| 27 Jul 2026
Storage and connectivity across river corridor and watershed scales in three headwater streams of the southeastern USA
Abstract. Watershed characteristics create a mosaic of potential storage zones, and linkages between these result in hydrologic connectivity between groundwaters and surface waters. Watershed storage and connectivity are fundamental controls on stream network dynamics, yet predicting the patterns of hydrologic responses like streamflow remains difficult due to landscape heterogeneity across multiple, interacting spatial scales. Here, we characterize storage and connectivity dynamics at the watershed scale and the hydrogeomorphic feature (HGF) scale using a network of steam and riparian groundwater monitoring wells in three watersheds spanning the Coastal Plain, Piedmont, and Appalachian Plateaus physiographic regions of the southeastern USA. At the watershed scale, we operationalized watershed storage as the slope of streamflow recession at the watershed outlet, and hydrologic connectivity as the relationship between outlet discharge and network length. At the HGF scale, we operationalized storage as the slope of water level recession at seven in-stream monitoring locations, and connectivity as the lateral hydraulic gradient and water level hysteresis relationships between the stream channel and adjacent riparian zone at three locations per watershed. At the watershed scale, we found patterns in storage and connectivity varied across physiographic regions. We also found distinct storage and connectivity dynamics across HGFs, many of which challenged existing conceptualizations for similar geomorphic settings. Notably, channel incision emerged as a key structural control, homogenizing storage and connectivity patterns across otherwise disparate watersheds and producing a common threshold response in watershed storage. Together, these results demonstrate that watershed-scale patterns do not reflect the heterogeneity observed at the HGF scale. Moreover, river corridor structure, particularly channel incision, is a key driver of storage and connectivity dynamics across physiographic settings.
This manuscript utilizes two years of instream stilling well and riparian well data to calculate metrics representing watershed and reach-scale (hydrogeomorphic feature, HGF) storage and connectivity. Connectivity is described at the network scale as longitudinal connectivity of the stream network. At the HGF scale they use the water table gradient between in-channel and riparian wells as a proxy for lateral connectivity. They conclude that there is inter and intra HGF variability in storage and connectivity that is not well represented with catchment metrics. They also suggest that incision is important for modulating storage and connectivity. This paper with major revisions could make a great contribution to the watershed hydrology community. It does need some editing, adjusting how HGFs are classified, and acknowledgement of data limitations for it to be an impactful paper that readers can follow.
The are many different terms and classifications used across this paper. As a result, it makes it hard to evaluate the validity of the claim that channel incision is an important factor in storage and connectivity across all watersheds. Only the coastal plain site has an HGF explicitly classified as incised. This is not to say that this conclusion isn’t correct or that there are not incised channels at all sites, but it is a prime example of the confusion created by using two different sets of classifiers for HGFs and mixing terminology. It is challenging to draw conclusions when each reach is in its own class. Overall, the authors should consider trying to simplify how they classify HGFs and consistently use a single descriptor term. For example, if incision is truly the most important factor and exists at all sites, can the HGFs be classified by degree of incision? The other geomorphic descriptors can be used to introduce the sites and describe HGFS but all subsequent analysis and plots could be presented using a singular more unifying descriptor.
On a related note, the authors describe a variety of storage zones, directions of connectivity, and ways to measure each in the introduction. They use a subset of those for their analysis but do not always use specific terms for what they are referring to. For example, “connectivity” is frequently used with no modifier so it’s unclear if they’re referring to network or longitudinal connectivity, lateral connectivity, vertical connectivity, hillslope-riparian-stream connectivity, etc. etc. Or “network extent dynamics” is used instead of “network connectivity”. This is also true for storage. To alleviate confusion and help readability of the mauscript, the authors should consider defining the terms for storage, connectivity, and metrics to describe them and use them consistently throughout.
Many of the papers the authors cite (Spence, Jencso and McGlynn, Tromp Van Meerveld and McDonnell, McNamara et al., and McGuire et al., etc.) focus on connectivity and storage from the hillslopes to the streams, or entirely in a hillslope: Tromp Van Meerveld and McDonnell. In this study, wells were installed only in the riparian area and stream and connectivity was defined based on the hysteresis index where more similar event responses in riparian and instream wells were considered well connected. They do not have data to determine how well connected the hillslopes are to the riparian areas and streams. It is however highly likely that hillslopes are influencing the connectivity observed here as well as the storage metrics. They themselves are a storage zone feeding these downgradient storge zones. The size and gradient of hillslopes likely also impact the differences across sites. Just looking at the maps, hillslopes vary in the amount of area they occupy relative to the HGFs across sites, i.e. there is varying drainage density. The authors can’t go get data on hillslopes but they should be able to more explicitly include them in their discussion and interpretation of results.
