The Doniford gravels and the curse of the LGM bias

 

"Churned" deposits at the base of the gravel sequence at Doniford.  At the base are Lias shale and limestone beds.  therev isb evidence here of a permafrost environment, with the operation of both fluvial and poeriglacial processes?  But are there any signs here of glacial action?

One of the great curses of modern stratigraphic studies is the abundance of "recent" sediments as compared with those which are ancient.  Recent events (such as glacial episodes) tend to wipe out most of the evidence of preceding similar events -- and we must be very careful that we do not automatically assume that the LGM glacial episode (MIS- 2) was longer and more powerful than its predecessors simply because LGM sediments and erodional traces are more frequently discovered and examined.  So was the Late Devensian glacial episode the most important one in the Quaternary record in the British Isles?  There are many who think that it was -- especially in the western parts of the glaciated or ice-affected area.    In the Midlands and in the east, the evidence is more equivocal, and the evidence seems to suggest that pre-Devensian glacial episodes were far more powerful and prolonged that that of the LGM.

The principle of LGM bias is one that we have to think about carefully when we examine Quaternary sites where all the evidence seems to suggest ba simple straigraphic sequence in which all the deposits seem to ralate to a single glacial cycle.  This is what I found over and again, in site after site, when I did my doctorate research on the Pembrokeshire coast in 1962-1965.  The overall sequence was this:

7.  Modern soil horizon

6.  Recent colluvium -- Holocene

5.  Fossil ice wedges and "churned" deposits relating to Late Glacial climate changes

4.  Fluvioglacial deposits and flowtills etc -- ice wastage phase

3.  Irish Sea till -- LGM glacial episode (MIS-2)

2.  Periglacial deposits including slope breccia and colluvium -- Early and Middle Devensian? MIS-4 and MIS-3)

1.  Raised beach deposits -- assumed to be Ipswichian in age (MIS-5)

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

The known sequence of marine isotope stages:

MIS 2 – 29,000 yrs BP Late Devensian or LGM -- peak glaciation. COLD

MIS 3 – 57,000 mid-Devensian.  COOL

MIS 4 – 71,000 Onset of Last Glacial Period / Weichselian / Devensian / Wisconsin in North America.  COLD

MIS 5 – 130,000 Eemian interglacial, or Ipswichian in Britain -- many substages. WARM

MIS 6 – 191,000 Illinoian glacial in North America, Saalian in northern Europe and Late Wolstonian in Britain)  COLD

MIS 7 – 243,000 (Aveley Interglacial in Britain) WARM

MIS 8 – 300,000 Early Wolstonian in Britain. COLD

MIS 9 – 337,000 (Purfleet Interglacial in Britain) WARM

MIS 10 – 374,000 COLD

MIS 11 – 424,000 Hoxnian Interglacial in Britain WARM

MIS 12 – 478,000 Anglian Glacial in Britain, Elster glaciation in northern Europe. COLD

==================

Only in a very few locations in Pembrokeshire have I found direct evidence of ancient glacial deposits, heavily stained and cemented.  The key sites are Black Mixen near Lydstep, Ceibwr near Moylgrove, Witches Cauldron near Moylgrove, and Traeth Mawr, Newport.  There are other cemented deposits, but I'm uncertain about their significance and stratigraphic relationships.  We need some proper cosmogenic dating.........

Anyway, switching rapidly to the Somerset coast, one of the most intruguing sites is Doniford,  near Watchet on the coast of Bridgwater Bay.  Here there is an intriguing set of gravels resting on a rock platform made of Liassic shale and limestone beds.  The basal layer is referred to as the "cobbly gravel bed", about 3m thick.  Since glacier ice has indubitably affected the Devon coast near Saunton, the Somerset coast near Bristol and the Somerset lowlands at leat 15 km inland, one would expect to find traces of glaciation on both the north-facing and west-facing segments of the Bridgwater Bay coast.  If only things were that simple..........

