Chapter 8 The Quaternary history of the Isles of Scilly

Introduction

J.D. Scourse

The Isles of Scilly contain a wealth of exceptionally important and interesting Quaternary sediments and landforms quite out of proportion to their relatively small area. Situated 45 km west-south-west of Land's End (Figure 8.1), the group consists of some 100 islands, most of which are devoid of soil and terrestrial vegetation. Most major appraisals of the Quaternary stratigraphy and geomorphology of the islands (Barrow, 1906; Mitchell and Orme, 1967; Scourse, 1991) have been based on evidence from the five largest and permanently inhabited islands of St Mary's, St Martin's, St Agnes, Tresco and Bryher, and the larger uninhabited islands of Samson, St Helen's, Northwethel, Tean, Nornour, Great Ganilly, Great Arthur, Little Arthur and Annet (Figure 8.1).

The Quaternary significance of the islands lies in the evidence they contain for: 1. changes in relative sea level, in the form of raised beaches; 2. former periglacial conditions elucidated from sedimentary, geomorphological and palaeobotanical sites including the first radiometric dates for the coastal 'head' deposits of South-West England; 3. glaciation of the northern islands with a well-defined glacial limit straddling the archipelago, indicated by a range of glacial, glaciofluvial and associated aeolian sediments; 4. some of the finest granite landforms in the British Isles whose evolution and form can be demonstrated to be intimately linked to the periglacial and glacial history of the islands; and 5. the successful establishment of forest tree taxa on the islands during the Holocene post-dating the severance from mainland Cornwall by sea-level rise. Of particular significance and importance are the palaeobotanical records from periglacial sequences which provide evidence of the vegetation of the late Middle and early Late Devensian, a phase very poorly covered by sites elsewhere in Britain, and the glacial sequences, which, together with glacigenic sediments from the adjoining continental shelf, provide evidence of the most southerly advance of ice during the Quaternary in north-west Europe.

The solid geology of the islands is almost exclusively granite (Figure 2.2), Scilly representing the highest parts of an almost completely submerged elliptical cupola which is the westernmost extension of the Variscan batholith of South-West England. Barrow (1906) divided the granite into coarse- and fine-grained facies and believed the only existing exposures of country rock to be the highly tourmalinized slates (linos') of White Island, St Martin's. Although this material has been re-interpreted as either sheared or greisened granite (Hawkes, 1991), one small exposure of phyllitic country rock containing sporadic perthite megacrysts does occur on Shipman Head, Bryher (Hawkes, 1991). Although no in situ Mesozoic sediments occur on the islands, Barrow (1906) interpreted gravel, consisting largely of flint and greensand, on the summit of Chapel Down, St Martin's, to be of Eocene (?) age. This he correlated with the Eocene fluvial gravel capping the Haldon Hills in east Devon.

The homogeneity of the bedrock assisted in the early identification of pebbles of 'foreign' lithology on the northern islands. Smith (1858) recorded chalk-flints and greensand from Castle Down, Tresco. These were later interpreted by Whitley (1882) as glacial in origin. Barrow exhibited a striated boulder from the islands to the Geological Society in 1904 (Barrow, 1904), and he was later (1906) able to place the occurrence of the erratics within a stratigraphic framework. He took Chad Girt (see below), on White Island, St Martin's as his type-site for the Pleistocene of the islands. At Chad Girt, and at many other sites, Barrow observed that in cliff section the erratics were usually set within a silty matrix often cemented by iron oxides, and that this 'glacial deposit' was both underlain and overlain by head deposits, the whole resting on a raised beach. Barrow accepted Whitley's (1882) glacial hypothesis, but preferred deposition from 'floe-ice'.

