BLACKWATER ANASTAMOSIS: PART 3

See parts 1 and 2 here:

Blackwater Anastamosis Part 1

Blackwater Anastamosis Part 2

Usually a subchannel in an anastamosing river system (that is, channel other than the dominant one) or slough is correctly considered to be a former main channel that was partially abandoned or bypassed due to an avulsion (channel shift). That's the way I've treated them, at least implicitly, in parts 1 and 2. That would make them older than the main channel, which could explain why parts of them are obscured by apparently encroaching vegetation.

But what if the subchannels are actually younger than the main channel? What if they represent flow diversions (avulsions) that have persisted without eroding a distinct channel along their entire length?

This is at least conceivable because the channels often have no banks to speak of, just a transition along the channel margin from channel to wet swamp, and the floodplains and valley bottom have such low relief. Let's look at some examples. The images are derived from Michael Davias' ovoid basin survey, which is based on processing high resolution LiDAR images to identify possible Carolina Bay wetlands. Because the dataset is specifically designed to identify nearly flat, low relief features in coastal plain environment, it is also well suited for viewing fluvial and wetland landforms in the region. The imagery shown here is based on LiDAR digital elevation models with horizontal resolutions of 1 m or better. 

Below is a section of the Little Pee Dee River valley. The whole valley bottom shown has no relief, other than the hummocky patterns common in alluvial swamps, and some visible channels and lakes. The lakes, which appear disconnected (and also appear so in other maps and aerial imagery), are actually connected to each other and to the Little Pee Dee at both ends (I know because I've paddled them) and convey flow at river stages well below flood levels. Furthermore, most of the valley bottom shown, though mostly forested, is nearly always inundated and conveying flow.

Section of the Little Pee Dee River valley, South Carolina. The differences in texture in the floodplain are due to different resolutions of available LiDAR data. The color legend also applies to the other images below. 

Another section of the Little Pee Dee is shown below. Jordan "Lake" is actually part of a continuous flow path connected to the river on both ends, with numerous other channels and water bodies that do not show up even in this high resolution image.

Section of the Little Pee Dee River valley, South Carolina. The differences in texture in the floodplain are due to different resolutions of available LiDAR data.

Lest you think this is confined to the Little Pee Dee River, here are a couple of other examples. The section of the Black River, North Carolina shown below, like the other examples, is essentially one valley-wide flow system of channels of various size (only a few visible), floodplain depressions, and swamps that usually convey flow.

Part of the Black River, North Carolina. 

Holly Shelter Creek, a tributary of the Northeast Cape Fear River, N.C., shows some fascinating patterns, transitioning from a low sinuosity channel to a strongly meandering planform, to a complex mosaic of mostly indiscernible (on this and other images) channels to its confluence with the river. 

Holly Shelter Creek, North Carolina. Arrow is the confluence with the Northeast Cape Fear River.

At all of these sites, and others like them, it is easy in a kayak or small boat to leave the main channel to explore numerous side channels, or to just paddle through the swamp forest. It is also very easy to get lost while doing so, though usually following the visible flow will get you back to the main channel. 

Paddling through the forest.

This sort of thing is all that prevents you from paddling almost anywhere in the valley bottoms.

We'll now explore several scenarios as to these channels and flow paths arise. One is a variation of the standard avulsion-and-anastamosis scenario discussed in parts 1 and 2. Flow diversions to trigger avulsions should be relatively easy, as there are no or low banks, so a crevasse is not required. A logjam--not uncommon in these systems--or simply overflow of the low banks is all it takes to deflect flow from an existing channel. Because of the low sediment loads, few of the new channels infill, and thus nearly all persist. 

A variation on this theme is simply overflow through the swamp forest. Overbank flow in these systems is quite common. In this scenario there is initially un- or poorly concentrated flow through the swamp forest. Gradient and resistance selection (here I can plug my Abiotic Selection in Earth Surface Systems book) will inevitably cause the flow to become more concentrated as the flow preferentially follows locally steeper (admittedly a subtle thing in this environment) and lower-elevation paths, and avoids blockages from vegetation and woody debris. Though not to the extent in steeper settings with less vegetation, these selected flow paths tend to be self-reinforcing, via basal erosion, removal of woody debris, and inhibition of tree establishment via substrate instability and removal of seeds. The wider, visible sections, often referred to as lakes or sloughs, are probably pre-existing depressions and paleochannels. Other possible origins are creation by flood scour in backswamp swales or depressions, beaver ponds, or in historically recent times, depressions associated with logging or sand mining.

