MYSTERIOUS WAYS & CONNECTION SELECTION

 Recently Oxford University Press published my first book written for a general readership, as opposed to a scientific research monograph. 

The description is below, and you can get it in hardcopy or E-book directly from the publisher here. You can also get it through Amazon, and a free chapter (till May 2026) here.

In one section of the book, I discuss the advantages of high connectivity in environmental systems. “Everything is connected to everything else” is called the First Law of geography, ecology, and environmental science, but why are things so highly connected? For the full answer, read the book. For an illustration of the advantages of high connectivity, I used the lower Waccamaw River, South Carolina—thus my excuse for plugging the book in the Swamp Things blog. 

Over a period of slightly less than three years, the lower Waccamaw River experienced the three highest flows ever recorded, during major floods in October 2015, October 2016, and September 2018. In 2015 an “atmospheric river” event pumped moisture from a tropical system well to the south and sent a firehose of wet air into South Carolina, causing extreme rainfall, runoff, and river flooding. Hurricanes Matthew and Florence in 2016 and 2018 included not only extreme river discharges but also storm surges from downstream estuaries. Irrespective of these large, high-energy flows (and in contrast to the severe impacts on humans and the built environment), ecological, geomorphological, and hydrological changes were minimal. The Waccamaw took a lickin’ and kept on tickin.’

 

Myrtle Beach Sun-News photo by Jason Lee of Waccamaw River flooding in Conway, SC on 9 October 2016. Two National Guard soldiers waded through flood waters going door to door to check on residents.  

How? Why? Because of the high connectivity among hydrogeomorphic components of the system.  Having better sense than to launch my kayak into the teeth of a hurricane, I did not directly observe what was happening in the river valley during the floods, but have made many field observations since. There is also ample aerial imagery—including some during the floods made by the U.S. Geological Survey and the National Oceanic and Atmospheric Administration specifically to assess storm and flood damage. These give a reasonable picture of what was going on at a broad scale, if not a feel for exactly how water flowed through swamps at a specific location. 

Google Earth image of part of the lower Waccamaw River.

Based on field observations and imagery, I demarcated the eight types of hydrogeomorphic elements as shown below. 

 

Hydrogeomorphic elements of the lower Waccamaw channel/wetland complex (copy of Table 8.2 from Mysterious Ways). 

Observed hydrologic connections and water exchanges among hydrogeomorphic elements of the lower Waccamaw River. Entries represent fluxes from the row to the column element (copy of Table 8.3 from Mysterious Ways).

Not many blank boxes, huh? There is a high degree of interconnectivity in this system, shown diagrammatically below. Almost everything is connected to almost everything else. The exchanges among elements are in all instances two-way, with the net direction of flux depending mainly on river stages and whether they are rising or falling, but also influenced by astronomical tides, local runoff, storm surges, and wind. This enabled the Waccamaw to absorb the flood and storm surge impacts by storing water and delaying flow through wetlands, activating spillways to transport excess water, reverse flows in some components, and conduct two-way exchanges of water during rising and falling river flows. 

Connectivity graph for hydrogeomorphic elements and water exchanges in the lower Waccamaw (Figure 8.6 from Mysterious Ways). 

The book gives some additional analysis, but beyond a demonstration of the benefits of connectivity, there is an important lesson about how vital it is to preserve the complex of swamps, back-channels, and other features of these lower coastal plain rivers. 

MEANDERS & DEFLECTED TRIBUTARIES

 Did you ever notice that along winding, meandering rivers, tributary streams almost never connect on the interior, point bar section of meander bends; always on outer bends or straight reaches? Basic principles of hydrology and fluvial geomorphology regarding flow dynamics in bends readily explain why it is difficult for tributaries to form on bend interiors, and why an inner-bend location is disadvantageous, and an outer-bend site is hydraulically advantageous for tributary junctions. But these generalizations largely apply to winding rivers with channel margin bars, often sandy or gravel, that are readily mobile, and banks that are not too difficult to erode. 

Junction of Ashes Creek and the Northeast Cape Fear River, at the apex of a river meander bend. The leaves coming out of the creek show the flow dominance by the river.

But those conditions are rarely present on bends of the swamp-flanked rivers of the lower coastal plain. Both banks, including point bars on bend interiors, are typically fully and densely vegetated right up to the river’s edge. The bars are also often composed on cohesive fine-grained sediments rather than more readily moved sand. These rivers have been called “vegetation bound” by some scientists due to their perceived lateral immobility. 

In addition, these lower river reaches are characterized by coastal backwater effects as astronomical tides, wind tides, and storm surges slow, block, and reverse downstream flow. In addition, banks are also very low, and sometimes nonexistent, with just a gradual transition from open water to trees standing in water to wet, frequently inundated floodplain. 

Lower Waccamaw River, S.C.

