The Physical Evolution of the North Avon Levels a Review and Summary of the Archaeological Implications

Home , Archaeomagnetic dating, Arlingham, Awkley, Meare Lake Village, Terminus post quem, Wessex

By Michael J. Allen and Robert G. Scaife The Physical Evolution of the North Avon Levels: a Review and Summary of the Archaeological Implications

by Michael J. Allen and Robert G. Scaife

with contributions from J.R.L. Allen, Nigel G. Cameron, Alan J. Clapham, Rowena Gale, and Mark Robinson

with an introduction by Julie Gardiner

Wessex Archaeology Internet Reports

Published 2010 by Wessex Archaeology Ltd Portway House, Old Sarum Park, Salisbury, SP4 6EB http://www.wessexarch.co.uk/

Wessex Archaeology Limited is a Registered Charity No. 287786

Contents

List of Figures List of Plates List of Tables

Editor’s Introduction, by Julie Gardiner ...... 1

INTRODUCTION The Severn Levels ...... 5 The Wentlooge Formation ...... 5 The Avon Levels ...... 6 Background ...... 7

THE INVESTIGATIONS The research/fieldwork: methods of investigation ...... 10 The Sites ...... 11 1. Awkley Lane ...... 11 The sequence ...... 11 Chronology ...... 12 Sampling the sequence ...... 16 Textural features of the minerogenic sequence, by J.R.L. Allen ...... 18 Pollen analysis, by Robert G. Scaife ...... 18 Diatom analysis, by Nigel G. Cameron ...... 30 Molluscs, by Michael J. Allen ...... 32 Insect remains, by Mark Robinson ...... 32 Waterlogged plant macrofossils, by Alan J. Clapham ...... 34 2. Vimpenny’s Lane ...... 37 The sequence ...... 37 Chronology ...... 40 Sampling the sequence ...... 40 Textural features of the minerogenic sequence, by J.R.L. Allen ...... 40 Pollen analysis, by Robert G. Scaife ...... 42 Diatom analysis, by Nigel G. Cameron ...... 49 Molluscs, by Michael J. Allen ...... 52 Insect remains, by Mark Robinson ...... 52 Waterlogged plant macrofossils, by Alan J. Clapham ...... 53 3. Hallen Marsh ...... 54 The sequence ...... 54 Sampling the sequence ...... 56 Textural features, by J.R.L. Allen ...... 60 Pollen analysis, by Robert G. Scaife ...... 60 Diatom analysis, by Nigel G. Cameron ...... 67 Molluscs, by Michael J. Allen ...... 67 Charcoals, by Rowena Gale ...... 68 4. Northwick ...... 69 The sequence ...... 69 Molluscs, by Michael J. Allen ...... 72 5. Awkley Interface ...... 74 The sequence ...... 74

3 Molluscs, by Michael J. Allen ...... 78 Comment ...... 78

DISCUSSION AND REVIEW. The Physical and Vegetational Evolution of the Avon Levels Stratigraphic Overview ...... 79 The Wentlooge Formation ...... 79 Sea-level change and palaeo-geographic reconstruction...... 80 Patterns of consistency: the mapped sediments ...... 82 Environmental change in the Avon Levels ...... 85 Pollen taphonomy and interpretation ...... 85 The late prehistoric vegetation of the Avon Levels ...... 87 The regional pollen database ...... 87 Avon Levels sites ...... 88 The changing wetland/floodplain/estuarine habitat ...... 94 Summary ...... 96 The mineral, estuarine and freshwater sediments ...... 97 Archaeology and environment: marsh landscape and archaeology……………………………….. 98 Some archaeological implications ...... 98 History of human exploitation patterns and behaviour ...... 99 Why does settlement occur? ...... 103 What caused Iron Age abandonment? ...... 103 A change of use ...... 105 Land claim ...... 105 Archaeological significance ...... 106 Conclusions ...... 106 Acknowledgements ...... 106 Appendix: analytical techniques ...... 108 Bibliography ...... 111