Brunson Swamp, a Little Pee Dee River subchannel. The water looks still, but it was flowing when the photo was taken.

Another possibility is that the secondary flow paths are reoccupations of former channels. Some avulsion regimes in alluvial rivers are dominated by channel-switching, often involving paleochannels (Phillips, 2009). The latter represent low spots in the floodplain that can create local slope advantages, and sometimes have more easily erodible substrate. 

A final(?) possibility is that the channel complex is controlled or influenced by braided patterns formed during the Pleistocene. Braided channels are a characterized by multiple, intertwining channels separated by bars that are mostly un- or sparsely vegetated. Both the bars and channels are shiftier than in anastamosing systems. Sandy braided channels existed in many streams of the southeast Atlantic coastal plain during the Pleistocene under different conditions of climate, streamflow, sediment supply, and sea-level than is now the case (Leigh et al., 2004; Leigh, 2006). 

Here is a brief summary of the proposed mechanisms for development of the blackwater anastamosing channels. Based on experience (both my own and from the literature) I will not be surprised if additional explanations are identified, and I will be shocked if only one explanation (whichever it is) fits all the cases, even in the Carolinas.

1. Avulsion away from subchannel. The latter was once the main channel, and an avulsion switched flow to the current dominant channel. The subchannel is persisting as an anabranch or is in the process of infilling. 

2. Avulsion into subchannel. Flow diversion into the new flow path is underway.

2A. Anastamosis avulsion (formation of a new channel that reconnects with the dominant channel downstream). 

2B. Relocation avulsion underway--the newer pathway will eventually become dominant.

2C. Temporary avulsion--newer flow path is infilling and will eventually infill or become isolated from the main channel except during floods.

3. Overflow through the swamp. This is the scenario described earlier where general overflow of the channel margins forms concentrated preferential flow paths downvalley. 

4. Multiple channels controlled by underlying Pleistocene braided channels. 

In the 4th installment of this series, I will identify some possible ways to support or falsify these hypotheses--that is, if I can figure them out. 

--------------------------

References

Leigh DS. 2006. Terminal Pleistocene braided to meandering transition in rivers of the southeastern USA. Catena 66: 155–160.

Leigh DS, Srivastava P, Brook GA. 2004. Late Pleistocene braided rivers of the Atlantic Coastal Plain, USA. Quaternary Science Reviews 23: 65–84.

Phillips, J.D. 2009. Avulsion regimes in southeast Texas rivers. Earth Surface Processes and Landforms 34: 75-87. 

TAEDIUS YELLOW DUST

Here in the Carolina lowlands we are in the tail end of the annual coating of everything with pine pollen, mostly from lobolly pine (Pinus taeda, the scientific name being the source of the title pun, which beat out "Appallin' Pollen" for that dubious honor). As I write, oak pollen is still going strong . . . 

A hideous yellow soup of pine pollen in a forest canal in Charleston County, S.C.

A "bathtub ring" of pollen marks a recent fall in water level of Grinnell Creek, N.C. 

Loblolly pine is native to this region. Its natural habitat is on the fringes of and on high spots in swamps and other wet forested sites, where it is still easy to find--in fact the champion tree for the species is in a bottomland forest in the Congaree River system, South Carolina. We also have longleaf pine (Pinus palustris), pond pine (P. serotina), slash pine (P. elliottii), and probably a bit of shortleaf (P. echinata). But mostly, Pinus taeda is the leada. 

A "champion tree" (largest known living specimen) loblolly pine in floodplain of the Congaree River system in Congaree National Park, S.C. (photo from Parcation.com). 

Loblolly is pretty much everywhere in these parts. If you let a farm field or pasture lay fallow, in a few years some loblollies will have come in. Vast areas are in plantations owned by major timber companies (Weyerhauser and International Paper being predominant in the region) or by individual landowners. More vast areas, and many small forest and woodland patches, are in second (or third, or fourth) growth areas and in sites disturbed by storms, fires, and whatnot. 

