Introduction
The steady rise in global sea level following the end of the last glacial maximum (LGM) reflects the decay of the world’s great ice sheets (Peltier and Fairbanks, Reference Peltier and Fairbanks2006), of which the Laurentide was the largest (Peltier, Reference Peltier2004). Yet, increasing numbers of well-dated moraine records from around the globe reveal that terrestrial ice masses melted rapidly during the last glacial termination (Termination 1), with a considerable proportion of deglaciation coinciding with Heinrich Stadial 1 (HS1: ca. 18.2–14.7 ka) (Denton et al., Reference Denton, Lowell, Heusser, Schlüchter, Andersen, Heusser, Moreno and Marchant1999, Reference Denton, Putnam, Russell, Barrell, Schaefer, Kaplan and Strand2021; Linge et al., Reference Linge, Olsen, Brook, Darter, Mickelson, Raisbeck and Yiou2007; Hall et al., Reference Hall, Porter, Denton, Lowell and Bromley2013, Reference Hall, Lowell, Bromley, Putnam and Alan2025; Putnam et al., Reference Putnam, Schaefer, Denton, Barrell, Andersen, Koffman and Rowan2013, Reference Putnam, Denton and Schaefer2023; Moreno et al., Reference Moreno, Denton, Moreno, Lowell, Putnam and Kaplan2015; Bromley et al., Reference Bromley, Schaefer, Hall, Rademaker, Putnam, Todd, Hegland, Winckler, Jackson and Strand2016; Wilson et al., Reference Wilson, Ballantyne, Benetti, Small, Fabel and Clark2019; Foreman et al., Reference Foreman, Bromley, Hall and Jackson2022, Reference Foreman, Bromley, Hall and Rodríguez2025; Strand et al., Reference Strand, Putnam, Sambuu, Putnam, Denton, Schaefer and Radue2022). Heinrich Stadial 1 did not result simply in widespread cryospheric loss, but broadly as shifts in global wind patterns (Moreno et al., Reference Moreno, Villa-Martinez, Cardenas and Sagredo2012; Pesce and Moreno, Reference Pesce and Moreno2014; Denton et al., Reference Denton, Putnam, Russell, Barrell, Schaefer, Kaplan and Strand2021) and hydroclimate (Dupont et al., Reference Dupont, Schlütz, Ewah, Jennerjahn, Paul and Behling2010; Martin et al., Reference Martin, Blard, Lavé, Condom, Prémaillon, Jomelli and Brunstein2018; Du et al., Reference Du, Russell, Liu, Otto-Bliesner, Oppo, Mohtadi and Zhu2023), and, potentially, reorganization of North Atlantic Ocean circulation (McManus et al., Reference McManus, Francois, Gherardi, Keigwin and Brown-Leger2004). In their recent synthesis of stadial climate, Denton et al. (Reference Denton, Toucanne, Putnam, Barrell and Russell2022) suggested that this apparently unified glacial response was driven by sustained global warming during HS1.
This scenario of stadial deglaciation contradicts the traditional view of North Atlantic climate, in which stadial climate is characterized by extreme year-round cooling (Martin et al., Reference Martin, Buizert, Edwards, Kalk, Riddell-Young, Brook, Beaudette, Severinghaus and Sowers2023). Founded on paleoecological reconstructions from adjacent landmasses (Guiot et al., Reference Guiot, de Beaulieu, Cheddadi, David, Ponel and Reille1993; Sánchez Goñi et al., Reference Sánchez Goñi, Turon, Eynaud and Gendreau2000) and the original interpretation of ice core and marine isotope data (Grootes et al., Reference Grootes, Stuiver, White, Johnsen and Jouzel1993; Alley et al., Reference Alley, Brook and Anandakrishnan2002; McManus et al., Reference McManus, Francois, Gherardi, Keigwin and Brown-Leger2004; NGRIP Members, 2004), this model of severe stadial cooling of northern middle and high latitudes is often linked to freshwater-driven perturbations of Atlantic meridional overturning circulation (AMOC), and coincident with Southern Hemisphere warming and rising concentrations of atmospheric greenhouse gases (Monnin et al., Reference Monnin, Indermühle, Dällenbach, Flückiger, Stauffer, Stocker, Raynaud and Barnola2001; EPICA Community Members, 2006). To reconcile these apparently conflicting scenarios, recent studies have proposed that Northern Hemisphere stadial cooling is strongly biased towards winter months (Björck et al., Reference Björck, Bennike, Rosen, Andresen, Bohncke, Kaas and Conley2002; Denton et al., Reference Denton, Alley, Comer and Broecker2005), with summertime warmth and melting of land-based ice arising from increasing global temperatures (Denton et al., Reference Denton, Toucanne, Putnam, Barrell and Russell2022; Toucanne et al., Reference Toucanne, Vázquez Riveiros, Soulet, Blard, Migeon, Rigalleau and Roubi2026), sustaining a long-term perturbation of the thermohaline circulation in the northern North Atlantic. If true, thermal seasonality anomalies in the circum-North Atlantic are potentially an intrinsic characteristic of stadial climates.
Viewed in greater detail, the overall retreat of land-based ice during HS1 was punctuated by prominent yet short-lived moraine-building episodes, suggesting that remnant ice masses were briefly in states of quasi-equilibrium with climate. Well-dated examples of intra-HS1 moraine-building events include the Prospect Hill advance in the Southern Alps (ca. 16.3 ka: Putnam et al., Reference Putnam, Schaefer, Denton, Barrell, Andersen, Koffman and Rowan2013), the Veluyoccocha advance in the eastern Peruvian Andes (ca. 16.2 ka: Bromley et al., Reference Bromley, Schaefer, Hall, Rademaker, Putnam, Todd, Hegland, Winckler, Jackson and Strand2016), the Wester Ross Readvance in western Scotland (>15.8 cal ka BP: Simms et al., Reference Simms, Best, Shennan, Bradley, Small, Bustamante, Lightowler, Osleger and Sefton2022), and the Killard Point readvance in eastern Ireland (ca. 16.4–17.3 cal ka BP: McCabe et al., Reference McCabe, Clark, Clark and Dunlop2007). A similar event was reported recently from the Maumturk mountains in western Ireland (Foreman et al., Reference Foreman, Bromley, Hall and Rodríguez2025), where rapid post-LGM thinning and retreat of the Connemara ice center was interrupted by a brief readvance approximately 16.7 ± 0.6 ka (Figs. 1 and S3).
Location of the study site in its (A) broader island of Ireland and (B) regional Connemara contexts. (C) Position of the WCMS and KT esker relative to the major bedrock lithologies and Lough Corrib (LC), Ross Lake (RL), and Ballyquirke Lough (BL). Also shown are the locations of previous 10Be data discussed in the text: BD = Baile Doite (Foreman et al., Reference Foreman, Bromley, Hall and Jackson2022); MY = Moycullen (Roberts et al., Reference Roberts, Ó Cofaigh, Ballantyne, Burke, Chiverrell, Evans and Clark2020); and MT = Maumturk Mountains (Foreman et al., Reference Foreman, Bromley, Hall and Rodríguez2025). 10-m digital-elevation model used with permission from Tailte Éireann.
Figure 1 Long description
A) Map of Ireland with a highlighted area in the Connemara region. B) Detailed map showing Galway Bay and the position of MT (Maumturk Mountains) marked with a blue dot. C) Geological map of Connemara showing various bedrock lithologies: Carboniferous Limestone, Galway Granite, Metamorphic Complex and WCMS. KT esker is marked in pink, with RL (Ross Lake), BL (Ballyquirke Lough) and LC (Lough Corrib) labeled. BD (Baile Doite) is marked with a yellow dot and MY (Moycullen) with a green dot. The map includes a scale bar indicating distances in kilometers.