4 List of Figures

Figure 1. The Inner Bristol Channel and Severn Estuary showing the distribution of post-glacial alluvium comprising the Severn Estuary Levels Figure 2. Sedimentary features comprising the Wentlooge Formation Figure 3. The geology of the Avon Levels showing the location of the archaeological and sedimentological investigations Figure 4. Awkley Lane Wentlooge Formation profiles Figure 5. Radiocarbon determinations in relation to their OD and the middle Wentlooge wood and fen peat Figure 6. Awkley Lane; particle size triangle Figure 7. Awkley Lane; pollen diagram Figure 8. Diatom analysis; all sites Figure 9. Vimpenny’s Lane Wentlooge Formation profiles Figure 10. Vimpenny’s Lane; particle size triangle Figure 11. Vimpenny’s Lane pollen diagram Figure 12. Hallen Marsh: Summary plan of the excavation Figure 13. Hallen Marsh; full site profiles Figure 14. Hallen Marsh Wentlooge Formation profiles Figure 15. Hallen Marsh; particle size triangle Figure 16. Hallen pollen diagram Figure 17. Northwick; summary plan of the excavations Figure 18. Northwick Wentlooge Formation profiles Figure 19. Awkley Interface composite transect Figure 20. Relative heights of soils and stasis horizons from various locations Figure 21. a. Sea-level curve using the existing framework of Haslett et al. (1997) from a synthesis of data obtained by Kidson and Heyworth; 20b. Seaa-evel curve using the the corrected RSL graph for Caldicot Pill on the north side of the Severn Crossing Figure 22. Correlation of the sedimentary facies from the reported sites Figure 23. Location of archaeological sites in the Avon Levels Figure 24. Suggested model for the exploitation of the Avon Levels, Neolithic to late Roman period

List of Plates

Plate 1. Sampling the minerogenic sediment sequence in monoliths at Awkley Lane Plate 2. Awkley Lane excavation showing the fully excavated and exposed upper and middle Wentlooge sequence Plate 3. Vimpenny’s Lane excavation showing the fully excavated and exposed upper and middle Wentlooge sequence Plate 4. Round-house 1 at Hallen after excavation Plate 5. Round-house 2 at Hallen after excavation

List of Tables

Table 1. Radiocarbon determinations and calibrated results from the middle Wentlooge Formation at Awkley Lane Table 2. Mollusc data from Awkley Lane Table 3. Coleoptera and other insects from Awkley Lane Table 4. The waterlogged plant remains from the wood and fen peat of the middle Wentlooge Formation at Awkley Lane Table 5. Radiocarbon determinations and calibrated results from the detrital peat at Vimpenny’s Lane Table 6. Mollusc data from Vimpenny’s Lane Table 7. The waterlogged plant remains from the detrital peat (context 211) from Vimpennys Lane Table 8. Mollusc data from Hallen Marsh Table 9. Charcoal from the round-house at Hallen Marsh Table 10. Mollusc data from Northwick Table 11. Mollusc data from Awkley Interface, trench 4 Table 12. Stasis horizons in the middle Wentlooge sequence in relation to their OD height and radiocarbon determination

6 The Physical Evolution of the North Avon Levels: a Review and Summary of the Archaeological Implications

by Michael J. Allen and Robert G. Scaife

with contributions from J.R.L. Allen, Nigel G. Cameron, Alan J. Clapham, Rowena Gale, and Mark Robinson

with an introduction by Julie Gardiner

Editor’s introduction by Julie Gardiner

The construction of the new motorway links for the Second Severn Crossing and the resultant infill and developments which would come in their wake provided the opportunity for a comprehensive programme of archaeological desk-based research, auger survey, building recording, excavation, and palaeo-environmental work (The Second Severn Crossing English Approaches project). The defined area of study covered approximately 54 km², bounded respectively to the north and east by the M4 and M5 motorways, to the south by Bristol, and to the west by the Severn Estuary. This work was completed in 1994.