In a nutshell, these lower coastal plain swamp rivers are quite a bit different than most alluvial rivers, and some of the generalizations about tributary junctions and meander bends may not apply. For example, on a sandy alluvial river meanders often start with a channel margin bar along what will become the bend interior. If there is a tributary junction, and if the main stream flow is significantly stronger (as it generally is), the bar will deflect tributary flow downstream. As the bend develops, the tributary is deflected toward the downstream end of the bend. If there is an incoming tributary on the outer bend, by contrast, the erosion on the cutbank shortens its channel, thereby steepening it, and promoting its staying on the outer bend. Sandy, unvegetated point bars are rare in the lowermost river reaches (fluvial-estuarine transition or tidal freshwater zones). 

From my paddling these rivers, my impression was that the pattern (tributary junctions only on outer bends or straight reaches) generally holds in these systems, especially for larger tributaries, but you can see some small channels on the low, wet, bend interiors. 

So, with some additional fieldwork and GIS analysis, I examined 121 tributary junctions along the lower reaches of seven South and North Carolina Rivers, as well as a number of other river bends that did not appear, on maps and imagery, to have tributaries.

Indeed, the no-confluences-on-inner-beds pattern holds true in these environments. None of the 121 tributary junctions occurred on bend interiors, with about half on outer bends and half on straight reaches of the main stream. The small channels occasionally found on inner bends all turned out to be local distributary/tributary channels where water flowed into the floodplain bend interior during high water and back again as river stages fall. None extended beyond the local floodplain.

A reversing (distributary/tributary) channel on the lower Neuse River. 

I found 17 cases where a tributary approached the interior of a meander bend on the trunk stream and turned abruptly toward the downstream end of the bend. I investigated these as possibly deflected tributaries. In six of these cases evidence was insufficient to determine whether the tributary direction changed after the river bend began developing, or the potential causes of diversion. These were cases where human alterations of terrain or drainage had occurred, or where digital elevation model data or imagery were of poor quality. All 11 other cases showed strong evidence of diversion by a similar cause. As the trunk stream bend developed, the tributary continued to occupy the original, pre-bend channel rather than maintaining its connection to the laterally migrating main channel. Or, the tributary mouth apparently migrated with the trunk stream channel for a while, and then flow was captured by a pre-bend paleochannel of the river. In either case, the tributary reconnects to the river near the downstream end of the meander bend. 

Deflected tributary on the lower Cape Fear River. 

Confluence of Bigham Branch and the lower Pee Dee River. Image on left is slope map derived from digital elevation model data. 

In this case the creek extended as the Cape Fear River bend developed, and was then captured by a paleochannel. 

One key lesson is the importance of antecedent topography in these settings. On outer bends, the pre-bend topography is eroded away as the bend develops. On inner bends it can be preserved, and this is especially the case in the study area. The dense plant cover (and often fine, cohesive soils) slow down lateral migration of the river, but allow for preservation of paleochannel features. 

Another is the local nature of gradient selection. In many cases extension of the tributary across the growing bend interior would have allowed for a shorter, overall steeper path to the river. However, flowing water can’t “see” anything but its immediate surroundings, and the remnant river channels provided a steeper, deeper, easier path at the point where they were encountered. 

This work was recently published as the article referenced below, from which the figures above came (it is open access, so you can get it for no cost). The abstract is also below.

Phillips, J.D. 2025. River meanders, tributary junctions, and antecedent morphology. Hydrology 12, 101. https://doi.org/10.3390/hydrology12050101

SWAMP BEAVERS

 In 1897, all native beavers (Castor canadensis) in North Carolina were gone; hunted and trapped to local extinction. In 1939, the N.C. Department of Conservation and Development got 39 beavers from Pennsylvania and released them in the Sandhills area. In 1979, after reporting on several incidents of beavers blocking drainage canals in eastern North Carolina, I wrote a column for the small daily newspaper I worked for suggesting that this was Mother Nature’s revenge for the water quality, water chemistry, and habitat damage done by artificial drainage ditches and canals in the region (it was not well received by the local agribusiness and land development establishment). By the early 1990s, scientific articles were appearing about the recovery of beaver populations in the southeastern U.S. (1).

Beaver lodges on Pinetree Creek near Vanceboro, NC (top) and on the Black River in Bladen County, NC (bottom). 

The story in South Carolina is similar—gone due to trapping and hunting by the end of the 1800s. The U.S. Fish and Wildlife Service reintroduced beavers in S.C.’s Pee Dee region in 1940. Around the same time, Georgia beavers began to recolonize the Savannah River drainage along the GA/SC border. Both populations expanded their range, according to the S.C. Department of Natural Resources, moving toward each other. Beavers are now found in all 46 S.C. counties. 