To better understand the glaciologic and climatic significance of these short-lived yet prominent reversals in stadial conditions, it is necessary to constrain both the timing and geographic extent of intra-stadial moraine systems. Here, we describe a laterally extensive end moraine system located in the western Lough Corrib, western Ireland, which we term the West Corrib moraine system (WCMS) (Fig. 1). Specifically, we report the geomorphological characteristics of the moraines, including a sedimentological analysis of constituent tills, and their age as inferred from cosmogenic beryllium-10 (10Be) surface-exposure dating. We then discuss the glaciologic relevance of the WCMS in the context of previous work in this area (Roberts et al., Reference Roberts, Ó Cofaigh, Ballantyne, Burke, Chiverrell, Evans and Clark2020; Foreman et al. Reference Foreman, Bromley, Hall and Jackson2022, Reference Foreman, Bromley, Hall and Rodríguez2025), as well as the broader climatic implications of this event for our understanding of sub-millennial climate shifts during HS1.
Geographic and geomorphologic setting
The WCMS is located on the western periphery of the Lough Corrib basin in County Galway (Fig. 1), where it forms a discontinuous suite of moraine ridges and mounds that can be traced for approximately 10 km from the village of Rosscahill in the northwest to Ballyquirke Lough in the southeast. Acknowledging that previous mapping by Charlesworth (Reference Charlesworth1928, Reference Charlesworth1929) described glacial deposits throughout the Lough Corrib basin, we note that his surveys did not extend southward to cover our study area and thus, to our knowledge, this study constitutes the first scientific reporting of the WCMS landforms. Underlain by Carboniferous limestone, the Corrib basin is bounded to the west and southwest by the Galway granite uplands and to the northwest by Dalradian metamorphic lithologies (Fig. 1), and while the WCMS itself is positioned entirely upon the limestone lowlands, the horizontal distance between the moraines and outcrops of granitic and metamorphic rock types is generally less than a kilometer. The moraine complex can be broken down into three sections, which are described in detail in the results: a northern section adjacent to Ross Lake; a middle section centered on Killarainey; and a southern section adjacent to Ballyquirke Lough. Chronologically, the WCMS is bracketed by two published 10Be datasets: east of the moraines, exposure ages from granite erratics perched on limestone bedrock at Baile Doite (Fig. S1) were reported by Foreman et al. (Reference Foreman, Bromley, Hall and Jackson2022), while Roberts et al. (Reference Roberts, Ó Cofaigh, Ballantyne, Burke, Chiverrell, Evans and Clark2020) reported 10Be ages for boulders on the granitic uplands west of the moraines (Fig. 1).
Adjacent to the northern section of the WCMS, the Killanin-Tullokyne (KT) esker extends >5.3 km east from Killanin (Fig. 1). This prominent landform has been mapped and described by multiple works over the last century (Sollas, Reference Sollas1896; Gregory, Reference Gregory1921; Synge, Reference Synge1979; Hennessy et al., Reference Hennessy, Carlin, Cunniffe, Feeney and Gormally2010; Feely, Reference Feely2024) and is generally associated with eastward drainage of glacial meltwater out of Connemara. We revisited the KT esker as part of this study to provide broader sedimentologic and geomorphic context for the WCMS.
Methodology
Glacial-geomorphic mapping
We conducted ground surveys between 2024 and 2025 to establish the extent and configuration of this moraine system, during which we covered each landform on foot and mapped the orientation and relief of the principal ridges and described the physical characteristics of sedimentary exposures in the moraines. These data were then plotted on a 10-m-resolution digital elevation model (DEM) provided by Tailte Éireann in QGIS (3.40.7) to build a regional glacial-geomorphic map of the WCMS. Moraine profiles were reconstructed in two ways: at Killarainey, we capitalized on the availability of LIDAR imagery to measure profiles in QGIS via the profile tool; at Rosscahill and Ballyquirke, for which LIDAR imagery is not available, we made ground surveys across the moraines with a Trimble GNSS.
10Be surface-exposure dating
We established the age of the WCMS using 10Be surface-exposure dating. Specifically, we collected ∼1 kg of rock from the upper surfaces of five granite boulders embedded in the crest of the Rosscahill moraine. Targeted boulders exhibited no apparent indication of excessive surface weathering or post-depositional movement, and were of sufficient relief (>0.5 m) to have minimized any effect of shielding by, for example, snow cover or vegetation. As part of this study, we also collected a further sample from a granite erratic at Baile Doite (Fig. S1) to expand the dataset published originally by Foreman et al. (Reference Foreman, Bromley, Hall and Jackson2022). Photographs of target boulders are provided in Figure 2. For each sample, we collected position and elevation data with a Trimble GNSS and quantified topographic shielding with a clinometer and the University of Washington online shielding calculator (http://stoneage.ice-d.org/math/skyline/skyline_in.html).
Photographs of erratic boulders on the Rosscahill moraine crest that were sampled for cosmogenic 10Be surface-exposure dating. The site is currently forested, and all boulders are mantled with a cover of moss.
Samples were crushed and pulverized in the Palaeoenvironmental Research Unit, University of Galway, after which we extracted quartz from the 250- to 850-μm size fraction and purified the separates with successive leaching in HF/HNO3 (Kohl and Nishiizumi, Reference Kohl and Nishiizumi1992). All samples were spiked with a custom-made phenakite carrier, with concentrations of 280 ± 10 ppm (Phena 2D), 368 ± 10 ppm (Phena 4), and 940 ± 10 ppm (Phena 7), and dissolved in concentrated HF. Carrier concentrations were measured by ICP-OES repeatedly over time to quantify evaporation effects. We used ion chromatography to isolate beryllium in the University of Maine Cosmogenic Isotope Laboratory, whereafter BeO targets were measured at the Center for Accelerator Mass Spectrometry (CAMS), Lawrence Livermore National Laboratory, relative to the 07KNSTD3110 standard (10Be/9Be = 2.85 × 10−12). Three procedural blanks gave 10Be concentrations ranging from 3970 to 8790 10Be atoms per blank (mean = 6633 ± 2450 10Be atoms), equivalent to <1% of the 10Be concentration of a typical sample measurement. All sample metadata and beryllium data are given in Table 1.
10Be sample details and nuclide data for Rosscahill and Baile Doite samples.
Table 1 Long description
Columns list landform, sample ID, CAMS ID, latitude, longitude, elevation in metres, sample thickness in centimetres, density in grams per cubic centimetre, shielding, quartz weight in grams, 9Be added in micrograms, 10Be to 9Be with plus or minus 1 sigma in units of 10 to the minus 14, and 10Be concentration with plus or minus 1 sigma in units of 10 to the 4 atoms per gram quartz. Rosscahill moraine includes four samples, RSC 24 01, RSC 24 02, RSC 24 03, and RSC 24 05, at latitudes 53.38132 to 53.38429 and longitudes minus 9.22755 to minus 9.22806, with elevations 18 to 28 meters. For these Rosscahill samples, thickness ranges from 1.3 to 3.0 centimeters, density is 2.7, shielding is 0.999, quartz weight ranges from 14.4948 to 15.4466 grams, and 9Be added ranges from 0.245 to 0.256 micrograms. Rosscahill 10Be to 9Be values range from 5.51 plus or minus 0.02 to 6.37 plus or minus 0.02, and 10Be concentrations range from 6.45 plus or minus 0.02 to 6.85 plus or minus 0.02. Baile Doite erratic includes one sample, ROX 23 01, CAMS ID BE53958, at latitude 53.36225 and longitude minus 9.17919, elevation 21 meters, thickness 3.2 centimeters, density 2.7, shielding 0.999, quartz weight 15.2362 grams, and 9Be added 0.240 micrograms. For ROX 23 01, 10Be to 9Be is 6.37 plus or minus 0.02 and 10Be concentration is 6.65 plus or minus 0.02.