By the early 20th century, much of the U.S. south was heavily eroded land and degraded soils, made so by cut-out-and-get-out lumber industry practices and a lack of soil and water conservation practices on farmland. This was most obvious in the Piedmont but was also the case on the coastal plain. Soil conservation and reforestation programs began in the 1930s, and one tree that could and did grow on poor soils and degraded sites was loblolly pine. I still find it useful in that regard. In September, 2018 Hurricane Florence produced a big washout at my place. I filled it in (we're talking dump trucks and front loaders here, not shovels and wheelbarrows), but I couldn't get anything to grow on it, even after I asked for advice from the county extension agent. Finally (in late 2019) my son Damien and I went into the adjacent forest and dug up some loblolly seedlings and planted them on the fill. They took off, and by 2025 the stand needed thinning (actually by 2024, but I did not get around to it until 2025). Florence also eroded bluffs of the Neuse River estuary, exposing sediment and soil ranging from impermeable clay to sand. The loblollies colonized it, as shown below.

                            Pinus taeda established on eroded bluffs after Hurricane Florence.

Erosion of the bluffs is ongoing, and accelerating due to climate change driving more frequent storms and sea-level rise. Thus, on many reaches the post-Florence pines have been undercut, particularly after a series of northeasters (meteorologists now like to call them coastal lows, but to folks here they are still northeasters or mullet blows). 

Post-Florence pines undermined by a series of northeaster storms in late 2025.

A couple of side notes: I mentioned above that in the early 20th century much of the south was eroded, worn-out land that nobody much wanted by the 1930s. Many of the National Forests in the south were established in the 1930s by acquiring this land that no one would pay much for. This is the case for Francis Marion National Forest on the S.C. coastal plain, and Croatan N.F. on the N.C. coastal plain, established 19 days apart in 1936. Other examples: Bienville National Forest, Mississippi, 1936; Daniel Boone N.F., Kentucky, 1937; Jefferson N.F., Virginia, 1936; Sam Houston and Sabine National Forests, Texas, both 1936; Talladega N.F., Alabama, 1936. A few national forests in the south were established from 1907-1927; a few after the 1930s. 

Highly resinous southern yellow pines such as Pinus taeda were not suitable for paper until some technological problems were solved. Carl Dahl invented the kraft process in 1879, but the big breakthrough was from Georgia chemist Charles Herty who found a way to make quality paper from loblolly and other southern yellow pines in 1932. Loblolly became not only a way to stem erosion and begin healing worn-out land, but a way to make money and help pull southern states out of the depression. Commercial silviculture operations can typically harvest trees for pulpwood every 25-30 years in the Carolinas (longer for sawtimber), but I've heard of rotations as low as 20 years. 

I have no great love for the timber industry, or for the sticky-resin, cone-shedding, poor firewood loblolly pines (I've said on many occasions: a loblolly is better than no tree at all, but just barely). But I have to respect the ecological fitness of the unstoppable tree that came out of the swamps and took over much of the southern landscape--yellow pollen and all. 

BLACKWATER ANASTAMOSIS: PART 2

 This continues the post started here, where it was explained that some blackwater rivers of the Carolinas (and probably elsewhere) have anastamosing patterns that cannot be explained by the typical avulsion-based conceptual model for alluvial rivers. 

When a channel splits, or bifurcates, either both of the channels (old and new) persist, or one or the other gradually infills and becomes abandoned as a slough or paleochannel (which may be reactivated during floods). If both channels persist and rejoin downstream, this is an anastomosis avulsion. If both persist but never rejoin this is a distributary avulsion (think deltas) or occasionally a watershed fragmentation avulsion. Otherwise, one channel or the other is gradually abandoned, generally becoming plugged and disconnected first at the upstream, and later at the downstream end. 

Anastamosing patterns in the delta of the Great Pee Dee River (left), South Carolina. The channel splits are a combination of distributary avulsions (e.g., see diversion at the arrow to the Waccamaw River channel) and anastamosing avulsions (e.g., Cooter Creek). 

In some cases, however, in the blackwater rivers, channels do not appear to become abandoned and disconnected. They may sometimes appear to be isolated in air photos and other imagery because of forest canopies, but water flows into and out of these features at normal flow levels; not just floods. The photo below shows water flowing from the Little Pee Dee River into a channel called Jordan Lake. On some images this connection is invisible, and on some maps no connection is shown. But as the photo shows, the flow is quite vigorous. On the date of the photo stages and discharges on the Little Pee Dee at both the upstream and downstream gages were below median levels, and way below designated flood levels. You can also see the trees and cypress knees, and in the summer the channel is almost entirely covered by the tree canopy. 