An assessment report was prepared (Wessex Archaeology 1994), in which a Revised Project Design proposed a series of research aims and themes and the preparation and publication of a monograph text comprising a series of chronological chapters discussing the development of the Avon Levels landscape. There followed a lengthy period during and after the construction of the motorway links and new bridge during which the post-excavation programme was not commissioned. This happened in 2000, by which time considerable further archaeological work had taken place on both sides of the Severn Estuary and many of the most important results published (eg, Rippon 1996; Locock 1997; Locock et al. 1998; Bell et al. 2000). These more recent archaeological works and publications allowed for a more focused and critical appraisal of the research themes defined in 1994. The relative importance and information potential of some aspects of the English Approaches project could be seen to have been enhanced, and others diminished, in the light of new information and ongoing research.

The themes deemed to remain of greatest importance and potential were: The characterisation and nature of later Bronze Age to Romano-British settlement and settlement patterns. The characterisation and chronological development of the sedimentological sequence during the Holocene. The reconstruction of the evolution of the palaeo-environment.

New proposals for publication were drawn up (Wessex Archaeology 2000), consisting of a series of papers, of which the two main ones, destined for appropriate academic journals, would deal specifically with the first of these themes on the one hand, and with the latter two on the other. The ‘archaeological’ paper was completed in 2001 and published in the Proceedings of the Prehistoric Society (Gardiner et al 2002). A suitable publication vehicle for the second, much longer, paper was identified and favourable discussions with the Editors ensued before writing commenced. The draft 1 of the second paper was also completed in 2001. Unfortunately, in the meantime, the journal underwent an editorial review and altered its remit to the extent that the paper was no longer deemed suitable for inclusion.

The length and complexity of the paper was such that it was difficult to identify another journal that could take it in its entirety, yet it was not an attractive proposition for a monograph. Various protracted negotiations were entered into regarding suitable formats and re-arrangements of the information, including a multi-media approach to publication. In the end it was decided that the strength of the paper was in the integration of the detailed analyses with wider thematic discussions and that its integrity would be unnecessarily compromised by the various proposals to ‘disaggregate’ it that had been considered. The decision has therefore been taken to publish it here, in full, online.

The text that follows is that which was drafted in 2001. It had already been subject to external academic refereeing pending submission to the journal and the referees’ comments, and those of its major contributors, have been largely incorporated. Financial contraints have meant that no new or additional work has been undertaken and no account has been taken of any archaeological (or sedimentary) work that has been undertaken since completion of the draft, or of any more recently obtained sea-level index points. We are confident that none of this will have substantially altered the arguments and conclusions presented here though some refinements at the detailed level have certainly been achieved. The only updating that has been undertaken has been in terms of bibliographic references (several cited ‘in prep’ and ‘forthcoming’ papers have been published in the interim) and the radiocarbon dates obtained as part of the project have been recalibrated using Oxcal 4.1.

INTRODUCTION

This paper reviews the results of research into the sedimentary sequences examined as a result of mitigation for the approach roads of the Second Severn Crossing. Although most of the work lay in strictly archaeological investigations, the brief was broad enough to allow excavation and detailed examination of long sedimentary sequences with peat horizons that occurred along the road corridor. This paper provides the first opportunity to present the detailed results of the palaeo- environmental analysis of the later Holocene sedimentary sequence of the region. These interpretations are of value in understanding the physical development and evolution of the Avon Levels and the wider Severn Estuary Levels. They also provide the foundation for wider discussion of the social development and patterns of human exploitation of the area, particular in later prehistory. These archaeological elements are outlined here (see discussion) and dealt with in more detail elsewhere (Gardiner et al. 2002).