Measurements were made relative to the 07KNSTD AMS standard, which is dilution 01-5-4 from Nishiizumi et al. (Reference Nishiizumi, Imamura, Caffey, Southon, Finkel and McAninch2007) with a known ratio of 2850 × 10−15; shielding corrections were calculated with the UW topographic shielding calculator (https://stoneage.ice-d.org/math/skyline/skyline_in.html); sample ROX-23-01 was processed with procedural blank B161 (280 ppm: 6633 ± 2450 10Be atoms); samples RSC-24-01 and 02 were processed with B171 (368 ppm: 7140 ± 1840 10Be atoms); samples RSC-24-03 and 05 were processed with B189 (940 ppm: 3970 ± 1620 10Be atoms)
a Beryllium ratios reported with no blank correction
b Sample concentrations reported with blank correction
To calculate surface-exposure ages from the measured 10Be/9Be ratios, we used version 3 (v.3) of the University of Washington (UW) online calculator (https://hess.ess.washington.edu) in conjunction with the Rannoch Moor (RM) production rate (Putnam et al., Reference Putnam, Bromley, Rademaker and Schaefer2019). This production rate was calibrated against landforms of ages constrained by 14C dating and is appropriate for the low-elevation, mid-latitude site in our study. The UW calculator generates exposure ages according to three current scaling schemes (time-independent ‘St’, time-dependent ‘Lm’, and Lifton-Sato-Dunai ‘LSDn’ scaling), which are given in Table 2. Since St scaling gave the closest agreement between 10Be and 14C data at the calibration site (Putnam et al., Reference Putnam, Bromley, Rademaker and Schaefer2019), we report apparent exposure ages in section 4 according to that scheme. Nonetheless, we highlight that the output from all three schemes agrees within 1%, confirming that our interpretations are not affected by choice of scaling.
Cosmogenic 10Be surface-exposure ages and internal [external] uncertainties calculated with the Rannoch Moor production rate (Putnam et al., Reference Putnam, Bromley, Rademaker and Schaefer2019) and the three current scaling models used in v3. of the UW online calculator.
Table 2 Long description
Cosmogenic beryllium-10 surface exposure ages are listed in ka for each sample, with values reported for three columns: St, Lm, and LSDn, each given with internal uncertainty and an external uncertainty. Rosscahill moraine includes samples RSC 24 01, RSC 24 02, RSC 24 03, and RSC 24 05, with St values of 16.9, 16.1, 16.7, and 16.3 ka, and corresponding Lm values of 16.9, 16.2, 16.8, and 16.4 ka, and LSDn values of 16.8, 16.1, 16.6, and 16.2 ka. Baile Doite includes ROX 23 01, ROX 20 01a, ROX 20 03a, and ROX 20 04a, with St values of 16.8, 17.6, 17.7, and 16.8 ka; Lm values of 16.8, 17.7, 17.8, and 16.9 ka; and LSDn values of 16.7, 17.6, 17.7, and 16.8 ka. Moycullen includes T5MOY01b through T5MOY05b, with St values of 16.1, 16.0, 16.7, 17.5, and 18.1 ka; Lm values of 16.2, 16.0, 16.7, 17.6, and 18.1 ka; and LSDn values of 16.2, 16.0, 16.7, 17.5, and 18.1 ka. Across rows, St, Lm, and LSDn values are close for each sample, typically differing by up to 0.2 ka. Internal uncertainties range from 0.2 to 0.7 ka, and external uncertainties range from 0.5 to 0.9 ka.
a Denotes age published previously by Foreman et al. (Reference Foreman, Bromley, Hall and Jackson2022)
b Denotes recalculated age from Roberts et al. (Reference Roberts, Ó Cofaigh, Ballantyne, Burke, Chiverrell, Evans and Clark2020)
Individual exposure ages are reported with both internal and external uncertainties as generated by the UW calculator (Table 2). For age populations, we report the mean age of samples from a specific surface (e.g., moraine crest) along with the standard deviation (SD), which estimates the probability with which one might predict the age of a single boulder on a landform. We then propagated the SD with the 2.7% production rate uncertainty (Putnam et al., Reference Putnam, Bromley, Rademaker and Schaefer2019) to visualize the effect of that systematic uncertainty on the chronology. Although all sampled surfaces show evidence of post-depositional weathering, we did not correct for erosion in our calculations because the rate is likely to have varied significantly among target boulders. However, we note that applying an erosion correction of 1 mm/ka would generally serve to make our reported ages slightly (∼1%) older. Finally, to facilitate comparison between the new WCMS ages and the previously published 10Be data of Roberts et al. (Reference Roberts, Ó Cofaigh, Ballantyne, Burke, Chiverrell, Evans and Clark2020), we recalculated the latter using the same production rate and scaling as for our own data. 10Be data reported by Foreman et al. (Reference Foreman, Bromley, Hall and Jackson2022, Reference Foreman, Bromley, Hall and Rodríguez2025) were also calculated in this manner. All input data for the UW calculator are provided for recalculation purposes in Table S1.
Sedimentological characterization
Moraine sedimentology—specifically, clast lithology and morphology—affords information on both the transport history of glacially derived material and the glaciology of the respective ice mass. We collected clasts from six exposures along the WCMS (Rosscahill: RC1–RC3; Ross Lake: RL; Killarainey: KR1–2; Ballyquirke: BQ) for characterization according to current glacio-geomorphological indices. We also collected two samples from the KT esker (KT1–2) for comparison against the moraine samples.
For all clast samples, we employed the C40 index to quantify differences between blocky and platy clast forms (Benn and Ballantyne, Reference Benn and Ballantyne1993, Reference Benn and Ballantyne1994). Defined as the percentage of clasts with c/a axial ratio ≤ 0.4, the C40 index provides a first-order indication of transport time and/or erosion intensity within the glacial system. Low C40 values correspond to blocky forms, suggesting either longer transport periods or a higher degree of abrasion by temperate ice, whereas higher C40 values correspond to platier clast forms, which generally reflect shorter transport periods, supraglacial transport, minimally erosive (i.e., cold-based) glacial regimes, or a combination thereof. We derived C40 index values for each exposure by measuring each (n = 50) clast with a caliper on three axes: longest (a), middle (b), and shortest (c). No individual axis exceeded 90 mm or was <10 mm; these same 50 clasts per site were used for the following two analyses.
Clast angularity/roundness is a widely used indicator for the intensity of abrasion experienced by glacial material during transport and emplacement. Clasts collected from landforms such as moraines are grouped into six categories, ranging from Very Angular to Well Rounded. Highly angular clasts (VA – very angular; A – angular) tend to represent transport via predominantly supra- or englacial pathways, which afford limited opportunity for clast modification. Such forms are most common in cold-based regimes (Tonkin et al., Reference Tonkin, Midgley, Graham and Labadz2017) and those glaciers receiving high proportions of debris from talus and rockfalls (Benn and Ballantyne, Reference Benn and Ballantyne1993). Conversely, rounded (R) and well-rounded (WR) forms generally indicate the entrainment of fluvial sediments and/or intense subglacial abrasion and modification of material in temperate or polythermal regimes (Lukas et al., Reference Lukas, Benn, Boston, Brook, Coray, Evans and Signer2013). The proportion of VA and A clasts is represented by the RA value, while the proportion of R and WR clasts is represented by the RWR value. Between these two extremes, subangular (SA) and subrounded (SR) forms are most common in clasts that have undergone active transport and modification via subglacial erosion (Benn, Reference Benn, Evans and Benn2014).
Lithology also has been shown to influence clast morphology, with some trending naturally to different forms and often at different rates, independent of glacial regime (Lukas et al., Reference Lukas, Benn, Boston, Brook, Coray, Evans and Signer2013). To account for any such effect, we used an optical microscope on freshly broken clast faces to verify the lithology of each clast measured. This information is also central to establishing the original source of moraine and esker material, and thus estimating former ice-flow direction. Having established clast lithologies for the seven exposures, we compared those data to the Geological Survey of Ireland 1:100,000 bedrock shapefile (https://data.gov.ie/dataset/bedrock-geology-1100000-ireland-roi-itm?package_type=dataset).
Results
Morphology of the WCMS
Altogether, the WCMS spans a distance of almost 10 km between Rosscahill and Ballyquirke (Fig. 3). In each of the three sectors outlined in Figure 3, the morphology comprises 1–3 sub-parallel ridges exhibiting well-defined crests and abundant glacially molded surficial boulders. At the southeastern end of the WCMS, the Ballyquirke moraines form an approximately 2-km-long grouping of forested till ridges, each between 200 and 300 m in length and as much as nine meters in relief above the surrounding bog and fields. Several moraines in this group terminate in Ballyquirke Lough itself, where they form prominent features on the lake’s western and southern shores. Our cross-moraine transect at this site (Fig. 4) describes a roughly symmetrical slope form (Fig. 5). Two small bulldozer excavations north of the lake, including the BQ clast sample site (Fig. 4), revealed a predominantly clast-supported diamicton containing abundant angular limestone boulders and occasional granite boulders interspersed with pebble- to cobble-sized material (Fig. 4).