Water flowing from the Little Pee Dee River into Jordan Lake.

The image below shows that the Jordan Lake channel reconnects with the Little Pee Dee River just upstream of the U.S. Highway 378 crossing. The channel is also fed in its lower reaches by a diversion from the Tar Lake area. Multiple such features appear in the lower Little Pee Dee River area, and in other rivers (see previous post here).

Arrows show where Little Pee Dee River (dotted line) water is diverted into Jordan Lake (upper and lower arrows) and a diversion from the Jordan Lake channel to Brunson Swamp. 

Here is my proposed scenario:

1. An avulsion (diversion of flow) occurs.

2. The sediment load of the blackwater river is insufficient to plug or infill the "abandoned" channel at the upstream (or downstream) end. 

3. Some accretion and woody debris accumulation occurs, allowing vegetation to become established in the former channel, but flow through continues. 

4. The limited sediment load prevents the semi-abandoned channel from infilling, and also keeps floodplain elevations low enough to allow frequent, perhaps nearly constant, cross-floodplain flow. 

The flow path of the former channel thus persists, despite establishment of swamp vegetation.

This scenario can, I believe, explain these features. But there are other potential scenarios as well, while I'll explore in Part 3. 

Other examples from the Little Pee Dee River. Arrows show locations where water is diverted from the river into subchannels that are not always fully evident on imagery. 

Maps shown above are based on imagery from the U.S. Geological Survey National Map. 

BLACKWATER ANASTAMOSIS: PART 1

I've had a couple of paddling trips recently on the Little Pee Dee River, South Carolina, and its backwaters and side channels. Like some other blackwater areas in the coastal plain of the Carolinas, there is a mosaic of channels and floodplain swamp wetlands, all intertwined with side channels, and with most of the floodplains having water deep enough to paddle through if you can find a path between the trees--even at typical, normal water levels; not just during floods. 

Multi-channel patterns in low-gradient river reaches are usually described as anastamosing, and can generally be attributed to streams losing the ability to transport sediment as slopes decrease toward outer coastal plain and the coast. In such aggrading systems avulsions (channel shifts) occur, and often both the original and new channels remain, creating the anastamosing pattern. 

Images from the U.S. Geological Survey National Hydrography data set, which shows only channels, water bodies, and wetland areas. (A) is along the Horry/Marion County border, SC (coordinates at center 33.9133^o N, 79.2808^o W). (B) is the Black River in southeastern NC (center: 34.4064^o N, 78.1182^o W). (C) shows tributaries of the Lynches River, SC (33.8207^o N, 79.634^o W). Areas shown in A, B are about 16 km N-S by 10 km E-W; C is 4 X 10 km). In the Little Pee Dee and Black River sites there are many more channels (as well as floodplain lakes and flowing floodplains) than shown. That is probably also the case in Lake Swamp area, but I have not been there to confirm. 

But these blackwater rivers, where upland soils are mainly sandy, the watersheds are mostly forested, and the valley bottoms are largely swamp, carry very little suspended sediment (brownwater rivers such as the Great Pee Dee, Cape Fear, and Roanoke rise in the Piedmont or Appalachians and arrive in the lower coastal plain with enough suspended sediment to give the water a tan or brown color). While the blackwater rivers do move some sandy bedload, but they are not visibly aggrading. 

Google Earth^TM image showing the contrast between the brownwater Great Pee Dee and blackwater Little Pee Dee Rivers.

Another deviation from the normal anastamosing situation is the that avulsions usually begin as a crevasse, which is river flow breaking through the natural levee on the banktop. Often the water simply spreads out in the backswamp behind the natural levee, leaving a deposit called a crevasse splay. But sometimes the breakthrough flow remains concentrated and is able to erode a channel. If there is a slope advantage relative to the river (basically, some depression that is lower than the river level), a new channel can form. In the case of a relocation avulsion, the new channel becomes the main channel and the old one may gradually infill. If both the old and new channels persist and reconnect downstream, that's an anastamosing avulsion. I worked a lot on avulsions in Gulf Coastal Plain rivers back in the day; references are at the end.

A figure from Phillips, 2014 (see below) from the Neches River, Texas, where the standard model of avulsion and anastomosis based on an aggrading system and crevasses is applicable. 

In many cases the Little Pee Dee (for instance) has no levee and no real bank--even at typical, normal water levels there is a just a flooded transition from channel to swamp. No crevasses in that situation; water can overflow the channel almost everywhere. 