There is a wealth of literature concerning the evolution, salt-marsh accretion, geomorphology and nature of the Severn Estuary. In particular this has concentrated on the sedimentary history of coastal sites, especially around the inner Severn Estuary, eg, Arlingham (Allen 1990a), Longney (Allen & Fulford 1990a), Elmore (Allen & Fulford 1990b), Slimbridge Warth (Allen 1986), and the Welsh Severn Estuary, eg, Rumney Wharf (Allen 1987a; 1996) (see Fig. 1). More recently, the archaeological significance of these areas has been demonstrated, from early Holocene times (Mesolithic) to the Iron Age in the Gwent Levels, Wales, at Goldcliff (Bell et al. 2000), and the Roman to medieval history in both the Welsh and Inner Severn Estuary (Allen and Fulford 1986; 1988; Allen & Rae 1987; Rippon 1991; 1994; 1996; 1997). Most of this work has its origin in the palaeo-environmental and chronological studies in the Somerset Levels initiated by Godwin (1943).

3 INSET River Severn Gloucester Elmore Arlingham Longney

Awre

Slimbridge INNER ESTUARY Hill Gw ent Levels Coldicot Level Oldbury

Wentlooge Level Goldcliff Redwick Rumney Avon Levels Avonmouth Fig.3 Cardiff MIDDLE Bristol ESTUARY

North Somerset Level

Inner Bristol Channel Brean OUTER ESTUARY Somerset Levels Porlock

Zones of peat exposed intertidally Estuarine alluvium 0 10 km

Figure 1. The Inner Bristol Channel and Severn Estuary showing the distribution of post-glacial alluvium comprising the Severn Estuary Levels

4 The Severn Levels

The Severn Estuary and Inner Bristol Channel lie along the Severn Carboniferous Fault Zone (Wilson et al. 1988), and bedrock surface forms a broad flat-floored ‘outer valley’, which contains estuarine sediments through which the Severn and its tributaries cut narrow gorges during the late Devensian (and probably earlier) low-stand sea(s) (Allen 1990b). The soft Triassic to early Jurassic mudrock surface is scarred by numerous periglacial features generally ascribed to the Late Devensian, Dimlington Stadial (Andrews et al. 1984; Allen 1987b).

The catchment of the Severn is c. 11,400 km2 and includes much of mid-Wales and the West Midlands, and the Severn Estuary. The river valley is overlooked by high ground of Carboniferous rocks (Mendips and Bristol), mid Jurassic limestones (Cotswolds) and Old Red Sandstones and Carboniferous strata (Forest of Dean). These bound large areas of low-lying wetlands which occur on both sides of the estuary.

These wetlands comprise a total of c. 840 km2 of largely disconnected and generally intricate outcrops of post-glacial alluvium (c. 8 km3) declining from the large inland Somerset Levels to the very small area of the inner estuary (eg, Arlingham). The main wetlands are the Somerset Levels, forming the largest inland wetland fringing the Inner Bristol Channel. Along the coast of the outer estuary are the Gwent Levels, comprising the Wentlooge and Caldicot Levels on the Welsh coast, and the North Somerset Levels and the smaller Avon Levels marking the northern limit on the east coast. The Middle estuary comprises much narrower and less extensive wetlands including those supported by the intertidal rock platforms of the Oldbury Flats and Hills Flats, while the Inner estuary, extending 15–20 km down stream to Gloucester (ie Slimbridge to Gloucester), includes important but smaller wetlands such as those studied around Awre, Arlingham, and Longney (Fig. 1).

The Wentlooge Formation

These wetlands are supported by a post-glacial sediment sequence that is, on average, 10–15 m thick, largely comprising soft estuarine mineral sediments and peats and which arguably contains some of the most extensive and complete exposures of post-glacial shallow-marine sediments in the British Isles (Allen 1990c). In the Severn Estuary they are described as the Wentlooge Formation (Allen 1987a; Allen & Rae 1987; Allen 1990b) and are essentially sediments of high tidal mudflats and salt-marsh environments with peats overlain with recent estuarine alluvium. This sequence is remarkably uniform in general character throughout the Avon, Gwent, and Somerset Levels and has been informally divided into three principal component elements (see Fig. 2): lower Wentlooge - thin gravels and sands, organic-rich silty sands and ‘peats’ grading into bluish grey-greenish grey sand to clayey silts. The ‘sands’ (some contain Ipswichian shells) probably relate to cold climate estuarine sand deposition (Andrews et al. 1984) middle Wentlooge - occur at about -1 to +3–4m OD comprise alternation of organic (peat) beds with often laminated greenish grey sandy to clayey silts (c. 5500–250 cal BC (6500–2200 BP), with the organic facies representing development from wetland carr woodland with alder and oak to reed swamp upper Wentlooge - thick greenish grey sandy to clayey silts which grade into green-brown mottled silts which are unlaminated and poorly stratified