Beavers are iconic in several ways in my corners of the scientific world. Through their dam building and pond construction, they are the prototype (other than Homo sapiens, of course) ecosystem engineer—0rganisms that modify the environment to suit their own needs and affect resources and habitats for other organisms as well. They are also a prototype for the concept of the extended phenotype. An organism’s genotype is its genetic makeup, and its phenotype is how the genes are expressed in its physiology. The extended phenotype is in essence how its genes are expressed in the external environment. Beavers are also prominent in recent and ongoing efforts to use nature-based solutions in stream and wetland restoration. 

Beavers are a significant part of swamp fauna, but though their impacts are known to be significant, we don’t know a whole lot. The Classification of the Natural Communities of North Carolina (2), in the section on coastal plain floodplain forests, states:

The most poorly known natural dynamic process of floodplains is that of beavers. Beavers can dam small stream channels or may impound tributary streams or sloughs within large floodplains. Beaver ponds can raise the local water table beyond the extent of standing water. A beaver dam on an outlet slough (gut) through a natural levee can impound a large area of a complex pattern determined by microtopography. Beavers have been returning to North Carolina for several decades, after a much longer absence . . . . Little is known about their natural population dynamics, predation, disease, nor about past pond longevity and return intervals. An important question for small streams is whether all parts of a stream are suitable for pond building, so that beaver ponds appear randomly and eventually affect the whole area, or if certain favored sites are chronically ponded while others never are. In large river floodplains, only specific sites can be flooded by beaver dams; the natural levees, high ridges, and some backswamps and sloughs are not susceptible (p. 600).

 

Answering those questions is a tall order. Many beaver ponds in swamps are remote; you could not access them in anything other than a canoe or kayak or waders. Remote sensing experts can detect beaver dams and ponds from imagery, as a 1996 study on the Roanoke River, NC did (3). However, I looked at Google Earth^TM images that spanned a period on Otter Creek in Craven County, NC before, during, and after the appearance and disappearance of a beaver dam and pond (in an area near my home where I paddle several times a year). I could not detect the emergence, presence, or loss of the feature from the images due to the complex pattern of flooding that occurs independently of any beaver ponds. 

Beaver dam and pond on upper Otter Creek, a tributary of the Neuse River estuary, The dam was not present in early 2017, and present by March, 2020. It was still there in 2022, and gone by late 2024. It is not known if the pond was abandoned, washed out by a flood, or removed by property owners. 

The classification (2) includes at least 14 different descriptions of ecological, hydrological, geomorphological, and soil transitions associated with either the establishment or drainage/abandonment of beaver ponds in the N.C. Coastal Plain alone (more are described in the Piedmont and Mountain provinces). 

 

A 2015 study of sediment trapping in beaver ponds included 4 coastal plain sites in Pitt County, NC and coastal plain sites in Virginia (4). On the coastal plain, they estimated that one pond per kilometer of stream length would result in 19 million cubic meters a year of deposition in the VA and NC coastal plains. They also noted that when beaver ponds are constructed on artificially channelized streams, they often restore the natural hydrological and ecological functions of the floodplain by reconnecting the channel and floodplain. 

 

Dam on a slough of the Black River in Bladen County, NC.

I think it likely that beavers are responsible for some of what I call water savannas or water woodland; permanently flooded floodplain depressions with mostly or entirely large, older bald cypress (Taxodium distichum) and tupelo (Nyssa biflora; N. aquatica) trees. Construction of a beaver dam floods a section of floodplain forest, and most of the trees die off (or are chewed up by beavers). Mature tupelo and cypress, however, can survive constant inundation, and while not excluded from an occasional beaver chomp (see below) are not favored. In the lower coastal plain settings where I’ve observed the water savannas, sediment inputs are very low, and abandoned beaver ponds fill in only very slowly, if at all. Tupelo and cypress remain in the ponded floodplain. This is just my proposed story, mind you, but it is plausible and fits the evidence of many of my field observations. 

 

A selection of water savannas & woodlands in the lower Neuse River drainage basin.

 

 

Beavers are not believed to chew on bald cypress, and I had rarely seen evidence of it before, but on Core Creek near Cove City, NC in January there was extensive evidence of chewing on cypress knees. 

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

 

(1) Butler, D.R. 1991. The reintroduction of the beaver into the South. Southeastern Geographer 31, 39-43. 

 

(2) Shafale, M. 2023. Classification of the Natural Communities of North Carolina. Fourth Approximation. Raleigh: N.C. Natural Heritage Program. 

(3) Townsend, P.A., Butler, D.R. 1996. Patterns of landscape use by beaver on the lower Roanoke River floodplain, North Carolina. Physical Geography 17, 253-269. 

(4) Kroes, D.E., Bason, C.W. 2015. Sediment-trapping by beaver ponds in streams of the mid-Atlantic Piedmont and Coastal Plain, USA. Southeastern Naturalist 14, 577-595.