active paleochannel separated from the river by a set of meander belt ridges associated with the growth of the bend (Figure 5). Core and Halfmoon Creeks and Bigham Branch all feed water to the river paleochannels to maintain the embayments at the downstream end of the respective bends.

The Island Creek confluence with the Northeast Cape Fear River was shown as an example of a possibly deflected tributary in Figure 2. Figure 6, a DEM-derived slope map, shows a more geometrically complicated meander bend than the others discussed. Island Creek’s original connection with the river, shown as a paleochannel in Figure 6, persists as a high-flow channel at large discharges, but is not inundated at low flows. A pre-bend river paleochannel is evident, and the lower Island Creek now occupies this channel. This was an avulsion of the Creek, with the old river channel providing a more efficient flow path as the meander extended.

The Cape Fear River at Browns Creek shows evidence of a ridge associated with the natural levee of the channel alignment before bend growth (Figure 7). In this case, the highly sinuous tributary seems to have persisted for a time after the river bend began extending before deflection downstream, occupying the pre-bend river paleochannel. This was apparently an avulsion of the creek into the old river channel.

The role of a pre-bend paleochannel is also evident where Frenchs Creek flows into the Cape Fear River. In this case, a gap in the ridge associated with the pre-bend natural levee shows the original confluence (Figure 8), where the old channel runs along the valley side. Topography and soils (Figure 9) show this feature clearly. The soils mapped in the paleochannel (a mapping unit consisting of the Chewacla and Chastain series) are poorly drained, clayey soils. The outer parts of the bend interior, by contrast, are mapped as the Congaree series, a somewhat poorly drained soil that features a buried soil profile (soils are further described below).

The bend where Mulford Creek joins the Cape Fear River is the only case described here where there is a general downward slope from the base of the meander toward the river; in the other cases, the modern levee and more recent meander ridges are at a higher elevation than paleochannels or swales at the base of the meander. Note that the blue-line path of Mulford Creek across the bend shown on maps is in error; the creek actually follows the meandering path shown in Figure 10.

At five of the seven sites presented in detail, the straight-line distance from the point of apparent tributary deflection to the existing junction with the river is greater than the straight-line distance that would result from tributary extension across the bend interior. Bigham Branch is one exception. The other exception (Island Creek) shows a much shorter distance (85–90 m) via the former main connection (see Figure 6) than via the deflected channel (290 m). The ratio of the post-deflection straight-line distance to that of the hypothetical tributary extension is 1.37, or 1.12 if Island Creek and Bigham Branch are excluded. Straight-line distances are only a crude approximation, but these results do indicate that deflection did not occur simply to achieve shorter flow paths or steeper overall (deflection point to river) slope gradients.

Soils mapped at the 11 sites examined in detail are shown in

Table 2. Soils mapped on the bend interiors where deflected tributaries occur. Classification is at the subgroup level of the U.S. Soil Taxonomy. Texture sequence lists the textural classes encountered in the upper meter of a typical profile.

Series	Taxonomy	Texture Sequence	Tributaries
Chastain	Fluvaquentic Endroaquepts	Loam–clay loam–clay	Bigham Br., Browns Cr., Frenchs Cr., Mulford Cr.
Chewacla	Fluvaquentic Dystroquepts	Loam–silty clay loam–clay loam	Bigham Br., Browns Cr., Frenchs Cr., Mulford Cr.
Congaree	Oxyaquic Udifluvents	Loam	Bigham Br., Browns Cr., Frenchs Cr.
Johnston	Cumulic Humaquepts	Mucky loam–loamy fine sand–fine sandy loam	Gore Cr.
Masontown	Cumulic Humaquepts	Mucky loam–fine sandy loam–loamy sand	Halfmoon Cr., Core Cr.
Muckalee	Typic Fluvaquents	Loam–loamy sand–sandy loam	Halfmoon Cr., Core Cr., Island Cr., Jumping Cr., Beaverdam Cr., Lester Cr.

. The Masontown and Muckalee series are recent, poorly developed soils consistent with a growing meander, but are widely distributed along the Neuse, Trent, and other rivers. They are formed overlying stratified sandy sediments, but no sandy point bars were observed during extensive field observations in areas where these soils are mapped. The Johnston series is similar. The Chastain and Chewacla series are finer in texture and have mineralogical characteristics typical of alluvium where Piedmont-derived sediments comprise a significant proportion. In the Chastain series, the clay and loamy material may overlie sand, and the Bg horizons may be very firm. The Congaree series, whose properties also reflect a significant Piedmont contribution, consists of a poorly developed profile overlying buried Ab and Bb horizons. This may reflect a period of floodplain stability allowing pedogenic development followed by renewed deposition.

5. Discussion

The phenomenon of tributary junctions not occurring on interior bends of the main stream has long been known to apply to alluvial meandering rivers and has also been shown to apply to bedrock rivers with entrenched meanders [11,36]. As a result, this study suggests that the influence also applies to low-gradient, backwater-influenced coastal plain streams that differ from typical alluvial rivers, and that the phenomenon is likely universal. All confluences of larger tributaries in the study area occur on straight reaches or outer bends. This is consistent with the general pattern in alluvial rivers where hydrodynamics and morphodynamics inhibit the formation of tributaries on bend interiors due to a limited drainage area. In addition, in the wetland-dominated low-gradient systems of the study area low slopes slow surface runoff and high water tables inhibit channel incision by overland flow. There are only limited drainage areas available for tributary development on bend interiors in meandering rivers. The surface vs. subsurface runoff partitioning that hinders channel formation on bend interiors in the fluviokarst system studied in [11] is not applicable to the lower coastal plain reaches here, as the water table is often at or near the surface. Rather than diverting to the subsurface, runoff is ponded or occurs as diffuse surfaces flow.