(Left) Spatial distribution of 10Be surface-exposure ages from the Rosscahill moraines relative to published 10Be exposure ages from Baile Doite (Foreman et al., Reference Foreman, Bromley, Hall and Jackson2022; this study) and Moycullen (recalculated from Roberts et al., Reference Roberts, Ó Cofaigh, Ballantyne, Burke, Chiverrell, Evans and Clark2020). Both Baile Doite and Moycullen comprise populations of closely spaced erratic boulders represented by the single location markers above; details of Rosscahill sample locations are given the section closeup (Figure 4). The three main principal moraine sections are indicated by white rectangles: A = Rosscahill; B = Killarainey; C = Ballyquirke. Also shown is the KT esker (red). (Right) Camel plots for each of the three datasets. Thin black lines represent individual 10Be measurements as Gaussian curves; thick red lines represent each population’s summed probability. The mean age and 1σ uncertainty are denoted by vertical blue lines and yellow shading, respectively.
Figure 3 Long description
Map with three labeled locations and a scale bar. Text labels near the markers read, Rosscahill 16.5 plus or minus 0.4 ka, Baile Doite 17.2 plus or minus 0.5 ka and Moycullen 16.9 plus or minus 0.9 ka. A north arrow marked N appears on the left. A scale bar at the lower left shows 0, 1.5 and 3 kilometer. The map includes additional place text, Lough Corrib. Three white rectangles labeled A, B and C mark areas on the map and a small inset map appears inside the rectangle labeled C. Three stacked distribution plots, each with multiple thin curves and one thicker curve. The horizontal axis is labeled, AGE (ka), with tick labels 10, 12, 14, 16, 18, 20 and 22. The vertical axis represents relative probablity. The top plot is labeled, Rosscahill 16.5 plus or minus 0.4 ka and shows a main peak centered near 16.5. The middle plot is labeled, Baile Doite 17.2 plus or minus 0.5 ka and shows a main peak centered near 17.2. The bottom plot is labeled, Moycullen 16.9 plus or minus 0.9 ka and shows a broader main peak centered near 16.9. Each plot includes a vertical line at the stated mean value and a shaded band around that line.
(A) Ballyquirke moraines (box C in Figure 3) at the southern end of the WCMS, with the relative locations of the profile transect, till exposure (yellow circle), and BQ clast sampling site (green circle) indicated. Bulldozer excavations in the Ballyquirke moraine ridge revealed a clast-supported diamicton dominated by limestone clasts (B), but including abundant boulders of Galway granite (C) and coarse-grained quartz sand. (D) Rosscahill moraines at the northern end of the WCMS (box A in Figure 3), showing the locations of the till section, clast-sampling sites RC1–3 and RL, the profile transect, and boulders sampled for 10Be surface-exposure dating (blue circles). (E) Northward view of the principal Rosscahill moraine showing the landform at its widest point, where samples RSC-24-01 and RSC-24-02 were collected. (F) A narrower section of the same moraine, with the proximal and distal slopes visible on the left and right, respectively.
Detailed map showing part of the northern Killarainey (KR) moraines, based on Geological Survey Ireland LIDAR imagery, showing the locations of three (A–A′, B–B′, C–C) transverse moraine profiles generated using the profile tool in QGIS (3.40.7). The location of this map in the context of the wider Killarainey moraine section is denoted by the dashed white box in Figure 6A. Also shown are the transverse profiles for the Ballyquirke (BQ) and Rosscahill (RC) moraine ridges, locations of which can be found in Figure 4 (boxes A and D), created via Trimble GNSS ground survey.
Figure 5 Long description
The map shows part of the northern Killarainey moraines with highlighted areas and transverse profiles labeled A to A, B to B and C to C. These profiles are displayed on the left, showing elevation in meters over distance in meters. The profiles illustrate the shape of the KR moraines. Additional profiles for the Ballyquirke (BQ) and Rosscahill (RC) moraines are shown below, labeled W to E and SW to NE, respectively. The map includes a scale bar indicating 100 meters and an arrow pointing north. The highlighted areas represent the moraines′ locations, with the profiles providing cross-sectional views of their structure.
Between the village of Moycullen and Ross Lake, the Killarainey sector (Fig. 6) exhibits the most laterally continuous and best-exposed moraines of the WCMS (Fig. 3). Accounting for gaps due to non-deposition, erosion, and human activity, the primary ridge at Killarainey is almost 2.5 km long and as much as ∼10 m tall, ranging in width from 50 to 250 m. Three transverse profiles across the most prominent moraine segments revealed relatively uniform steep slopes and narrow crests (Fig. 5). As shown in Figures 6 and S2, man-made excavations in the Killarainey moraines revealed a predominantly clast-supported diamicton containing numerous large (as much as 2 m diameter) boulders of granite, gneiss, and limestone lithologies, as well as extensive units of tightly packed pebble- to cobble-sized clasts interspersed with coarse-grained quartz-rich sand. As at Ballyquirke, there is no evidence in the Killarainey exposures of a significant waterlain component, such as laminated sand, silt, or clay units.
Part of the Killarainey section of the WCMS (box B in Figure 3) showing (A) locations of cross-moraine profiles, KR1 and KR2 clast sampling sites (green circles), till exposure (yellow circle), and location of the LIDAR imagery from Figure 5 (dashed white box). (B) Close-up view of the sedimentary exposure depicted in panel (C), revealing a clast-supported diamicton. (C) Part of the main Killarainey moraine ridge viewed from the north, showing the north-facing exposure depicted in panel (B); inferred ice-flow direction was from right (west) to left (east). (D) The same moraine ridge as in (C), viewed from the west.
Centered on Ross Lake, the northernmost sector includes the Rosscahill moraines (Fig. 3), the most prominent of which is a 12-m-tall, ∼125-m-wide ridge that is traceable for 800 m through Rosscahill Wood and, at its northern extremity, appears to overlie the western limit of the KT esker. The principal Rosscahill moraine is accompanied by numerous lower-relief (3–10 m tall), NW–SE-oriented ridges located throughout the Ross Lake basin (Fig. 3). Owing to the effect of dense vegetation cover on satellite reception, we conducted only one transect across the Rosscahill moraine. This transect runs roughly SW to NE (see Fig. 4) across a relatively broad section of the moraine ridge and revealed an asymmetrical form, the southwestern slope being steeper than its northeastern counterpart (Fig. 5). North of the transect, however, the moraine ridge becomes narrower and taller and exhibits slopes of roughly equal gradient (Fig. 4F). While the absence of bulldozed excavations in the Rosscahill landforms precludes a detailed view of the moraine interiors, a forestry roadcut in a small (<2-m relief) moraine ridge immediately west of the primary moraine revealed a bouldery clast-supported diamicton dominated by limestone material. This roadcut is also clast sample site RC1 (see below).
East of the Rosscahill moraines, the KT esker comprises a sinuous ridge that rises as much as 8 m above the surrounding farmland and peat bogs. Classified as a short-bead esker representing continuous subglacial tunnel fill (Geological Survey Ireland, 2014), the landform’s undulating topography and the occurrence of at least one subaqueous fan system (Ower fan) suggest the esker reflects successive open- and closed-tunnel settings (Knight, Reference Knight2019). While a detailed examination of the esker morphology is beyond the scope of this paper, multiple exposures in northern and southern flanks of the esker reveal a diverse sedimentological composition, including a boulder-rich basal diamict, thick (>1 m) units of pebble- and cobble-rich sediments, sub-meter-thick sand layers, and stratified units of silts, sands, and gravels (Hennessy et al., Reference Hennessy, Carlin, Cunniffe, Feeney and Gormally2010; Fig. S2). Recent transverse excavations reveal evidence for post-depositional slumping of material down the esker’s southern flank.
