Reply to comment on ‘Rivers of the Variscan Foreland: fluvial morphodynamics in the Pennant Formation of South Wales, UK’ Open Access

In Wood et al. (2022), we used a well-established framework for palaeohydraulic reconstruction (see further detail below) based on bedform analysis and grain size. This approach utilizes the advances in palaeohydrology which have been developed and tested in the five decades since the publication of the thesis of Jones (1977), to which the comment of Jones and Collinson repeatedly refers (e.g. Leclair and Bridge 2001; Trampush et al. 2014; Bradley and Venditti 2017; Mahon and McElroy 2018; Shibata et al. 2018; Ganti et al. 2019; Greenberg et al. 2021; Long 2021; Lyster et al. 2021, 2022b). We note, however, that the comment of Jones and Collinson first asserts that we neglected to consider sedimentary facies in our study of the Pennant Formation. It is important to stress that our work did not occur independently of sedimentary facies analysis in the field, and new facies descriptions are presented where appropriate (e.g. wood debris packages, Wood et al. 2022, fig. 11; see also McLeod et al. 2023). We did not repeat a compilation of lithofacies associations for the Pennant Formation in our work as this has been done in detail by previous authors, including the authors of the comment who provide detailed descriptions of the Lower Pennant Formation (Rhondda, Brithdir and Llynfi members; Jones 1977). Our motivation, instead, was to complement previous work with the use of palaeohydrological techniques across the whole Pennant Formation.

The locations outlined in previous works aided in outcrop selection for our work, although we note that changes in engineering and land management in South Wales now limit access to some localities. For example, the Earlswood Road exposure (Lat: 61.634°, Lon: −3.839°), which has previously been considered a type section of the Rhondda Member (e.g. Waters et al. 2007), is along an inaccessible stretch of the M4 Motorway that has seen extensive slope stability engineering (e.g. netting and concreting), rendering it unsuitable for data collection at present.

For this study, we sought to characterize the hydrology of the Pennant Formation across a maximum possible spatial and temporal extent. Therefore, we used field sites in both the South Wales and Pembrokeshire coalfields (Wood et al. 2022, fig. 1), and across all five members of the Pennant (Llynfi, Rhondda, Brithdir, Hughes and Swansea; Wood et al. 2022, fig. 2). We collected field data at 18 localities of varying exposure quality (Wood et al. 2022, fig. 4). This included recording a total of n = 2077 cross-set heights, n = 1038 palaeocurrent measurements and n = 116 architectural fluvial element thicknesses, including channel fill packages and lateral accretion packages. We are aware that previous work on the Pennant is dominated by observation in the Rhondda Member (e.g. Kelling 1968, 1974; Jones 1977; Jones and Hartley 1993) which outcrops frequently in the South Wales Coalfield, so we took care in this study to minimize this preservation bias in our results by sampling all members across the lateral extent of the Pennant Formation.

Our database expands the existing number of measurements on bedforms, palaeocurrents and architectural elements in this formation by two orders of magnitude. Consequently, we feel our database is suitably representative of the Pennant Sandstone across South Wales and across the c. 7 Myr depositional history of the unit. This allowed us to consider spatiotemporal trends in the deposition without placing undue significance on some of the best-preserved, or most frequently visited, exposures. Each outcrop selected for the study exhibits measurable cross-sets as well as channel architecture (e.g. barforms and channel packages) to ensure that cross-set-derived flow depths can be reasonably validated using previously applied techniques (e.g. Jones 1977; Jones and Hartley 1993).

We were of course interested to read in the comment about individual localities such as the ‘quarry above Cymmer’ or ‘Blaenrhondda Road’, which feature in the thesis of Jones (1977). While we have collected data in this region, the provided details are not sufficient to make a definitive comparison of specific cross-sets or architectural packages with our published data.