Channel edge of the Little Pee Dee near Conway, SC--no real bank; no levee. 

If the traditional anastamosing model does not fit these situations, how do the multichannel planforms arise?

We will start exploring this in Part 2. 

Published work by JDP on avulsions and multi-channel patterns in Gulf Coastal Plain Rivers:

Phillips, J.D., 2014. Anastamosing channels in the lower Neches River valley, Texas. Earth Surface Processes and Landforms 39:1888-1899 

Phillips, J.D. 2013. Hydrological connectivity of abandoned channel water bodies on a coastal plain river. River Research and Applications 29: 149-160. 

Phillips, J.D. 2012. Logjams and avulsions in the San Antonio River delta. Earth Surface Processes and Landforms 37: 936-950. 

Phillips, J.D. 2011. Universal and local controls of avulsions in southeast Texas rivers. Geomorphology 130: 17-28. 

Phillips, J.D. 2009. Avulsion regimes in southeast Texas rivers. Earth Surface Processes and Landforms 34: 75-87. 

STRAIGHTENING THE WACCAMAW

Look here. The upper Waccamaw River in South and North Carolina, the section not affected by tides, has reaches that are quite sinuous, interspersed with sections that are relatively straight, with no meanders.

 

Two reaches of the Waccamaw River near the North Carolina/South Carolina border near Longs, S.C.

Sometimes such differences are attributable to tectonic effects, with an abrupt transition from a strongly meandering channel on the steeper side of a fault or flexure, and a much straighter channel on the more gently sloping side. A river's power depends on its discharge (or in relative terms, velocity) and slope. When the slope is steeper the river may have more energy and power than it needs to transport whatever sediment is available, and (according to the laws of thermodynamics) that excess energy must be dissipated in some way. Where it is dissipated against the banks, there may be erosion and lateral channel migration. This frequently occurs in the form of bends and meanders, and the channel sinuosity increases. There are various explanations for how flow hydraulics get this done, but independently of that, many studies have shown that in many cases a meandering path is the most efficient way to dissipate energy. Of course, flowing water and stream channels don't care how (or even if) energy is dissipated, but selection processes operate in abiotic as well as biological systems, and tend to select for maximum efficiency. As the meandering pattern is often optimal in this respect, erosion and deposition patterns that create meandering are selected for--that is, once initiated they tend to be preserved and sometimes enhanced and repeated (like all selection, this is probabilistic and not deterministic). Science nerds can read all about it in my Abiotic Selection in Earth Surface Systems book (Springer Nature, 2025). 

 

Bald cypress trees along the Waccamaw River, N.C. 

In trying to explain this curious pattern, the first thing I looked at was slope, controlled either by tectonic features (these are present in the Carolinas coastal plain) or paleotopography inherited from the ancestral Waccamaw River, which appears to have carried more flow than the modern version. But no luck. I also looked for potential variations in erodibility of the bank materials--they exist (they always do), but do not correspond with the straight vs. meandering reaches. But recently Gerald Nanson (a geomorphologist and river scientist in Wollongong, Australia) and I were exchanging some thoughts via e-mail about river channel patterns. Gerald, along with He Qing Huang, has done some of the most innovative and iconoclastic work on this over the years. With respect to rivers in particular and science in general, Gerald and I partially agree on almost everything and totally agree on very little, which makes for interesting and productive exchanges.

I sent a few maps of the upper Waccamaw to him and described some of my observations while kayaking. He suggested that tributaries might be bringing in a large load of sand. This would take up some of the river's excess energy to transport it, thereby reducing the bank erosion necessary to produce meander bends, and making the straighter channel the more efficient way to dissipate energy. I was pretty sure this was not correct (and I am now more so), but Dr. Nanson has not had the opportunity I have to see those tributaries and the swampy, mostly forested watersheds that are not delivering much sand.

The Waccamaw is a blackwater river with little suspended sediment. Most sediment transport is as sand bedload, as seen here. 

But the basic idea that an extra input of sand could trigger a switch from a sinuous, winding path to a straighter one got me thinking about what could deliver a load of sand. The answer is river erosion of its valley walls, which are predominantly sandy soils. It turns out that the meandering sections are mainly winding through the middle of the swampy stream valley (unconfined, in fluvial jargon), while the straight sections occur where the stream is abutting a valley wall, with erosion and bank failures delivering sand. 