10Be surface-exposure chronology
The four boulders on the principal Rosscahill moraine crest yielded 10Be surface-exposure ages ranging from 16.1 ± 0.6 ka to 16.9 ± 0.4 ka, and together represent a normal population with a mean age of 16.5 ± 0.4 ka (Fig. 3). Completing the Baile Doite dataset, sample ROX-23-01 gives an exposure age of 16.8 ± 0.5 ka. Together with the three previously published ages from this site (Foreman et al., Reference Foreman, Bromley, Hall and Jackson2022), the Baile Doite dataset exhibits a consistent yet slightly bimodal distribution, with a mean age of 17.2 ± 0.5 ka (Fig. 3). Finally, the recalculated Moycullen dataset of Roberts et al. (Reference Roberts, Ó Cofaigh, Ballantyne, Burke, Chiverrell, Evans and Clark2020) yields ages spanning 16.0 ± 0.7 to 18.1 ± 0.6 ka, with a mean age of 16.9 ± 0.9 ka (Fig. 3).
Sedimentology
Clasts obtained from the moraine and esker exposures represent a mixture of local rock types, including the underlying limestone and various igneous, metamorphic, and sedimentary lithologies cropping out west and north-west of the moraine complex (Table 3; Fig. 7). Three exposures (RC1, RC3, RL) are dominated by a light-gray limestone with visible calcite crystals (Table 3; Fig. 7), whereas the RC2, KR1, and KR2 samples are primarily metamorphic and igneous in composition. At the southern end of the WCMS, the BQ exposure contains predominantly limestone material, but with a significant (28%) component of granitic material. We also note that the limestone clasts in the RC2, KR1, and BQ exposures are generally darker and more fossiliferous than those observed in RC1, RC3, and RL. The quartzite within RC2, KR2, BQ, and KT 1-2 was of a notably gritty type, which is most closely associated with the Lakes Marble formation in the Connemara region, in contrast to other, purer quartzite formations in the area.
Clast lithologies and proportions for the moraine and esker exposures, along with the approximate positions of likely parent bedrock sources.
Table 3 Long description
Clast lithology counts are listed for nine exposure sites, RC1, RC2, RC3, KT1, KT2, RL, KR1, KR2, and BQ, alongside a column describing bedrock outcropping position relative to the exposure sites. Limestone counts are RC1 49, RC2 9, RC3 48, KT1 32, KT2 15, RL 50, KR1 4, KR2 9, and BQ 34, with bedrock noted as Underlying and East. Granitic Gneiss, Orthogneiss is RC2 14, KT2 8, KR1 28, and KR2 5, with bedrock noted as All West. Granite, Galway Granite and Oughterard Granite is RC1 1, RC2 9, RC3 1, KT1 7, KT2 3, RL 0, KR1 12, KR2 24, and BQ 14, with bedrock noted as All West. Gritty Quartzite, Lakes Marble is RC2 3, KT1 6, KT2 7, KR2 8, and BQ 2, with bedrock noted as Northwest. Sandstone Gneiss, Paragneiss is RC2 7, RC3 1, KT1 2, KT2 13, and KR2 2, with bedrock noted as West and Northwest. Diorite, Galway Granite is RC2 5 and zero at the other sites, with bedrock noted as West; Hornfel, Hornfels is RC2 1 and KT2 1 with bedrock noted as West; Schist, Lakes Marble is RC2 2 and KT1 3 with bedrock noted as Northwest; Pure Quartz, Undifferentiated is KT2 2, KR1 5, and KR2 2 with bedrock noted as All West; Dolerite, Undifferentiated is KT2 1 and KR1 1 with bedrock noted as All West.
Clast lithologies for the seven moraine exposures (RC1–RC3, RL, KR1, KR2, BQ) and from the esker exposures (KT1, KT2). Material falls into 10 broad lithological groups, eight of which are traceable to the approximate bedrock source area. Exposures of quartz and dolerite veins are common throughout Connemara, therefore clasts of those materials are not traceable to specific sources. Any regions without assigned bedrock units indicate lithologies that were not represented in clast sampling. Digital bedrock units sourced from the Geological Survey of Ireland 1:100,000 shapefile (https://data.gov.ie/dataset/bedrock-geology-1100000-ireland-roi-itm?package_type=dataset).
Figure 7 Long description
The map depicts clast lithologies in the Connemara region, with north oriented at the top. It includes a legend identifying rock types: limestone, orthogneiss, granite, schist (Lakes Marble), quartzite (Lakes Marble), paragneiss, undifferentiated quartz, dolerite and hornfels. Pie charts at locations RC1, RC2, RC3, KT1, KT2, BQ, RL, KR1 and KR2 show lithological composition percentages. Limestone dominates RC1, RC3, RL and BQ. Orthogneiss is significant in RC2 and KR1, while granite is notable in KT1, KT2 and KR2. The map features a scale bar up to 5 km and a north arrow. The spatial pattern shows limestone covering large areas, with other lithologies more localized. The pie charts summarize clast percentages, providing insight into the geological composition at each site.
Of the seven moraine exposures, RC2 exhibits the most varied composition; source areas for this material lie to the northwest, west, and southwest of the exposure (Fig. 7). In contrast, RL is the least varied, being composed entirely of the local limestone (Fig. 7). One dolerite clast and five undifferentiated pure quartz clasts occur within the KR1 exposure. Given that dolerite dikes are common features of the Galway granite, while quartz is ubiquitous throughout Connemara’s metasedimentary and igneous provinces, both lithologies suggest sources to the WNW of the WCMS.
While all seven moraine exposures returned low C40 values (none >16%), our angularity analysis revealed a clear division that correlates well with lithological composition. The first group (RC1, RC3, and RL) is characterized by high proportions of limestone clasts (≥ 96%), which exhibit relatively high angularity (high RA) values (Fig. 8). In contrast, the second group (RC2, KR1, KR2, BQ) contains higher concentrations of metamorphic and igneous lithologies, which generally exhibit lower RA index values. The samples within the second group include RWR clasts, as well as a higher proportion of subangular clasts (Fig. 9), all above 50% except KR2, which instead has increased subrounded clasts (26%) rather than increased angular clasts. No site had noticeably larger or smaller clasts than any other, and angularity did not correlate with size.
Ternary diagrams and clast form counts for the three moraine exposures (n = 50 per exposure) with higher angularity. All exposures returned low C40 index numbers typical of subglacial till, according to Sneed and Folk (Reference Sneed and Folk1958) and Benn and Ballantyne (Reference Benn and Ballantyne1993, Reference Benn and Ballantyne1994). Created using TRI-PLOT v 1.4.2. RA = proportion of VA (very angular) and A (angular) clasts; RWR = proportion of R (rounded) and WR (well-rounded) clasts.
Figure 8 Long description
Three ternary plots labeled Rosscahill 1, Rosscahill 3 and Ross Lake. Each ternary plot contains a triangular outline with internal grid lines and multiple plotted points. Each ternary plot includes the text, C40 equals 0, RA equals 28, RWR equals 0 for Rosscahill 1; C40 equals 7, RA equals 35, RWR equals 0 for Rosscahill 3; and C40 equals 0, RA equals 36, RWR equals 0 for Ross Lake. Each ternary plot shows the ratio of c to a as the left side of the triangle, the ratio of b to a as the right side of the triangle, and a minus b over a minus c as the bottom of the triangle. Three bar graphs labeled Rosscahill 1, Rosscahill 3 and Ross Lake. Each bar graph has the vertical axis label, Number, with tick marks from 0 to 50. Each bar graph has category labels along the horizontal axis: VA, A, SA, SR, R, WR. In Rosscahill 1, the bars are at VA 0, A 45, SA 30, SR 5, R 0, WR 0. In Rosscahill 3, the bars are at VA 0, A 35, SA 15, SR 3, R 0, WR 0. In Ross Lake, the bars are at VA 0, A 35, SA 15, SR 0, R 0, WR 0.
Ternary diagrams and clast form counts for the four moraine exposures (n = 50 per exposure) with lower angularity. All exposures returned low C40 index numbers typical of subglacial till, according to Sneed and Folk (Reference Sneed and Folk1958) and Benn and Ballantyne (Reference Benn and Ballantyne1993, Reference Benn and Ballantyne1994). Created using TRI-PLOT v 1.4.2. RA = proportion of VA (very angular) and A (angular) clasts; RWR = proportion of R (rounded) and WR (well-rounded) clasts.