In their comment, Jones and Collinson (2025) express an overarching concern about the applicability of palaeohydrological methods, which seems independent to our specific application of these approaches to the Pennant Formation. Before responding to their specific concerns, we want to highlight again that the field of palaeohydrological reconstruction has matured and expanded markedly over recent decades (see the review of Long 2021), meaning that a large number of methods have now been developed and tested to reconstruct morphological and hydrodynamic parameters from sedimentary deposits. These include methods to reconstruct river morphology, which include dune heights, lengths and widths (Paola and Borgman 1991; Leclair and Bridge 2001; Bradley and Venditti 2017; Cardenas et al. 2023), channel depths (Hajek and Heller 2012; Bradley and Venditti 2017; Foreman et al. 2023; Alpheus and Hajek 2024), channel widths (Toonen et al. 2012; Greenberg et al. 2021; Long 2021), channel slopes (Paola and Mohrig 1996; Duller et al. 2010; Trampush et al. 2014; Lyster et al. 2021), and channel sinuosity and channel river planform (Parker 1976; van den Berg 1995; Ghosh 2000; Crosato and Mosselman 2009; Galeazzi et al. 2021; Lyster et al. 2022b). These methods also encompass multiple approaches to reconstruct channel hydrodynamics, including flow and sediment transport conditions (Dade and Friend 1998; Carling 1999; Ferguson and Church 2004; Wright and Parker 2005; Mahon and McElroy 2018).

Wood et al. (2022) is not unique in its application of these methods; sedimentary researchers regularly combine methods into multi-step workflows to tackle broad questions of landscape evolution (e.g. Holbrook and Wanas 2014; Lin and Bhattacharya 2017; Sharma et al. 2017; Ganti et al. 2019; Lyster et al. 2021; Hartley and Owen 2022). To date, this type of quantitative palaeohydrology has provided novel insights into pertinent research topics, including river response to climate change (e.g. Foreman et al. 2012, 2023; Colombera et al. 2017; Barefoot et al. 2021; McLeod et al. 2023; Owen et al. 2023), past hydroclimate (e.g. Lyster et al. 2023), pre-vegetation landscape dynamics (e.g. Ganti et al. 2019; Valenza et al. 2023), river dynamics on Mars (Wilson et al. 2004; Dietrich et al. 2017; Morgan and Craddock 2019; Hayden et al. 2021; Birch et al. 2023) and more. We firmly believe that the opportunities afforded by these methods extend beyond their associated uncertainties and limitations, which we describe in more detail in subsequent sections.

In the comment of Jones and Collinson (2025), the authors make a number of criticisms of our palaeohydrological approach. Unfortunately, we do not believe many of these criticisms are well founded. Here we aim to address the specific concerns that Jones and Collinson raise on the applied methodologies in turn.

Extensive work has demonstrated that bar thicknesses are effective palaeoflow depth proxies (Mohrig et al. 2000; Hajek and Heller 2012; Alpheus and Hajek 2024). In the absence of fully preserved bars we can use cross-bedding to estimate palaeoflow depths as cross-set thicknesses scale predictably with formative dune geometry (Paola and Borgman 1991; Bridge 1997; Leclair et al. 1997; Bridge and Tye 2000; Leclair and Bridge 2001; Leclair 2002), which, in turn, scale with formative flow depth (Yalin 1964; Akram 1971; Allen 1978; van Rijn 1984; Julien and Klaassen 1995; Jones et al. 2009; Bradley and Venditti 2017). These methods are not without limitation and the palaeohydrology community is aware of these limitations, with entire subfields of study devoted to reconciling them. This includes work to evaluate and adapt cross-set–dune height scaling relations for non-steady-state bedform preservation, which arises due to the presence of morphodynamic hierarchy (e.g. superimposed bedforms and barforms) and discharge conditions (e.g. transport stage, flow variability; Jerolmack and Mohrig 2005; Reesink et al. 2015; Myrow et al. 2018; Ganti et al. 2020; Leary and Ganti 2020; Das et al. 2022, 2024; Mahon et al. 2024). Other workers have aimed similarly to evaluate and adapt dune height–flow depth scaling relations to stratigraphy (Bradley and Venditti 2017, 2019; Reesink et al. 2018). With these limitations in mind, the palaeohydrological community often incorporates multiple methods to estimate palaeoflow depth (Lyster et al. 2021; Foreman et al. 2023) with careful propagation of uncertainty through workflows (e.g. Ganti et al. 2019; Lyster et al. 2021; Valenza et al. 2023) and consideration of how reconstructed palaeoflow depths may differ for scenarios of bedform preservation in non-steady-state conditions (Lyster et al. 2022a; McLeod et al. 2023).