The results are also consistent with a general pattern whereby tributary confluences on outer bends of the trunk stream are favored. The similar number of outer-bend vs. straight-reach connections suggests a possible tendency toward preferential connections on outer bends, given the greater length of straight reaches along the rivers. This could be related to meander bend extension truncating outer-bend tributaries, thus increasing their slope and facilitating their persistence. The stability of these confluent bends associated with the flow dynamics may also be a factor, as field evidence was observed of exactly the kind of flow dynamics described at stable confluent meander bends, where contrasts in the tributary and main stream flow due to leaves and other floating organic matter in the tributary made the flow interactions visibly apparent (e.g., Figure 11).

The factors above are sufficient to explain the lack of tributary development in bend interiors, but do not explain what happens to tributary confluences on initially straight reaches where lateral migration and bend development on the trunk stream away from the tributary mouth occurs. Development of mobile channel margin and point bars that accompany meander growth on many rivers [3,11,32] deflect incoming tributaries downstream, but this mechanism is generally absent in the study area. Coastal plain rivers of the Carolinas are characterized by low suspended and bed sediment loads in their lower reaches. This is an inherent trait of the blackwater rivers, but even in the Piedmont-draining brownwater rivers much of the sediment is trapped in floodplains upstream of the fluvial–estuary transition zone, and in upstream reservoirs. The low sediment supply and dominance of fine-grained and organic sediment limit formation of channel margin or tributary mouth sandbars that can be readily reworked to deflect tributaries. Thus, the presence of deflected tributaries approaching trunk stream inner bends indicates other mechanisms. All the cases examined in detail here indicate the persistence of tributary flow in pre-bend channels (now paleochannels) of the trunk stream, or avulsion into such paleochannels.

The presence and persistence of the paleochannels reflects the hydrologic and geomorphic setting of the study rivers. Low slopes and associated stream power and shear stress limit sediment transport capacity, and the extensive floodplains provide ample storage opportunities [12,14,15,35]. These combine in many cases to produce floodplain sediments and soils that are high in fine-grained silts and clays and in organic matter, and low in sand. This allows abandoned channels to persist on the floodplains, and low sediment input slows infilling. These paleochannels provide local slope and hydraulic radius advantages for the tributaries at developing bends, and capture or retain their flow, diverting the channel away from the bend interior toward the downstream end, an example of local hydraulic selection.

This phenomenon illustrates two key points. The first point is the role of antecedent morphology in the development of tributary junctions. In the case of outer bends, where lateral migration of the main channel encroaches on the tributary, antecedent morphology is erased by erosion, and truncation of the channel increases slope, which generally increases shear stress and stream power, reinforcing the tributary channel location. On bend interiors, the pre-bend channel tends to persist in some form, providing an advantageous alternative route compared to channel extension across the bend interior. This contrasts with many alluvial rivers, particularly sand and gravel-bed systems, where the growth of point bars associated with bend extension may preserve former levees, but generally not former channels.

The second key point is the role of local selection. Distances from the point where tributaries are deflected downstream to the receiving channel are often greater than those associated with extension across the point bar/bend interior, and the inflection point-to-river scale may have lower slopes. However, at the original (pre-bend) junction, there is likely to be a local slope advantage as well as an occupiable channel. The tributary can only “see” the local advantage, not the potential reach-scale option of channel extension. Some treatments of tributary/main stream interactions emphasize a broader, reach-scale efficiency of potential tributary flow paths rather than local slope variations (e.g., [2,36]).

As mentioned in the introduction, if tributary deflection and a lack of tributary development at trunk stream bends occurs in fluvial systems in different environmental settings and with multiple mechanisms, equifinality is indicated. Thus, a situation contrary to the general trend can indicate clues useful for interpreting fluvial system evolution and changes. This is evident in Hackney and Carling’s studies on the Mekong River [32]. To examine this implication in the study region, two situations of tributary connections on inner bends were identified on the Black River, NC, which was not included in the initial study design (the Black River of South Carolina was included).

The first example (Figure 12) shows a tributary that may have been deflected upstream rather than downstream, and has unusually linear ridges, and the second example is a small inner-bend tributary (both highlighted by boxes). These sites occur within an area where tectonic influences on fluvial morphology have been identified [37], though this site has not been previously analyzed. The atypical tributary junctions, coupled with the regional evidence of tectonic effects, led to a search for potential influences of faults and lineaments, resulting in the identification of the feature shown on the right side of Figure 12.

Though testing and confirmation of the role of geological structures will require further investigation, these examples show how atypical tributary junctions at meander bends indicate possible influences of local, idiosyncratic factors.

6. Conclusions

Low-gradient, backwater-affected coastal plain rivers are like other fluvial systems in that tributary junctions occur on outer bends of the rivers or straight reaches and do not occur on the inner, point bar side of meander bends. As in other rivers, there is limited drainage area on bend interiors for tributaries to develop, as well as minimal slope gradients and high water tables that inhibit channel incision. Also, in common with fluvial systems in general, truncation of tributary channels by lateral migration of outer river bends (cutbanks) increases tributary slopes, promoting their persistence.

In the study area, tributaries that once connected to the main river on straight reaches are apparently deflected downstream as bends develop and the main channel migrates laterally away from the tributary mouth. This occurs as the smaller streams continue to occupy the old river channel as lateral migration occurs, or due to avulsions into the pre-bend river paleochannels due to local hydraulic advantages. The results point to the importance of inherited morphological features and local hydraulic selection, as well as the role of morphological clues in hydrological and geomorphological interpretations.