In 1957 Dollar attributed the St Martin's gravels to the Pliocene, and in 1960 Mitchell further revised their probable age to Lower Pleistocene, citing a fluvial origin. Mitchell and Orme (1965, 1967) undertook a major re-examination of the Pleistocene stratigraphy of the islands and identified a more complex sequence than Barrow:

6. Upper Head Weichselian ( = Devensian)

5. Raised Beach (Porth Seal) Eemian ( = Ipswichian)

4. Glacial Deposit Gipping ( = Wolstonian)

3. Lower Head Gipping ( = Wolstonian)

2. Raised Beach (Chad Girt) Hoxnian

1. Shore platform

(revised stage names from Mitchell et al., 1973b)

They divided Barrow's (1906) 'glacial deposit' into two facies, till and outwash gravel, and identified an ice limit running across the northern islands based on the distribution of these sediments (Figure 8.1). By comparing their stratigraphy with similar sequences in mainland South-West England and in Ireland, Mitchell and Orme suggested a Gipping ( = Wolstonian) age for the glacial event. This interpretation was largely based on the suggestion that the erratic-free Chad Girt raised beach is Hoxnian in age by correlation with other raised beaches of supposed Hoxnian age at similar elevations elsewhere, for example, the Courtmacsherry raised beach in southern Ireland. Direct correlation of lithostratigraphic units with chronostratigraphic stages on a one-to-one basis then dictated that the Porth Seal raised beach should be Eemian ( = Ipswichian) in age. The interpretation of a Wolstonian glacial limit on the islands as a result became firmly established in the literature (cf. Catt, 1981; Lowe and Walker, 1984), despite some speculation that the glacial material on the Isles of Scilly might be younger in age (John, 1971; Synge, 1977, 1985).

Bowen (1969, 1973b) questioned Mitchell's and Orme's interpretations, proposing that the lenticular form of the glacial material and its geomorphic association with coastal valleys was more consistent with an origin as soliflucted till. He later (1981) suggested that the critical stratigraphy identified by Mitchell and Orme (1967) at the Porth Seal site was 'inferred and superposed' and argued that granite corestones had been misinterpreted as a marine deposit. Bowen (1973b) regarded the single raised beach as Ipswichian in age, the soliflucted glacial deposits having been originally emplaced in the Wolstonian.

Coque-Delhuille and Veyret (1984, 1989) proposed a much more southerly ice limit on the islands. They concurred with Mitchell and Orme (1967) and Bowen (1973b) that the glaciation was Wolstonian in age.

The dating of the glacial material on the Isles of Scilly by Mitchell and Orme (1967) and Bowen (1973b) was therefore heavily dependent on the number and age of stratigraphically juxtaposed raised beach units, and lithostratigraphic correlation with neighbouring regions.

Scourse (1991), in the most recent re-examination of the islands, proposed a revised local stratigraphy independent of the stratigraphies erected in neighbouring regions. He reported organic sediments (Figure 8.2) from a number of sites which provided a chronology through multiple radiocarbon determinations, and which yielded palaeobotanical evidence critical in palaeoenvironmental reconstruction. Thermoluminescence (TL) dates (Wintle, 1981) and optical dates (Smith et al., 1990) further assisted in the establishment of a radiometric chronology for the sequence.

Scourse (1991) defined eight lithostratigraphic units of member status which he incorporated into two lithostratigraphic models for the 'southern' (extra-glacial) and 'northern' (glaciated) Isles of Scilly (Figure 8.3). The southern limit of the Hell Bay Gravel defines the boundary between these two areas, and corresponds closely to the ice limit identified by Mitchell and Orme (1967; (Figure 8.1)). The same lithostratigraphical units are used, albeit in slightly modified form, in the Geological Society's recently revised correlation of British Quaternary deposits (Campbell et al., in prep.).

Overlying the raised beach sediments of the Watermill Sands and Gravel in the southern Isles of Scilly is the Porthloo Breccia, a variable unit of soliflucted material derived entirely from the weathering of the granite bedrock (Scourse, 1987). The organic deposits were found towards the base of this unit at five sites (Figure 8.1); Carn Morval [SV 905 118], Watermill Cove [SV 925 123], Toll's Island [SV 931 119], Porth Askin [SV 882 074] and Porth Seal [SV 918 166]. Scourse interpreted these organic sequences as the infillings of small ponds or lakes impounded by active solifluction sheets or lobes.

Radiocarbon dates from these organic sediments are critical since they pre-date the units associated with the glacial advance (Figure 8.3), the Scilly Till, the Tregarthen Gravel, the Hell Bay Gravel and the Old Man Sandloess. The radiocarbon determinations indicate deposition of the organic material between 34 500 + 885/− 800 (Q–2410) and 21 500 + 890/−800 (Q–2358) BP. They provide a maximum age for the glacial event and the first radiometric dates for the coastal 'head' deposits of South-West England.