1 Note: Throughout this paper post-glacial refers to post-Devensian glacial and, similarly, ‘alluvium’ is used as shorthand for estuarine alluvial or intertidal sediments.

The construction of drainage systems and sea defences over the last two millennia has halted the development of the Wentlooge Formation. On the seaward side of the sea defences three other estuarine formations have developed: The Rumney, Awre, and Northwick formations (Allen & Rae 1987; Allen 1987a; Allen 1988), which are largely medieval to post-medieval in date.

These post-glacial sediments represent fluvial and marine environments which accumulated during major marine transgression. Human activity increasingly influenced the character of these ‘marginal’ wetlands being created by a ‘lengthy interplay between human and natural forces’ (Allen 1990c). As such ‘Much can be learned from the sedimentary and archaeological record about the way in which coasts and estuaries have varied in the past, the way in which man has responded to or caused such changes and with what consequences for both himself and for the ‘natural’ environment’ (Pye & Allen 2000, 1). The important and significant evidence of prehistoric communities from the Somerset Level (Coles & Coles 1986) has, with little doubt, distracted investigators away from the potential and significance of the wetlands as a whole (Allen 1990c, 33).

The Levels form a landscape which superficially appears flat and low-lying (generally less than 7.5 m OD) and uniform, but which may be divided into a number of character areas (cf. Rippon 1994). The-low-lying nature, fine-grained soils, flatness, and susceptibility to flooding provide wetland sites of ecological importance. Since 1987 a number of Site of Special Scientific Interest have been registered in the Levels, and a series of important archaeological sites have been discovered with well-preserved palaeo-environment material.

The Avon Levels

The opportunity to examine the evolution of the Avon Levels (Fig. 3) was presented as a result of archaeological work conducted prior to the construction of the new road bridge across the Severn Estuary (the Second Severn Crossing) and the approach roads from the south (M5) and east (M4). Initial evaluation work isolated ten archaeological or palaeo-environmental sites which were excavated in 1992–4. Until that time the Avon Levels were not well known archaeologically (Barnes 1993, 5), but as the archaeological and palaeo-environmental profile has been raised since the fieldwork (1992–3), further investigations have started to reveal a more consistent, if ephemeral, presence of archaeological sites buried within the alluvium (eg, Locock et al. 1998; Locock 1999; and for the Welsh side of the estuary see, for instance Bell et al 2000; Rippon 1996). An important medieval landscape exists (see Rippon 1997) in both physical and documentary evidence as surface features and sites, as exemplified by investigations at, for instance, Rockingham Farm (Locock & Lawler 2000).

In this paper, we present the results of the most systematic investigation of the middle/upper Wentlooge Formation undertaken to date. We deal with four main investigations of sediment profiles up to 5 m thick which are reviewed in relation to previous work and related spatially through a large-scale auger survey totalling c. 3.6 km. We are primarily concerned with the middle and upper Wentlooge Formations, the peat within which is dated in the Avon Levels from c. 4500 cal BC (5600 BP), interrupted by episodes of minerogenic sedimentation. Peats were succeeded by the upper Wentlooge clays at c. 1500–500 cal BC. This work does not deal with the later Rumney, Awre, and Northwick formations (sensu Allen 1988) which are restricted to the coastal margins and estuarine fringes seaward of the seabank. The more overtly archaeological aspects of these sites are

6 presented elsewhere (Gardiner et al. 2002), as summaries of the investigations and models of the social and community use of the landscape.