The gravels in the Doniford - Watchet area have been studied for many years, most notably by the Wedlakes and by Gilbertson and Mottershead in 1975.  See this:  https://cdn.fieldstudiescouncil.net/fsj/vol4.2_100.pdf

Campbell et al (in the GCR series; SW England volume) described the exposures in 1998, and there is now a new study by Basell et al, published in August 2025:

Resolving the late Pleistocene (MIS3–1) sedimentary sequence from Doniford, UK: Implications for British-Irish ice sheet extent, megafaunal history and hominin occupation
L.S. Basell et al, 2025
Quaternary Science Reviews 368 (2025) 109509

https://doi.org/10.1016/j.quascirev.2025.109509

ABSTRACT

We present a new Optically Stimulated Luminescence (OSL) chronology, detailed sedimentological evidence, new palynological data and a new Palaeolithic artefact from a classic site known for over 100 years. The flu-vioperiglacial sedimentary sequence and chronology irrevocably indicates that the BIIS did not reach the north shore of the SW peninsular of the British Isles in MIS 3–2. Both the sedimentology and palynology suggests cool–cold steppic conditions rather than polar desert. The discovery during this project of a new unrolled bout coup´e biface is significant since it adds another westerly example of this characteristic Middle Palaeolithic form associated with Neanderthals, that is relatively common in Britain (compared to Europe). The chronology and sedimentology also confirm the likely reworking of the cold fauna and Acheulian artefacts from regional floodplains prior to MIS 3. The site highlights the archaeological potential of actively eroding cliffs for expanding knowledge of hominin occupations of south–western Britain, near to an ‘edge’ of the Middle Palaeolithic world.

Much of the paper is devoted to the archaeological finds from the gravel exposures, but there are a number of points that have a bearing on our understanding of the Quaternary climate oscillations that have affected this part of the coast.  Quote from the text:

Our model has clear glacial implications as the latest modelling suggests the BIIS reached its maximum extent (LGM) at 26 ka (Hughes and Gibbard, 2015; Scourse et al. 2021; Clark et al., 2022). Interestingly, the maximum modelled LGM ice cover incorporates Lundy and reaches the north coast of Devon and Cornwall BP (Hughes and Gibbard, 2015; Clark et al., 2022, Fig. 8). GL05062 (25 ± 3 ka) which dates DM–Unit 3n, covers this period, but it is most likely that DM–Unit 3 was deposited either prior to the LGM and/or after this as the ice began to retreat and water flow reactivated. With the ice margins so close both to the west and the north of Doniford, at the LGM the landscape would have been frozen and inactive (as is seen in other UK fluvial systems e.g. Rivers Axe (Dorset/Somerset/Devon) and Trent)). The solifluction deposit DM–Unit 4 is consistent with post LGM climatic amelioration but being undated can only be said to be Late Devensian. 

Another quote: 

The Bristol Channel including Lundy Island provides evidence of early Devensian ice dynamics (Rolfe et al. 2012; Gibbard et al., 2017) though this is debated (Carr et al., 2017). The basal date and the erosive boundary at the base of the Doniford Member supports such an early glacial ice advance. The Doniford Member, Helwell Bay Member and potentially the Swill Member are consistent with a proximal ice margin. This is evidenced on the ground and via modelling. The gravels record the repeated periglacial processes combined with pulses of sediment deposition during periods of glacial retreat or seasonal thawing.

The  OSL dates used in the paper are difficult to interpret, and the authors clearly have difficulty in incorporating into their modelled or proposed chronology the proposed Early Devensian glaciation of Lundy Island (see other posts on this).  But what we can say with reasonable confidence is that the evidence from Doniford confirms that glacier ice did not reach this part of the Somerest coast during the Devensian LGM.  But it was not far away, with an ice edge somewhere in the Bristol Channel........

As for the date of the glacial incursion which DID affect the Somerset coast, having read the article, we are not much wiser. There is a possibility that at the base of the Doniford sequence the gravels incorporate cobbles and boulders that are derived from older glacial deposits.  Although no far-travelled erratics have been identified, the wide range of stone shapes is suggestive of the presence at some stage of glacier ice.  The basal gravels are poorly bedded and poorly sorted, as noted by Gilbertson and Mottershead in 1975 as they pondered on a partly glacial origin.   I do not see any evidence here that there was an Early Devensian incursion of ice across the coast; and the most likely scenario still seems to involve an extensive glacial episode in Late Wolstonian time, at the time of MIS-6, maybe around 200,000 years ago.  

Evidence from Eastern England now suggests two Wolstonian glacial episodes, one in MIS-8 and the other in MIS-6.  The evidence in the field is confusing and sometimes contradictory -- and it remains to be seen whether these two episodes can be discerned in the record for Western Britain.