Figure 9 Long description
A ternary scatter plot labeled Rosscahill 2 with scattered points and a thick horizontal band across the triangle. Text reads, C40 equals 8, RA equals 13, RWR equals 1. All four scatter plots are labeled with the ratio of c to a as the left side of the triangle, the ratio of b to a as the right side of the triangle, and a minus b over a minus c as the bottom of the triangle. The plotted points are concentrated in the middle portion of the triangle, with fewer points near the corners. The image B showing a ternary scatter plot labeled Ballyquirke with scattered points and a thick horizontal band across the triangle. Text reads, C40 equals 1, RA equals 19, RWR equals 1. The plotted points are concentrated in the middle portion of the triangle, with fewer points near the corners. The image C showing a bar chart labeled Rosscahill 2. The x axis categories read, VA, A, SA, SR, R, WR. The y axis label reads, Number, with tick labels 0, 5, 10, 15, 20, 25, 30, 35. Bar heights: A equals 15, SA equals 35, SR equals 2, R equals 1. The highest bar is SA at 35 and the smallest nonzero bar is R at 1. The image D showing a bar chart labeled Ballyquirke. The x axis categories read, VA, A, SA, SR, R, WR. The y axis label reads, Number, with tick labels 0, 5, 10, 15, 20, 25, 30. Bar heights: A equals 20, SA equals 25, SR equals 5, R equals 1. The highest bar is SA at 25, followed by A at 20. The image E showing a ternary scatter plot labeled Killarainey 1 with scattered points and a thick horizontal band across the triangle. Text reads, C40 equals 2, RA equals 17, RWR equals 0. The plotted points are concentrated in the middle portion of the triangle, with fewer points near the corners. The image F showing a ternary scatter plot labeled Killarainey 2 with scattered points and a thick horizontal band across the triangle. Text reads, C40 equals 5, RA equals 18, RWR equals 0. The plotted points are concentrated in the middle portion of the triangle, with fewer points near the corners. The image G showing a bar chart labeled Killarainey 1. The x axis categories read, VA, A, SA, SR, R, WR. The y axis label reads, Number, with tick labels 0, 5, 10, 15, 20, 25, 30. Bar heights: A equals 15, SA equals 30, SR equals 1. The highest bar is SA at 30. The image H showing a bar chart labeled Killarainey 2. The x axis categories read, VA, A, SA, SR, R, WR. The y axis label reads, Number, with tick labels 0, 5, 10, 15, 20. Bar heights: A equals 18, SA equals 20, SR equals 15. The highest bar is SA at 20 and SR at 15 is closer to A at 18 than in the other bar charts. Across the four sites, the bar charts show SA as the highest category in all four charts, with values 35 for Rosscahill 2, 25 for Ballyquirke, 30 for Killarainey 1 and 20 for Killarainey 2. The ternary scatter plots and the bar charts share the same site labels, with the ternary plots showing point distributions within the triangle and the bar charts showing category counts for VA, A, SA, SR, R and WR.
Beyond the moraine landforms, the KT esker includes multiple lithologies derived from the northwest (Fig. 7), KT2 having some of the highest diversity among all our exposures. This general composition reflecting lithologies from the northwest has also been described by previous workers (Gregory, Reference Gregory1921; Hennessy et al., Reference Hennessy, Carlin, Cunniffe, Feeney and Gormally2010). The lithologies present are coherent among RC2, KT1, and KT2, appearing to represent a previous flowline out of the Connemara region during esker deposition. KT1 and KT2 exhibit generally low C40 values, low angularity, and a high proportion (≥ 50%) of subrounded clasts (Fig. 10), as would be expected of water-transported esker sediments.
(Upper) Example of in-situ clasts within the KT esker, showing clear bedding of coarse- and fine-grained glaciofluvial sediments and exhibiting the generally high degree of rounding of clasts. (Lower) Ternary diagrams and clast form counts for KT1 and KT2, according to Sneed and Folk (Reference Sneed and Folk1958) and Benn and Ballantyne (Reference Benn and Ballantyne1993, Reference Benn and Ballantyne1994). Created using TRI-PLOT v 1.4.2. RA = proportion of VA (very angular) and A (angular) clasts; RWR = proportion of R (rounded) and WR (well-rounded) clasts.
Figure 10 Long description
Sediment exposure showing many rounded clasts embedded in finer material, with a small circular marker. Two ternary diagrams labeled, KT Esker 1 and, KT Esker 2. Each ternary diagram is a triangle with many plotted dot markers and a thick horizontal line across the triangle. Text beside the KT Esker 1 ternary diagram reads, C40 equals 3, RA equals 11, RWR equals 1. Text beside the KT Esker 2 ternary diagram reads, C40 equals 6, RA equals 5, RWR equals 1. Each ternary diagram includes the text labels the ratio of c to a as the left side of the triangle, the ratio of b to a as the right side of the triangle, and a minus b over a minus c as the bottom of the triangle. Two bar graphs labeled, KT Esker and, KT Esker 2. For both bar graphs, the vertical axis label is, Number, with tick labels 0, 5, 10, 15, 20, 25, 30. For both bar graphs, the horizontal axis categories are, VA, A, SA, SR, R, WR. Bar values for the bar graph labeled, KT Esker: A about 12, SA about 15, SR about 25. Bar values for the bar graph labeled, KT Esker 2: A about 5, SA about 18, SR about 28. Across the two bar graphs, the highest bar is SR in both graphs, with SR higher in KT Esker 2 than in KT Esker. The A bar is lower in KT Esker 2 than in KT Esker.
Discussion
Reconstructing former ice-flow direction
In addition to Carboniferous limestone, all but one (RL) of the seven moraine clast samples contain varying proportions of crystalline material derived from bedrock sources located to the west and northwest of the WCMS (Fig. 7); in contrast, the RL sample is composed entirely of limestone. Coupled with the abundance of Galway granite boulders on the moraine surfaces, we consider that the lithological composition of the WCMS most plausibly represents deposition by ice flowing east out of south Connemara. Under this scenario, ice flowlines would have intersected many, but not all, of the crystalline bedrock sources indicated by our clast samples as well as the limestone bedrock underlying the moraines (Table 3; Fig. S2). Materials from more distal sources, such as quartzite of the Lakes marble formation, likely reflect entrainment and transport of pre-existing deposits during the subsequent WCMS episode.
Where clast lithology is predominantly limestone (e.g., RC1, RC3, and RL), we note that samples from adjacent sites often exhibit strong igneous/metamorphic signatures; for example, RC2 and RC3 are located at different points along the same moraine crest (Fig. 11) and thus represent the same glacial configuration. We also observe that the limestone components commonly exhibit the highest-angularity clasts in our dataset (Fig. 8), reflecting the propensity of limestone to break into highly angular shapes and suggesting minimal subglacial transport. The latter is to be expected, given the highly local origin of limestone material. Ultimately, we suggest that the variable composition of the WCMS likely reflects (1) the innately heterogenous nature of glacial debris, (2) the temporally variable availability of bedrock outcrops for subglacial quarrying, and (3) local differences in bedrock strength (potentially of different limestone varieties) that may have affected the relative proportions of limestone material in derivative tills.
Locations of the seven moraine exposures investigated in this study. Yellow circles denote samples dominated by highly angular limestone material, while green circles are those with lower angularity values and higher proportions of crystalline clasts. The dashed line represents the averaged ice limit during readvance with all exposures closely corresponding.
Figure 11 Long description
The map displays locations of moraine exposures marked by yellow and green circles. Yellow circles indicate samples with highly angular limestone material, while green circles represent samples with lower angularity and higher crystalline clasts. The dashed line shows the averaged ice limit during readvance. Key locations labeled include RC1, RC2, RC3, RL, KR1, KR2, BQ and BL. A scale bar indicates distances of 0.75 km and 1.5 km. Directional markers show north and west orientations. The larger Lough Corrib is visible on the right side of the map, with land features and a red line marking a specific boundary or feature. The map provides a detailed view of geological features and sample sites in the region.