 

Eroding bank along the Waccamaw River valley wall.

 

The Waccamaw River is still migrating laterally against the valley walls. Note the eroding bank valley wall in the background and the sand deposition in the foreground. 

The figures below show the same areas as the first two, but with a background of relief maps (derived from the U.S. Geological Survey National Map system). You can see the correspondence quite clearly. 

 

 

Lower panel of bottom image shows another section of the Waccamaw River, upstream of Conway, S.C. on a contour map rather than shaded relief. 

This may also be an explanation of another curious feature of the area, that I call cypress flats. These are broad areas of cypress trees in standing water that do not correspond with tributary mouths or prograding meanders. At least some of them, when you thread the kayak into them and get to dry land, are adjacent to valley walls with concave morphology indicating that they were once eroding. If I am correct, their previous erosion could deliver sand to the river that could be colonized by bald cypress (Taxodium distichum). Whether cypress could get established would depend on delivery of floating cypress seeds to the deposited sand, and the sand remaining in place and above the waterline long enough for seedlings to establish. That would in turn depend on the timing and sequence of valley wall erosion and high- and low-water events. 

 

Cypress flat along the Waccamaw River

SACKETT SUCKS IT DRY

 The Supreme Court’s 2023 Sackett v. EPA decision ruled in favor of two landowners backfilling a lot containing wetlands. The decision changed the definition of the term “waters of the United States”—which is used in the Clean Water Act—to exclude wetlands without continuous surface connections to larger, navigable bodies of water. In November 2025, the Trump administration’s EPA proposed to set new rules for water regulations that may be even looser than the updated Sackett definition.

A recent scientific study found that the new rules will leave an area of wetlands unprotected that cumulatively equals the land area of Wisconsin! This terrible news for clean water, fish and wildlife habitats, and the economic and ecosystem services provided by swamps, marshes, bogs, pocosins, floodplains, and mires. 

Backwater swamp along the Northeast Cape Fear River

While the loosened definition might not result in direct dredging and filling of the deepwater swamps that I focus on here, it would allow dredging and filling to encroach ever closer to tupelo-cypress swamps and other wetlands, inflicting numerous adverse impacts. 

Sad and tragic. 

FACTORS OF SWAMP SPATIAL DISTRIBUTIONS

 Regular readers of Swamp Things (both of you) may recall a post a few months back on The Factors of Swamp Formation. In it I applied the factorial conceptual model first developed in soil science and later applied in many other fields (especially ecology) to swamps and other wetlands. The follow-up was just published in the journal Hydrology. I initially submitted it with the "factors of swamp formation" title to resonate with the well-known factors of soil formation concept. However, reviewers pointed out, correctly, that the work deals more with spatial patterns and geographical distributions than with formation. 

The abstract is below:

Abstract

A state factor model of bottomland hardwood swamp formation is applied to a lower coastal

plain river in North Carolina, U.S., to explain variations in wetland hydrological, ecological,

geomorphological, and soil characteristics. Swamps and wetlands are a function of the

interacting influences of the state factors of climate, topography, hydrology, vegetation,

fauna, soils, geomorphic setting, and time. Five classifications of swamp and related

environments were applied to the study area, with the categories present determined

based on fieldwork. For each classification, the implicit, embedded state factors were

identified from the classification scheme itself. Relevant environmental gradients for the

study area were identified, and a spatial adjacency graph for the study area was developed

for each classification. The ability of the environmental gradients to explain the spatial

complexity of the pattern was assessed using spatial adjacency graph (SAG) analysis. All

the classification criteria are associated with the proposed state factors. SAG analysis

shows overdetermination, indicating that known gradients of causal factors are sufficient to

explain the overall pattern of spatial contiguity and that single-factor models of change are

not sufficient at the local scale. Results confirm studies showing that responses to seilla-level

and other changes are spatially patchy.

The article is open access, and you can get it via the link embedded in the citation:

Phillips, J.D. 2025 The factors of swamp spatial distributions. Hydrology 12, 332. https://doi.org/10.3390/hydrology12120332

Below, just for the heck of it, a couple of pictures from a recent paddling expedition on some anabranches and lakes of the Little Pee River, South Carolina. Though it's a bit hard to pick out at first, the top one shows a swamp tupelo (Nyssa biflora) growing right up on the root crowns of two large baldcypress (Taxodium distichum). The bottom photo is one of many cool looking cypress trees in the area.