All samples of organic sediment used for these radiocarbon determinations were taken from permanent open coastal sections which posed considerable problems of contamination by younger carbon derived from rootlet and groundwater sources. However, multiple dates of both humic and humin fractions of samples from different locations within the deposits enabled Scourse (1991) to identify the extent and sources of contamination and therefore to assess the reliability of the resultant determinations.

The pollen sequences from the Pleistocene organic deposits are all very similar in recording open grassland vegetation. Carn Morval on St Mary's (Figure 8.1) yielded the most detailed pollen profile, but has not been selected as a GCR site because its organic material has now been largely removed by coastal erosion. The organic beds at the site were situated towards the base of the Porthloo Breccia, and the pollen assemblages (Scourse, 1991; Fig. 10) are dominated by Gramineae (grasses), Cyperaceae (sedges) and other herb taxa. Pinus is the most important tree taxon, but Scourse (1991) interprets this as a long-distance component rather than indicating local presence; the increase in Pinus towards the top of the diagram is probably the result of the relative decline in the pollen productivity of the local herbaceous flora set against a relatively constant supply of long-distance Pinus.

These pollen sequences represent the earliest Quaternary vegetational record for South-West England west of east Devon. The spectra are broadly similar to others of the same age from elsewhere in north-west Europe (Bell et al., 1972; Morgan, 1973; West, 1977b).

In the southern Isles of Scilly, the Porthloo Breccia is overlain by the Old Man Sandloess, a coarse aeolian silt with subdominant fine sand and minor amounts of clay (Catt and Staines, 1982). This has yielded two TL dates, both of 18 600 ± 3700 BP (QTL1d and QTL1f) (Wintle, 1981) and two optical dates, 20 ka ± 7 ka and 26 ka + 10/− 9 ka BP (738al and 741al; Smith et al., 1990). This material occurs in a variety of facies related to different modes of reworking.

In the northern Isles of Scilly, the Porthloo Breccia is overlain by three units that are all related to a single glacial event. The Scilly Till, a massive, poorly sorted, clay-rich, pale-brown diamicton containing abundant striated and faceted erratics of northern derivation, occurs at Bread and Cheese Cove, at Pernagie Bar and White Island Bar (Figure 8.1). The precise depositional environment of this material is uncertain but Scourse (1991) argues, on the basis of sedimentological, structural, lithological and fabric (eigenvalue) data, that it is likely to be of lodgement origin. This conflicts with the suggestion by Eyles and McCabe (1989), echoing the 'floe-ice' hypothesis of Barrow (1906), that this material is of glaciomarine origin. At Bread and Cheese Cove the Scilly Till occurs in association with a matrix-supported sandy gravel, the Tregarthen Gravel, which has a similar erratic assemblage. The Tregarthen Gravel is best displayed at its type-site, Battery (Castle Down) on Tresco (see below — Battery; (Figure 8.1)).

Aeolian loessic processes in association with the glacial advance resulted in the deposition of the Old Man Sandloess in the southern Isles of Scilly. The relative coarseness of this material is interpreted by Scourse (1991) as a function of its proximity to glacially derived source material. The mineralogy of the Scilly Till is sufficiently similar to that of the Old Man Sandloess (Catt, 1986) to suggest a genetic link between the two units.

Overlying the Scilly Till and the Tregarthen Gravel in the northern Isles of Scilly is the Hell Bay Gravel, an extremely widespread matrix-supported gravel containing a similar assemblage of striated and faceted erratics to the underlying till, but alongside a considerable proportion of locally derived granitic material. The lithology and mineralogy of the Hell Bay Gravel is identical to colluvially reworked facies (facies D) of the Old Man Sandloess (Figure 8.3). This material represents an initial phase of solifluction, post-dating the glacial event, in which the Scilly Till, Tregarthen Gravel and Old Man Sandloess were mixed and transported downslope. In situations where these sediments were stripped from the land surface, weathered granite once again became the dominant raw material for solifluction, this subsequent phase being represented by the Bread and Cheese Breccia in the northern islands and the upper Porthloo Breccia in the south. The Hell Bay Gravel is synonymous with Barrow's (1906) iron-cemented 'glacial deposit'.