Although we surmise that the WCMS was deposited along the eastern margin of a residual Connemara ice mass, we recognize that previous workers have invoked different ice-flow configurations for the deglacial period. In his detailed investigation of the KT esker (which he termed the ‘Ross-Gortachalla’ esker), Gregory (Reference Gregory1921) invoked a lobe of Connemara ice flowing southward through the Lough Corrib basin and interpreted the KT esker as a water-lain deposit associated with the southern margin of that lobe. Subsequent work by Charlesworth (Reference Charlesworth1928, Reference Charlesworth1929) developed this model further, describing a regional ice mass draining south through Lough Corrib and citing moraine landforms immediately north of our study area. While neither author reported the moraines mapped in the present study, a south-flowing Corrib lobe as envisaged by Gregory (Reference Gregory1921) and Charlesworth (Reference Charlesworth1928, Reference Charlesworth1929) would, if true, imply that the WCMS was deposited by ice flowing from the east, counter to our interpretation.
At this stage, we consider that scenario improbable for several reasons. First, as detailed above, the composition of the WCMS has a strong Connemara signature, including boulders derived from the Galway granite pluton <2 km west of the moraines. Ice flowing south through the limestone Corrib basin would not have intersected many of those bedrock sources and thus is unlikely to have constructed moraines with the composition depicted in Figure 7.
Second, given the lateral continuity and topographic prominence of the WCMS, it would be reasonable to expect the former extent of a major Corrib ice lobe to be marked by correlative landforms elsewhere in the basin. To date, and despite ground survey and remote sensing analysis, we have not identified any such landforms.
Third, the composition and form of the KT esker, east of the WCMS, is incompatible with south-flowing ice in the Corrib basin. According to clast provenance, the KT esker reflects drainage out of Connemara (Gregory, Reference Gregory1921; Hennessy et al., Reference Hennessy, Carlin, Cunniffe, Feeney and Gormally2010; this study); no sources of these lithologies exist east of the esker. Although Gregory (Reference Gregory1921) interpreted the landform as a marginal (rather than subglacial) feature, exposures made in the esker over the intervening century exhibit clear units of stratified silt, sand, and gravel that are characteristic of subglacial esker morphology (Hennessy et al., Reference Hennessy, Carlin, Cunniffe, Feeney and Gormally2010; Fig. S2). Similarly, the sinuous, short-bead form and symmetrical prominence of the KT esker are consistent with a subglacial origin, rather than an ice-marginal deposit. Given the friable nature of glacio-fluvial features, we consider it unlikely that the KT esker would survive glacial overriding by south-flowing Corrib ice. Instead, we infer that the esker represents subglacial drainage beneath east-flowing ice during an earlier, more extensive phase of glaciation (e.g., Synge, Reference Synge1979). During the subsequent readvance, any westward extension of the KT esker inside the WCMS was likely destroyed or obscured. On these various bases, we ultimately reject the scenario of the WCMS being deposited by ice flowing south through the Loch Corrib basin.
Previous work in the west of Ireland has widely invoked a radially draining ice center located over Connemara during the late LGM (Gregory, Reference Gregory1921; Charlesworth, Reference Charlesworth1929; Synge and Stephens, Reference Synge and Stephens1960; Orme, Reference Orme1965; Greenwood and Clark, Reference Greenwood and Clark2009). The present study adds to a growing body of evidence for the persistence of this general configuration during deglaciation (Foreman et al., Reference Foreman, Bromley, Hall and Jackson2022, Reference Foreman, Bromley, Hall and Rodríguez2025) and is supported by striation data west of the WCMS (Fealy et al., Reference Fealy, Green, Loftus, Meehan, Radford, Cronin and Bulfin2009; Geological Survey Ireland, 2025). We note, however, that this interpretation does not align with some recent model-based reconstructions of ice sheet retreat in Ireland, which instead invoke eastward retreat of the ice sheet margin towards the Irish Midlands (Clark et al., Reference Clark, Ely, Hindmarsh, Bradley, Ignéczi, Fabel and Ó Cofaigh2022).
10Be chrono-stratigraphy
Having concluded that the WCMS was deposited by ice flowing eastward out of south Connemara, we now evaluate the new 10Be ages from the Rosscahill moraine in the context of the moraine-distal Baile Doite dataset (17.2 ± 0.5 ka) and the moraine-proximal Moycullen dataset (16.9 ± 0.9 ka) (Fig. 3). Assuming the WCMS represents a pause in, or reversal of, the overall deglaciation trend represented by the Baile Doite and Moycullen boulder deposits, the Rosscahill moraine (16.5 ± 0.4 ka) should fall chronostratigraphically between the distal and proximal populations. While the Baile Doite population is marginally older than the moraine, the three datasets do overlap within analytical uncertainty and therefore can be considered statistically indistinguishable (Fig. 3); combined, the resulting population (n = 13) has a mean age of 16.9 ± 0.7 ka (Fig. S3). At face value, the 1σ range provides a conservatively broad window (ca. 1400 years) in which deposition is likely to have occurred at each of the three sites. However, since the 1σ range is dominated by the broad age spread of the proximal Moycullen dataset (of which at least two ages exceed the mean age of the stratigraphically older Rosscahill moraine) and, to a lesser degree, by the bimodal Baile Doite dataset (Fig. 3), we posit that the true timeline of deposition at all three sites could have been considerably shorter. Further 10Be dating of sites distal and proximal to the WCMS will help improve the precision of this record. Meantime, we tentatively suggest that the WCMS represents a comparably brief interruption in the longer-term pattern of ice sheet decay that dominated HS1 (Foreman et al., Reference Foreman, Bromley, Hall and Jackson2022).
Beyond the Corrib basin, Foreman et al. (Reference Foreman, Bromley, Hall and Rodríguez2025) reported a set of 10Be-dated moraines in the Maumturk Mountains, central Connemara, that have a mean age of 16.8 ± 0.6 ka and thus overlap chronologically with the Rosscahill moraine chronology (16.5 ± 0.4 ka). Like the WCMS, the Maumturk moraines are positioned stratigraphically between full glacial conditions and widespread deglaciation, and were interpreted as reflecting a short-lived readvance of Connemara ice during the middle of HS1 (Fig. 11) (Foreman et al., Reference Foreman, Bromley, Hall and Rodríguez2025). Considering these clear temporal and stratigraphical parallels, we tentatively suggest that the WCMS and Maumturk moraines represent the same event in different sectors of the remnant Connemara ice center.
Climatic implications of the WCMS
The widespread contraction of Ireland’s terrestrial ice during HS1 (Wilson et al., Reference Wilson, Rodés and Smith2018; Foreman et al., Reference Foreman, Bromley, Hall and Jackson2022, Reference Foreman, Bromley, Hall and Rodríguez2025; Bromley et al., Reference Bromley, Bunce, Hall, Putnam and McDermott2026a) is mirrored in similar well-constrained records from mainland Europe (Schlüchter, Reference Schlüchter1988; Putnam et al., Reference Putnam, Denton and Schaefer2023), North America (Koester et al., Reference Koester, Shakun, Bierman, Davis, Corbett, Braun and Zimmerman2017; Bromley et al., Reference Bromley, Hall, Thompson and Lowell2019; Hall et al., Reference Hall, Lowell, Bromley, Putnam and Alan2025), the tropics (Bromley et al., Reference Bromley, Schaefer, Hall, Rademaker, Putnam, Todd, Hegland, Winckler, Jackson and Strand2016; Jackson et al., Reference Jackson, Kelly, Russel, Doughty, Howley, Chipman, Cavagnaro, Nakileza and Zimmerman2019), and southern middle latitudes (Hall et al., Reference Hall, Porter, Denton, Lowell and Bromley2013, Reference Hall, Denton, Lowell, Bromley and Putnam2017, Reference Hall, Putnam, Lowell, Denton, Russell, Soteres and Spoth-Ascencao2026; Putnam et al., Reference Putnam, Schaefer, Denton, Barrell, Andersen, Koffman and Rowan2013; Moreno et al., Reference Moreno, Denton, Moreno, Lowell, Putnam and Kaplan2015), suggesting that events in the NE North Atlantic were part of a broader spatial pattern (Toucanne et al., Reference Toucanne, Vázquez Riveiros, Soulet, Blard, Migeon, Rigalleau and Roubi2026). In the Northern Hemisphere, widespread shrinkage of land-based ice has variably been attributed to marine instability (Callard et al., Reference Callard, Ó Cofaigh, Benetti, Chiverrell, Van Landeghem, Saher and Livingstone2020; Ó Cofaigh et al., Reference Ó Cofaigh, Callard, Roberts, Chiverrell, Evans, Saher and Van Landeghem2021), glacial ‘starvation’ under excessively cold/dry conditions (Allard et al., Reference Allard, Hughes and Woodward2021), and amplified melting due to enhanced summer temperatures and anomalous thermal seasonality (Denton et al., Reference Denton, Alley, Comer and Broecker2005, Reference Denton, Toucanne, Putnam, Barrell and Russell2022; Putnam et al., Reference Putnam, Denton and Schaefer2023; Hall et al., Reference Hall, Lowell, Bromley, Putnam and Alan2025). In their assessment of the retreat of the Connemara ice center, Foreman et al. (Reference Foreman, Bromley, Hall and Jackson2022) concluded that, while North Atlantic climate may have been drier during HS1, the primary driver of stadial retreat was summertime melting, as reflected in elevated flux of European meltwater into the ocean at that time (Toucanne et al., Reference Toucanne, Soulet, Freslon, Jacinto, Dennielou, Zaragosi, Eynaud, Bourillet and Bayon2015; Martinez-Lamas et al., Reference Martinez-Lamas, Toucanne, Debret, Riboulot, Deloffre, Boissier and Cheron2020).