The watershed outlet is itself a single arbitrary point and HGF. The authors do acknowledge in the discussion that any instream measurement is both a function of local conditions and any upstream characteristics, but it is not fully integrated into how they interpret their recession metrics. Recession metrics at all sites are integrating all upstream storage, the outlet point does not uniquely integrate the watershed. This interpretation and discussion should be refined.
Minor Comments:
L13: network dynamics- Expansion and contraction or connectivity? Or streamflow? This sentence is confusing in general
L38: connectivity- defined and described in paragraph starting on line 60, but take a an extra clause or sentence to define here.
L103: The is paragraph could use some wrap-up. The intro ends rather abruptly.
L108: Each watershed… - delete this sentence. You repeat this and provide more details below. Here it sounds like this is all the information the reader will get on the instrumentation.
Figure 1: But the square symbol in the legend and/or move the legend closer to the map of AL.
Line 150: Storage in what zone and connectivity in what direction? It's listed out in the intro but lumped together again here.
Line 178: what is the difference between a cave-spring and spring discharge?
Section 2.4: The methods section is so long by this point, it’s hard to follow all the analysis and interpretation of metrics. Consider summarizing in a table and cutting down the text here and across all sections of the methods.
Line 237” groundwater-surface water..- this is vague. Would benefit from defining and using consistent terminology.
Line 240: extra comma
Line 249: how was 15 min data from the first year coarsened to hourly?
Line 268: Hydrologic state - a whole new term. frame within connectivity.
Line 289 and 290: “Surface water pools” - yet another term. use storage zone?
Results (or methods): Strongly recommend showing a time series of discharge or runoff and precipitation, possibly with the events highlighted. It’s helpful to get context and compare how each watershed responds to events and how flow changes seasonally.
3.1: “stream-riparian connectivity” – do you mean stream network connectivity here?
3.2: include “at HGF scale” in title to it’s clear what this section is about
Line 326: “are more likely to be driven by..”- interpretation, save for discussion
Line 329: “most likely driven by…” – also interpretation, move to discussion
Line 334: “It is worth noting..” - Not following what this sentence means. Is the order of magnitude comparable?
Figure 4: if this figure stays with all 8 different channel types, they should have labels indicating which watershed they are in. The reader is not as familiar with the channel types and watersheds as the author and needs some extra cues.
Line 348: “The only outlier…” - so is the outlet then representative of the whole watershed in the watershed level analyses?
Line 355: Stick with recession slope or bwl or some other more descriptive term defined in the methods and use it consistently.
Note on table 1: Should the S in LTMS be lower case? Also, when did you subset down to 8? are you not using all locations in each HGF in all analysis? If not, which are being used for what?
Line 384: “Further” - Editorial, but here and elsewhere some of these transition words are unnecessary
Line 384: “stream-riparian hysteresis” – another example of a place that could benefit from consistent terminology to keep things straight.
Line 391: “significant portion” – statistically significant?
Line 417: “quadrant” – you’re using a table, not a figure with quadrants. Recommend not using this term or converting the table to a figure (might be nice to see where they fall visually).
Line 432: observed is repeated.
Line 454: “re” is a typo?
Line 456: “ Taken together..”- This sentence is missing a key take home in the Jencso and McGlynn paper and may be using this citation incorrectly to make this point. The GTC/DFC metric is specific to the hillslopes or areas contributing to the streams in TCEF, not the valley bottom. There is also strong correlations between hillslope length and slope and underlying geology, where shorter lower slope hillslopes are underlain by sandstone, which strongly influences connectivity. There is a whole section of the discussion in that paper describing how geology influences connectivity, not just slope. This is an example place where the authors should adjust the discussion to include how previous work focuses on hillslopes and how in their study sites hillslopes may be influencing connectivity they observe between riparian wells and streams.
Section 4.2: They authors need to take connectivity of hillslopes or uplands into account. This is channel to riparian connectivity but riparian areas or wetlands should also receive contributions from upgradient hillslopes. Low relief watersheds/HGFs are more dominated by channel-riparian interactions than hillslope-riparian interactions. Especially if you bring drainage density into account across watersheds, there is just more area dominated by wetlands and riparian and upland hillslopes.
First paragraph of 4.2: this is where the issue of integrating everything upstream at a point and how that impacts interpretation of recession data is somewhat addressed. However, it should be addressed way earlier, in the methods. It should also actually be integrated in the interpretations of results, not just stated and then ignored in the subsequent paragraphs.
Section 4.3: see comment above, it is hard to evaluate this section when it is unclear how incised HGFs across watersheds actually are. Also, because it’s hard to follow what metrics are used for defining high or low connectivity and storage and what connectivity and storage is at each scale, this section is not particularly compelling.