There is a strong spatial relationship between the distribution of the glacigenic and glacial-derived sediments and marine bars, tombolos and granite landforms in the Scillies. Many marine bars occur within the limit of the Hell Bay Gravel, and two of these, Pernagie Bar and White Island Bar (Figure 8.1), are directly underlain by Scilly Till. Although most of the boulders comprising these bars are granite, many are erratics, dominantly flint, red siltstones and sandstones. Whereas contemporary marine processes clearly control their detailed morphology, the internal structure of the bars and their distribution in relation to the sedimentary units discussed above suggests a possible morainic origin (Scourse, 1991).

Scourse (1987) defined four tor forms from the Isles of Scilly, and there appears to be a good correlation between highly smoothed, eroded tors and the southern limit of the Hell Bay Gravel, suggesting that the glacial advance, though unable to penetrate far into the Scilly massif, was nevertheless capable of eroding the solid granite. Tor forms in the southern islands are, by contrast, 'mammilated' or 'castellated' (see below — Peninnis Head; (Figure 8.6)).

The evidence presented by Scourse (1991) therefore indicates that ice advanced as far as the northern Isles of Scilly during the Dimlington Stadial of the Late Devensian around 18 600 t 3700 BP. However, this may not have been the first glacial event to affect the islands because erratics are widespread in some exposures of the Watermill Sands and Gravel, the basal lithostratigraphic member of the Pleistocene sequence. The age of this earlier event remains uncertain.

Scourse's (1991) interpretation clearly conflicts in a number of important respects with Mitchell and Orme's (1967) sequence of events. The main differences include: 1. recognition of only one raised beach unit stratified with other sediments; 2. recognition of widespread loessic sediments, and the interpretation of a sedimentary suite of till, out-wash gravel and loess related to a single glacial event; and 3. independent radiometric dating rather than relative dating based on the inferred ages of raised beach units.

The interpretation of the Late Devensian glaciation of the Isles of Scilly, and the evidence on which it is based, have proved controversial. Scourse (1991) identifies the main points of contention as: 1. the reliability of the radiometric dates; 2. the validity of the lithostratigraphic correlations between sites and islands; and 3. the in situ status of the Scilly Till. Scourse (1991) discusses these points in turn and concludes that the evidence in support of the Late Devensian model is much more substantial than that proposed by Mitchell and Orme (1967) to support the hypothesis of Wolstonian glaciation.

Glacial and glaciomarine sediments have been discovered on the continental shelf adjoining the Isles of Scilly as far south as 49° (Scourse et al., 1990, 1991). These sediments have been grouped into a northern facies containing sparse reworked microfaunas, interpreted as either overconsolidated lodgement tills or proximal glaciomarine sediment, and a southern facies containing abundant cold-water microfaunas, interpreted as distal glaciomarine silty clays. Scourse et al. (1990, 1991) correlate both facies with the Scilly Till, thereby enabling a quantitative reconstruction of ice thicknesses, grounding-line, sea level and shoreline elevations in the Celtic Sea at c. 19 ka BP.

The Late Devensian ice advance responsible for the glaciation of the northern Isles of Scilly and the adjoining shelf is thought to have terminated in marine waters towards the shelf edge break and is likely to have constituted a thin lobate surge over deformable marine sediments at the southern terminus of the Irish Sea ice stream (Scourse et al., 1990, 1991; (Figure 8.4)).

Many of the GCR sites on the Isles of Scilly have been selected because they provide critical evidence relating to this glacial event and its dating; these sites include Bread and Cheese Cove (St Martin's), Watermill Cove (St Mary's), Porth Seal (St Martin's), Battery (Castle Down, Tresco), Castle Porth (Tresco) and Old Man (Gugh, St Agnes). Other sites have been chosen for their historical significance or because they contain evidence which has proved controversial in the evolving ideas on the Quaternary history of the islands; these include Chad Girt (White Island, St Martin's), Northward Bight (St Martin's) and Porthloo (St Mary's). Peninnis Head (St Mary's) has been selected because it contains spectacular granite landforms typical of features which lie outside the limit of the Hell Bay Gravel, and Higher Moors (St Mary's) because it contains the palaeobotanical evidence for Holocene forest development (Scaife, 1984, 1986).

References