In Wood et al. (2022), we explicitly compared multiple methods to estimate palaeoflow depths including barforms and channel fill measurements – we did not rely on cross-set–dune height and dune height–flow depth scaling relations alone. Where these scaling relations were used, we carefully propagated uncertainty via Monte Carlo methods, which included all relevant scaling relation bounds, while a companion study of McLeod et al. (2023) considered non-steady-state preservation scenarios. Considering established approaches to dealing with these uncertainties (e.g. Ganti et al. 2019; Long 2021; Lyster et al. 2021; van Yperen et al. 2024), and our own statistical approach to account for uncertainties, we are confident that our palaeoflow depth estimates are robust.

Our typical palaeoflow depth estimates of 2–3 m are discrepant with depths of >10 m inferred from channel fill packages by Jones (1977). We believe these discrepancies arise for three main reasons: (1) study area extent; (2) sampling bias; and (3) the extent to which depositional packages in amalgamated sands fully represent flow depths, as these are recognized by Jones and Collinson as the primary evidence for very deep rivers >10 m in the Pennant Formation. We address these reasons individually below.

Study area extent. We note that the work of Jones (1977) was spatially restricted to the Rhondda Valley and Earlswood Roundabout – areas that are dominated by the Rhondda Member with some exposures of the Llynfi and Brithdir members. Consequently, the channel package thicknesses they report do not reflect the whole stratigraphy of the Pennant. Nevertheless, we resolve that median flow depths of 2–3 m are consistent and stable across the whole Pennant including from our field sites in the Rhondda Valley (e.g. localities 2.1, 2.2, 2.3, 5.3 and 6.3 – fig. 1; Wood et al. 2022). To provide further insight into the flow depths resolved across the Pennant in the database of Wood et al. (2022), we provide a statistical breakdown in Table 1.

Alongside measuring cross-set heights to apply the scaling relationships outlined above of Leclair and Bridge (2001) and Bradley and Venditti (2017), we measured barform and channel fill package thicknesses, with the latter following similar criteria to that of Jones (1977). These provide reasonable architectural constraints on channel depth that were used to validate cross-set-derived flow depths. As we aimed for a conservative approach that reconciled these different observations, values of bedform-derived H that exceed 125% of the maximum measured package thickness were rejected in the analysis with a total of n = 89 cross-sets discounted for our representative reconstructions due to this (4% of total measured).

Across the Pennant Sandstone, we measured 27 channel fill packages (mean thickness of 2.9 m and maximum of 4.5 m) and 50 barform accretion packages (mean thickness of 2.2 m and a local maximum of 5 m; see supplementary material of Wood et al. (2022) for the full dataset). Examples of these packages are presented in figure 4 of the main text of that paper. Significantly, our measurements are consistent with observations made in Jones and Hartley (1993), which provide additional constraints on architectural fluvial elements across the Pennant Formation. Jones and Hartley (1993) report barforms of 1.4–4 m (reaching a local maximum of 7 m) across the lower Pennant (Llynfi and Rhondda members) while they report fining-upwards cycles, which the authors interpret as channel fill packages, with average thicknesses of 6.5 m in the Rhondda Member and 3.5 m in the Hughes Member. Their results describing barform heights do not differ significantly from those reported in Wood et al. (2022), and our measured packages and the implied flow depths from these agree well with values derived from our bedform approach (Table 1).