Background

The importance of understanding the precise nature of the estuarine environment is that it allows us to develop a picture of the sedimentological, physical, ecological and archaeological evolution of the Avon Levels and the wider Severn Estuary Levels. Sedimentological and geoarchaeological approaches have tended to concentrate on defining the origin of the sediment (marine, fluvial, estuarine), on modelling sedimentation, accretion and salt-marsh growth, and in examining changes in relative sea-level (RSL) and the effects of autocompaction.

Most of these studies define the nature of the developing landscape and surface morphology as either high intertidal zone environment (ie, mudflat to salt-marsh) or freshwater fen. There is certainly a shallow gradient between mudflat and salt-marsh and the definition of each environment is not as significant in geomorphological terms as it is to archaeology. The precise nature and location of these types of habitat are directly pertinent to the exploitation, settlement, use, and longevity of human activity in prehistoric and historic times and the present. It is precisely these aspects that we attempt to lay the foundations for here, and are explored in archaeological terms by Gardiner et al. (2002).

7 W E Hallen & Northwick Awkley Lane Awkley Awkley Interface Awkley Vimpenny's Lane Vimpenny's High tides

Upper

Middle

Low tides Lower Keuper Marls & Tea Green Marls

Upper Wentlooge Middle Wentlooge Lower Wentlooge Colluvium Peat Easturine Sands & Gravels

Figure 2. Sedimentary features comprising the Wentlooge Formation

8 353000 354000 355000 356000 357000 358000 359000 360000 Estuarine Alluvium Head Keuper Marl (Red Marl) 188000 Tea Green Marl Limestones & Sandstones Major Excavation Environmental Sequence 187000 Auger Transect Awkley Lane

Awkley Interface Northwick 186000

185000 ae

stol Ch nn l 184000 Bri

183000

Vimpennys Lane 182000

et exposed on fo eshore Pa r

181000

Hallen

180000

179000

178000

177000

Figure 3. The geology of the Avon Levels showing the location of the archaeological and sedimentological investigations

9 THE INVESTIGATIONS

The Research/fieldwork: methods of investigation

The investigations summarised below were conducted as an archaeological and palaeo- environmental programme of investigations (see Barnes 1993). All investigations were confined to the route of the M5 and M4 link roads, and their precise siting of was determined through observations in geotechnical pits, documentary research (Porter 1990; Russett 1990), and evaluation trenching (Lawler et al. 1992). Archaeological work involved the excavation of 10 sites, a detailed earthwork survey along the 14 km motor corridor, transplot of aerial photographs covering 54 km2 of the Avon Levels, a summary documentary survey, and a detailed photographic survey of the village of Redwick.

Investigations which provide information relevant to the physical and social evolution of the Avon levels are two deep sondages exposing long sediment sequences at Awkley Lane and Vimpennys Lane, two shallow sequences exposed in relation to excavations of archaeological sites at Hallen Marsh and Northwick, and sequences on the wetland margins at the foot of Awkley Hill at Awkley Interface (Fig. 3).

At two of these locations (Hallen Marsh and Northwick), full-scale open area archaeological excavation was conducted to reveal Iron Age and Romano-British activity in plan. Each was buried beneath relatively shallow (c. 0.5 m) alluvium. At two locations in the levels deep sondages up to 5 m deep were opened by machine to investigate the middle and upper Wentlooge sequence. These were conducted by opening trenches c. 10 m square and exposing the sedimentary sequence in a series of steps (see Plates 2 and 3). A third investigation comprising a series of machine cut trenches exposed the deposits at the foot of Awkley Hill revealing the alluvial–colluvial interface.

In all cases the sediments were exposed in section, drawn, and recorded, and in most cases full suites of samples were taken for the investigation of the pollen, diatoms, snails, waterlogged plant and insect remains, and soils. Where archaeological activity was encountered this provided some chronology, sparse charred remains, and discarded animal bones. Where organic and peat deposits were encountered samples were taken for radiocarbon dating. Samples were also taken through two sequences for archaeomagnetic dating of the fine-grained middle and upper Wentlooge sediments. Following assessment of the survival and suitability of various microsfossils and datasets to aid in the understanding of the evolution of the Avon wetland (Wessex Archaeology 1994; 2000), a selected programme of analysis was undertaken. Full analytical methods adopted are presented in the Appendix.