Within this broader climatic–cryospheric context, mounting evidence for short-lived readvances during HS1 raises the question of what climatic conditions such events represent. Was the WCMS formed under anomalously seasonal conditions, or does the readvance reflect a temporary breakdown of HS1 climate? Beyond our study area, deposits representing short-lived, mid-HS1-age readvances include the Wester Ross Readvance in NW Scotland (Robinson and Ballantyne, Reference Robinson and Ballantyne1979; Ballantyne et al., Reference Ballantyne, Schnabel and Xu2009; Simms et al., Reference Simms, Best, Shennan, Bradley, Small, Bustamante, Lightowler, Osleger and Sefton2022; Bromley et al., Reference Bromley, Hall, Putnam and Lowell2026b) and the Killard Point tills in NE Ireland (McCabe et al., Reference McCabe, Knight and McCarron1998), as well as mid-stadial moraines in central Connemara (Foreman et al., Reference Foreman, Bromley, Hall and Rodríguez2025), Donegal (Wilson et al. Reference Wilson, Ballantyne, Benetti, Small, Fabel and Clark2019), and the English Lake District (Wilson et al., Reference Wilson, Rodés and Smith2018). Similar events have been reported from sites farther afield, such as the HS1-age Prospect Hill readvance in the Southern Alps (Putnam et al., Reference Putnam, Schaefer, Denton, Barrell, Andersen, Koffman and Rowan2013) and Veluyoccocha event in Peru (Bromley et al., Reference Bromley, Schaefer, Hall, Rademaker, Putnam, Todd, Hegland, Winckler, Jackson and Strand2016).
While prior studies in Ireland and Britain have tended to correlate HS1-age landforms with stadial conditions (e.g., the ‘Killard Point stadial’) (McCabe et al., Reference McCabe, Knight and McCarron1998), often linking positive glacier mass balance to the notion of AMOC-driven cooling (Everest et al., Reference Everest, Bradwell, Fogwell and Kubik2006; Ballantyne et al., Reference Ballantyne, Schnabel and Xu2009; Wilson et al., Reference Wilson, Rodés and Smith2018), this association does not acknowledge the brevity of the readvance events themselves relative to the duration of HS1. This point is underscored by the WCMS, the age of which is indistinguishable within analytical uncertainty from those of adjacent deposits representing wholesale ice sheet retreat (Fig. 3). Had the readvance been of full HS1 duration, as would have been the case if stadials were cold and conducive to positive mass balance, we would expect a significant offset between the Rosscahill moraine 10Be ages and those from distal Baile Doite. Our chronology confirms that no such offset exists (Fig. 3).
Assuming that the seasonal expression of stadials in Europe’s maritime climate was instigated by the anomalous spread of winter sea ice, plausibly as a consequence of summer-temperature-induced melting and resulting freshwater export to the North Atlantic (Denton et al., Reference Denton, Toucanne, Putnam, Barrell and Russell2022; Toucanne et al., Reference Toucanne, Vázquez Riveiros, Soulet, Blard, Migeon, Rigalleau and Roubi2026), two factors might have caused the short-lived stabilization and/or readvance of Irish ice represented by the WCMS. First, a reversal in the expansion of winter sea ice could have allowed resumption of a more ‘normal’ mode of wintertime westerly airflow over the North Atlantic. Under such conditions, evaporation from the ocean surface would result in high precipitation (snowfall) over Ireland, thus nourishing the remnant ice masses; the same process might also reinstate milder winter temperatures than are associated with HS1. Second, a temporary reversal of anomalous stadial thermal seasonality also would have affected stadial summers. If HS1 summers were indeed warm, as suggested by the regional glacial record (Foreman et al., Reference Foreman, Bromley, Hall and Jackson2022, Reference Foreman, Bromley, Hall and Rodríguez2025; this study), any reversal could have caused a short-lived decrease in positive-degree days and in melt-driven ablation. Under such conditions, remnant ice masses in Ireland and throughout maritime Europe would be liable to stabilize and potentially advance. Acknowledging that these two scenarios are not mutually exclusive, we favor the second (reduced melt-season intensity) as the primary factor driving ice-margin stabilization and moraine construction, recognizing the dominant role of summer temperature in glacier mass balance (Oerlemans, Reference Oerlemans and Oerlemans1989, Reference Oerlemans2001, Reference Oerlemans2005; Zemp et al., Reference Zemp, Huss, Thibert, Eckert, McNabb, Huber and Barandun2019; Bromley et al., Reference Bromley, Putnam, Hall, Rademaker, Thomas, Balter‐Kennedy, Barker and Rice2023).
Conclusions
Deglaciation of the Connemara ice center during Heinrich Stadial 1 was interrupted by a short-lived pause or reversal, during which the extensive WCMS was formed in the western Lough Corrib basin. Sedimentological analysis of multiple moraine exposures, coupled with the lithology of glacial boulders on the moraine surfaces, confirms that the WCMS was deposited by ice flowing eastward out of Connemara. This configuration is in accord with earlier reconstructions of this part of the British–Irish Ice Sheet but does not agree with recent model-based simulations. Four new 10Be dates from the Rosscahill section of the WCMS indicate the readvance culminated ca. 16.5 ± 0.4 ka, in close agreement with bracketing 10Be data from sites distal and proximal to the WCMS. Together, these data confirm the short-lived nature of the glacial readvance. Stratigraphically and temporally, the WCMS is similar to well-dated intra-stadial moraine systems and deposits reported from elsewhere in Ireland, Britain, and more broadly. A salient characteristic of these correlative events is their apparent brevity—centuries at most—relative to the ca. 3 ka duration of HS1. We propose that the WCMS represents a brief reversal of the deglacial conditions that defined Ireland’s climate throughout HS1. Establishing the cause(s) and mechanisms of such short-lived perturbations is a priority for future research into abrupt climate change.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/qua.2026.10091.
Acknowledgments
This work was supported in part by the University of Galway and by co-funding from Research Ireland and Geological Survey Ireland (grant 21/FFP-P/10129). All data reported and discussed in this paper are listed in the references, figures, tables, and supporting information. All 10Be AMS measurements were made by Alan Hidy, Center for Accelerator Mass Spectrometry, Lawrence Livermore National Laboratory. We are most grateful to Becky Hedderman and Róisín Duggan for help with sample collection and preparation, Paulo Rodríguez for assistance with lithological identification, Billy Hamilton for GNSS assistance, Gerard Walsh for his kind permission to access the Ballyquirke moraine, and our reviewers for most excellent suggestions.
Open access