Having reiterated our understanding of the limitations when estimating palaeoflow depths, and our incorporation of multiple methods and uncertainty propagation, we are satisfied with our results. Even when considering upper bounds of palaeoflow depths, i.e. the uppermost estimates derived from uncertainty propagation, our mean reconstructed flow depth of 2.31 m (Table 1) would equate to a flow depth of 4.32 m – still far shallower than the 15 m flow depths resolved by Jones (1977).

Sampling bias. As Jones and Collinson explicitly state about the Pennant in their comment, ‘most trough cross-bed sets are fairly thin’, in comparison to an isolated 0.75 m thick set they observed in the Rhondda Member above Cymmer. This is in complete agreement with the dataset we collected across the formation. Anomalously large cross-sets are present in the Pennant Sandstone – for instance we documented a maximum cross-set height of 1.49 m in the Hughes Member in a quarry above Abercynon (Lat: 51.658°, Lon: −3.331°). However, across our n = 2077 measured cross-set maxima, our mean and median cross-set heights are 0.19 and 0.18 m, respectively.

Jones and Collison dismissed our reconstructed palaeoflow depths based on criticism of our methods. However, when applying the approach of Wood et al. (2022) to an anomalously large cross-set measured in Jones (1977), they were then content with the method's efficacy and palaeoflow depth estimate:

In a quarry above Cymmer, a highly truncated single trough set in the Rhondda Beds is 0.75 m thick parallel to the flow direction. Applying Wood et al.'s methodology to estimate dune height and their equation to estimate flow depth (Bradley and Venditti 2017) for such a set gives a depth of 10.2 m, a value consistent with depth estimates based on measurements of preserved channel fills.

This speaks to a larger issue of appropriate sample size and sampling bias, in proportion to the chosen research question, when observing and interpreting Pennant river deposits. We see the merit in Jones and Collinson's attempts to locate the very deepest channels that may have existed during the c. 7 Myr timescale of Pennant deposition. Axial Pennant channels do provide useful insights for palaeohydrology, especially for interpreting catchment-scale conditions. However, our goal was to characterize the full distribution of palaeohydrological conditions in Pennant rivers across space and time, and to capture their natural variability. As a result, we did not restrict field data collection to the largest cross-sets and barforms. With our interest in the entire system, it would be misleading if we suggested that anomalously large cross-sets, such as the 1.49 m set above Abercynon or the 0.75 m set above Cymmer, are most representative of the hydrodynamics of the Pennant rivers. With our broader geographical region, and data collection across all bedform and barform sizes, our wide-ranging findings for Pennant rivers are not directly comparable to the maximum depths of individual channels the work of Jones (1977) resolve, and we do not expect or anticipate agreement between findings.

Amalgamated sand packages. Finally, it is widely accepted that it can be challenging to pick out channel fill packages effectively in amalgamated sandy systems. Indeed, it is stated throughout the comment of Jones and Collinson that individual channel elements are often difficult to identify within the Pennant Sandstone. We completely agree – trying to deduce flow depths from bulk multilateral, multistorey channel architectures is hard, as there is a risk that composite forms are misidentified and depths overstated. We note the range of barform heights in Jones and Hartley (1993) for the Pennant, which are good markers for flow depth when fully preserved, are also somewhat smaller than the channel fill packages they report. It is therefore puzzling that Jones and Collinson would reject out of hand palaeohydraulic methodologies that are grounded in physics (Leclair and Bridge 2001; Trampush et al. 2014; Bradley and Venditti 2017) and shown to be applicable to sand-bed fluvial sediments (e.g. Ganti et al. 2019; Lyster et al. 2021) as credible methods for resolving conditions beyond the best-preserved exposures.