In addition to these major field interventions and analyses, the upper 2 m of the Wentlooge sequence was mapped by a large-scale hand auger survey comprising 411 records. This was conducted in 6 transects (Fig. 3) with augerholes at 50 m intervals covering an area of 480,000 square metres and the augered transects themselves represent a distance of about 3600 m. The principal aim was to map and model the upper Wentlooge sequence, as studied in detail in the excavated exposures, to provide the basis of a three dimensional model of the evolution of the Avon levels. Understanding of the physical evolution provides the basis for the examination of models of the archaeological use of the Levels as a whole (see Allen and Gardiner in Gardiner et al. 2002). The middle and upper Wentlooge Formations were of specific archaeological interest as these later units are better preserved and more closely integrated with wider scale patterns of exploitation and use of the Levels in the later Holocene (from c. 1000 BC).

10 The physical and vegetational evolution of the Avon levels provides an important contribution to the development of the Severn Estuary environment. It also provides the opportunity to examine the human interpretation and use of the landscape. Because of the three dimensional interpretations over a relatively wide area (c. 6 km2) afforded by the detailed and dated palaeo-environmental studies, in combination with the large-scale auger survey and sediment mapping, this information can be used to examine the resource potential for human occupation. It can help elucidate and explain the nature of human activity, the onset, and abandonment of both individual settlements and of phases of settlement activity. In so doing we are able to use this information to provide models of exploitation of the area in much wider social and geographic terms.

The sites

The sediment sequences and analysis from the five sites are presented individually before the data is combined and used in wider interpretations of the history of the Avon wetlands as a whole. The sites are as follows (see Fig. 3).

Awkley Lane: Deep environmental sequence in a back fen location to the east of the Avon levels near the foot of the rising land of Awkley Hill. Profile almost 5 m deep comprising a sequence of freshwater peats overlain by estuarine silts.

Vimpenny’s Lane: Deep environmental sequence nearly 1 km from the higher land to the east, but to the west of locations with deeper back fen freshwater peats. The 2.8 m exposed sequence included mainly estuarine silts and clays with a dated detrital Phragmites peat lens.

Hallen Marsh: A site situated well into the Avon wetlands to the southern end of the study area (Fig. 4) at which large scale archaeological investigations of Iron Age (100 BC) settlement was undertaken, but a short sequence of silts and clays up to 1m were exposed and examined. It is known that peats occur at depth but were not encountered in the archaeological investigations.

Northwick: Archaeological excavations of Romano-British (1st century AD) field systems in the centre of the widest expanse of wetlands. Only the upper 1 m of the sedimentary sequences was exposed and examined.

Awkley Interface: On the western slopes of Awkley Hill where the fine-grained deposits of the wetlands meet the terrestrial colluvial deposits relating to the surrounding higher ground. Limited evidence of localised Roman activity on a bench at the foot of the slope was recorded.

1. Awkley Lane

The sequence

Evaluation within the vicinity of Awkley Lane (ST 5932 8638) produced evidence of relatively deep, stratified and well preserved freshwater peats (Lawler et al. 1992, 57). This sequence was examined in more detail by the excavation of a trench 11 m square and about 4.5 m deep; a further 0.35 m was hand excavated through the base of the trench to sands of the lower Wentlooge Formation. The deposits recovered can be paralleled with those recorded in evaluation (Lawler et al. 1992, 57–8), but with greater differentiation in the deposits being recorded during excavation.