Median grain size (D₅₀) is a key parameter in reconstructions of fluvial systems in the geological record (e.g. Fedele and Paola 2007; Duller et al. 2010; Allen et al. 2013; Long 2021; Lyster et al. 2021; Valenza et al. 2023). In their comment, Jones and Collinson suggest we extrapolated a single grain size measurement across the whole Pennant Sandstone. We did not do this, and their suggestion of this misrepresents our methods. We direct the reader to the sentence in Wood et al. (2022) which states: ‘Where the height distribution of an individual cross-set was measured, the median grain size, D50, was also estimated using the Wentworth grain-size classification.’ We classified D₅₀ for every measured cross-set distribution, and our dataset describes over 2000 individual cross-sets. Further, we strongly reject the assertion that D₅₀ is ‘irrelevant’ to channel slope in modern or ancient sand-bedded rivers. It is important for readers that we explain clearly that this statement by Jones and Collinson is incorrect. It is extremely well documented that grain size influences slope in sand-bedded rivers, such as those that deposited the bulk of the Pennant Sandstone, as well as in gravel-bedded rivers (e.g. Dade and Friend 1998; Wright and Parker 2005; Sklar 2024).

In Wood et al. (2022) we apply the equation of Trampush et al. (2014) to reconstruct palaeoslope. This is a theoretically rooted but empirically derived equation that relates channel depth and channel bed grain size to local slope. Jones and Collinson commented that ‘no defensible reconstruction of S is available from the field data’ based on their dismissal of the role of grain size. This comment is based on a complete misappropriation of the work of Trampush et al. (2014). Jones and Collinson included a quotation from Trampush et al. to highlight the uncertainty of the applied palaeoslope equation. However, the quote is taken directly from the ‘Theory’ section of the paper, in which Trampush et al. (2014) described the uncertainty associated with using the widely known Shield's stress equation (equation 1 in Trampush et al. 2014). While Shield's stress-based approaches continue to be used in this field with success (e.g. Paola and Mohrig 1996; Duller et al. 2010; Lyster et al. 2021), we did not apply this equation in our reconstructions of the Pennant. We instead used the equation first outlined by Trampush et al. (2014), which was specifically derived to mitigate the uncertainties associated with a Shield's stress approach (equation 5 in Trampush et al. 2014; equation 3 in Wood et al. 2022).

The work of Trampush et al. (2014), contrary to the assertion of Jones and Collinson, explicitly shows the universal applicability of this empirical relationship across rivers of all transport modes, including sand bed rivers such as those recorded in the Pennant Formation. This was the key aim of their work. Their study shows predictable scaling of slope across a D₅₀ range of 0.01–222 mm and is an extremely well-utilized approach for calculating slope from sand-bedded fluvial stratigraphy (e.g. Chen et al. 2018; Mahon and McElroy 2018; Ganti et al. 2019; Long 2021; Lyster et al. 2021; Wu et al. 2022; McLeod et al. 2023; van Yperen et al. 2024).

Besides specifics of the palaeoslope equation, Jones and Collinson also criticize our application of this equation, and what they have interpreted to be our assumption that slope, calculated on the basis of local parameters, can be extrapolated along the length of the river. This is another unfortunate misrepresentation of our methods. We calculated palaeoslope at individual outcrops, and these estimates reflect local, reach-scale palaeoslope – we did not extrapolate slope estimates along the lengths of Pennant rivers. For each member of the Pennant Sandstone, we compiled our site-specific palaeoslope estimates (multiple estimates per site) and presented the grouped data (fig. 6 in Wood et al. 2022) to highlight the range of values reconstructed for each member, across the entire study area. Lowland rivers typically have slopes of order 10⁻⁴ to 10⁻⁵, and compiling these data allowed us to compare Pennant rivers, overall, with modern rivers. We then plotted our palaeoslope estimates in space (fig. 8 in Wood et al. 2022), trying to observe potential upstream to downstream decline in palaeoslope, which is reasonable to expect. We were not extrapolating our locality-scale slope estimate as a complete representation of entire systems.

Channel planform is principally governed by factors such as sediment supply, sediment grain size, channel stability, channel slope and more (Schumm 1985; Ferguson 1986; Heritage et al. 2001; Church 2006; Church and Rice 2009). In turn, reconstructing channel planform from sedimentary deposits provides insights into these governing factors and ancient river behaviour. Researchers often use planform interpretations to infer river response to climate change (e.g. Wang et al. 2024), and to characterize ancient landscapes (e.g. Ganti et al. 2019; Valenza et al. 2023). However, channel planform can be difficult to interpret from sedimentary deposits (Hartley et al. 2015, 2018; McMahon and Davies 2018), and researchers increasingly use hydraulic planform predictors, or planform stability fields, to predict planform (e.g. Henderson Francis 1963; Leopold and Wolman 1970; Parker 1976; van den Berg 1995; Eaton et al. 2010; Kleinhans and van den Berg 2011; Lyster et al. 2022b).

In Wood et al. (2022), we used stability fields originally proposed by Parker (1976), and refined by Lyster et al. (2022b), to support planform interpretation. We agree with the comment that interpreting planform from stratigraphy alone can be challenging, which is why we supplemented our facies-based observations (e.g. presence of well-defined lateral accretion surfaces) with this planform stability field analysis. In addition, we used multiple approaches to estimating channel width, including using the H:W scaling relationship of Jones (1977), to present a range of plausible planform solutions. Jones (1977) suggests flow depths in the Pennant of >10 m occur in a sand-bed, braided environment. We agree that the Pennant preserves sand-bed, multi-thread rivers. However, we highlight that not all multi-thread rivers are braided, and our field observations and planform stability reconstructions suggest the Pennant preserves multi-thread anastomosing rivers.

We chose to use the predictor of Lyster et al. (2022b) for two key reasons. First, this predictor aimed to resolve sampling biases associated with previous planform predictors (particularly the bias towards gravel-bed braided river data) and was derived from observations of 1688 rivers worldwide including 630 sand-bed examples (Lyster et al. 2022b). Second, this predictor allowed for discrimination between multi-thread planform types, with braided rivers and anastomosing/anabranching rivers treated as endmember multi-thread planforms (Lyster et al. 2022b). The compilation of modern rivers in Lyster et al. (2022b) shows sand-bed braided rivers reach maximum flow depths of 6 m at their largest examples. Considering modern anastomosing planforms, rivers with flow depths >10 m include downstream reaches of the Amazon, Yangtze and Brahmaputra (Lyster et al. 2022b and references therein). Across all sand-bed, multi-thread rivers in the compilation of Lyster et al. (2022b), a mean width of 1800 m is resolved. We did not observe evidence in the Pennant that would imply widths this great or in particular flow depths ∼2× greater than the largest braided rivers currently on Earth.

We are, however, confident in our interpretation that the Pennant Sandstone broadly reflects anastomosing multi-thread rivers with distinct single threaded reaches. The interpretation that the Pennant preserves this style of rivers is consistent with their sandy beds, low slopes and well-developed floodplains (e.g. coals and palaeosols), which are consistent with observations of anastomosing multi-thread rivers in modern swamps/wetlands (Gradziński et al. 2003; Wójcicki 2023; Zieliński 2024).

Our view is that if we do not attempt to progress sedimentological science beyond facies analysis and ‘developing a feel for the sizes of channels’, as advocated by Jones and Collinson, our understanding of these stratigraphic packages that record key climatic and tectonic intervals in Earth's history will remain incomplete.

Further to the scientific questions, which we hope are suitably answered here and for which we welcome further discussion, we feel that an important point in our response must address the manner in which Jones and Collinson construct and phrase their discussion of our study. The strength of language that Jones and Collinson used, which includes referring to our work as ‘naïve’, ‘bizarre’, ‘damagingly serious’, ‘reckless’, ‘not defensible’ and ‘crude’, would be considered by many to be unacceptable in modern scientific discourse. It is important to be challenged on published work and we strongly defend the right of Jones and Collinson to offer commentary on our paper. We were, however, disappointed to receive a comment with such a disparaging and dismissive tone. This is not constructive, and we direct readers towards a number of recent studies which have evaluated the effect of overly harsh and critical commentary and review on the scientific community as a whole, and on individuals (Silbiger and Stubler 2019; Mavrogenis et al. 2020; Bharti et al. 2024; Feinmann 2024).

We were extremely grateful to have had access to a physical copy of the thesis of Jones (1977), which is a substantive scientific contribution, and which significantly shaped the early stages of our Pennant research. We were especially surprised, given the lack of attention paid to this important sediment routing system over the last 50 years, to read that further work on this topic is seemingly not welcomed at all by those who had previously studied the formation. This is particularly true because Jones and Collinson provide little in the way of concrete evidence and supporting references to substantiate the scientific criticisms made in their comment, as we have described throughout. For instance, all references cited in their comment and not previously included in Wood et al. (2022) are from over 40 years ago – we do not want to detract from the value of these works, but likewise, it would be wrong to not acknowledge that fluvial sedimentology has advanced significantly over that time. Where more recent work is cited in the comment of Jones and Collinson (e.g. Trampush et al. 2014), it is misquoted. Our work was conducted by a group including early-career and established researchers, and while there must always be room in science for critique of methods and approaches, we cannot think of anything in our original paper that could remotely be considered ‘damaging’ or ‘reckless’.

In Wood et al. (2022), we showed the rivers preserved in the Carboniferous Pennant Formation of South Wales reflect stable anastomosing multi-thread rivers with distinct single-threaded reaches. Flow depth reconstructions, from both bedform scaling approaches and measurements of architectural channel elements, suggest that flow depths of 2–3 m are representative of the Pennant across its spatial extent and depositional history, with local palaeoslopes of 4–5  ×  10⁻⁴ (0.02–0.03°). While it is certainly possible that individual channels may have had depths greater than 2–3 m during the c. 7 Myr of deposition, the suggestion by Jones and Collinson (2025) that Pennant rivers were both braided and had flow depths of up to 15 m would imply hydrodynamic conditions incommensurate with all current rivers on Earth, as well as being at odds with the physics-based analysis of bedform and barform sizes we have made.

The body of work in the field of palaeohydrology continues to grow (e.g. Ganti et al. 2019, 2024; Leary and Ganti 2020; Straub et al. 2020; Greenberg et al. 2021; Long 2021; Arévalo et al. 2022; Lyster et al. 2022a, b, 2023; McLeod et al. 2023; Valenza et al. 2023; Colombera et al. 2024; Das et al. 2024; Mahon et al. 2024; Sharma et al. 2024; van Yperen et al. 2024; Shibata et al. 2025) and, as an active area of research, we are optimistic that these methods will continue to develop, uncertainties will be reduced, and the application of these approaches will facilitate deepened understanding of important sedimentary systems in conjunction with field- and outcrop-based studies and facies models. With this in mind, we were therefore disappointed to see that Jones and Collinson failed to acknowledge any of the transformative work done in the field of sedimentology, particularly over the past two decades. We warmly encourage further work on the Pennant Formation to tackle problems including linking field-sites in space and time to confidently identify downstream trends between isolated outcrops; resolving questions of discharge variability and flooding in the Pennant (cf. McLeod et al. 2023); and more closely relating the spatiotemporal evolution of the Pennant rivers to the development of the Variscan foreland basin and northward propagation of the Variscan Front.

The authors thank G. Hampson for useful discussions and the reviewers of Wood et al. (2022) for their helpful comments on the original study.

JW: conceptualization (equal), data curation (equal), investigation (equal), methodology (equal), project administration (equal), validation (equal), writing – original draft (equal), writing – review & editing (equal); JSM: formal analysis (supporting), investigation (supporting), methodology (supporting), resources (supporting), software (supporting), writing – original draft (supporting), writing – review & editing (supporting); SJL: formal analysis (equal), investigation (equal), methodology (equal), project administration (equal), supervision (supporting), writing – original draft (equal), writing – review & editing (equal); ACW: conceptualization (lead), funding acquisition (equal), investigation (equal), methodology (equal), project administration (equal), resources (equal), supervision (lead), writing – original draft (equal), writing – review & editing (equal)

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data sharing is not applicable to this article as no datasets were generated or analysed during the current study.