A series of fine grained, greyish brown silty clay, alluvial deposits (102–108) with very low magnetic susceptibility (4 to 6 SI10-8Kg-1) dominated the upper 1.8 m of the profile (6.41 to 4.31m OD). Between 4.31 and 4.16 m OD two putative stabilisation horizons (107 and 109) were recognised in the field on the basis of their darker hue (possibly organic content), but this was not reflected in the magnetic susceptibility profile (Fig. 4), although a gradual rise in enhancement was seen with depth. Below 4.16 m OD a further 1.4 m of light grey (10YR 6/1) mixed silty clay alluvium with a greyish and blue hue were present. The magnetic susceptibility profile shows much higher levels averaging c. 18 SI10-8Kg-1. Enhancements of up to 38 SI10-8Kg-1 at about 2.9 m are not reflected in the sedimentary record and may be due to post-depositional effects of waterlogging. From 2.30 to 3.65 m OD (110) fine (1–2mm) weak laminations with occasional sand and silt partings indicate a series of flood events (flood couplets) or possibly indicates a former tidal boundary, with evidence for the presence of freshwater (J.R.L. Allen, ers, comm.). A large isolated and waterlogged and weathered ‘bog’ oak log near the base of this unit dated to 2460–2140 cal BC (3816±50 BP, NZA-12532). This probably represents higher energy flood detritus. Occasional fine organic matter, possibly Phragmites stem fragments, are distributed throughout this alluvium.

The sediment changes abruptly to a series of humified detrital monocot and wood peats, the surface of which displayed a distinct ridge in the recorded section (Fig. 4). The ‘ridge’ was clearly marked by a thin band (20 mm) of loose, black peat, presumably a desiccated surface, beneath which was 80 mm of alluvium hardly distinguishable from that above. This band rises to 3.04 m OD but dips away to the west, indicating some truncation and modification of the upper peat surface (see Fig. 4).

Nearly 1.4 m of stratified peat occurs below 2.96 m OD. The upper deposit peats are black, well- humified fen peat (113) over wood peat (121), and a dense black compressed and partially dried band (120) at 4.10 m OD seals a black well humifed peat (115). Reddish, oxidised humified fen peat (116 and 118) occurs at the base and is separated by a very thin (20 mm) lens of grey silty alluvium at 1.99 m OD. The base of the peats were exposed at 1.57m OD by a small hand excavated pit in the base of the trench revealing an inorganic grey silty sand below.

A full suite of samples was taken from the deposits, including mollusc columns, pollen monoliths, samples for particle size analysis and samples for investigation of palaeomagnetic dating.

Chronology

The surface of peat at 2.62m had been dated in the evaluation exercise to 2920-2600 cal BC (Lawler et al. 1992, 115). A series of nine AMS results from the peat and the bog oak log provide a sequence for the rate of peat accumulation (Table 1; Fig. 5). The start of peat accumulation (1.44m OD) is dated to 4530–4350 cal BC (NZA-12774 5603±50 BP). The formation of 0.9m of humifed monoct (?fen) peat occurred steadily over a period of about 1½ millennia until the wood peat (121) at 3.04 to 2.42 m OD: 3090–2700 cal BC (NZA-12528 4286±55 BP). A depth of c. 1.3 m of wood peat developed in a period of a couple of centuries between 3090–2700 cal BC to 2860–2470 cal BC. Alluviation occurred and was then followed by a thin a final peaty horizon (111, dated to 2900- 2600 cal BC) prior to return to salt-marsh and estuarine conditions and a fully alluvial sequence.

The oak log, contained in excess of 40 rings but no match for the pattern could be established in dendrochronological analysis. It was concluded that that the tree had been subject to considerable

12 WA Awkley Lane GGAT G030

Mollusc & soil column

6 m O.D. Pollen Zones

104

5 m O.D.

105 5 106 107 108 109 4 m O.D.

110

3 m O.D. Log

112 111 4190+_ 70BP 113 4190+_ 60BP 113 0 10 20 30 40 50 x10-8 SI/kg 121 3

114 120 2 m O.D. 115 116 2

118 Alluvial silty clays (brownish red hue) Insect sample 1 Alluvial silty clays (bluish grey hue) Plant macro fossils sample ?Stabilisation horizon C14 sample 119 Peat (wood peat vs humified peat) 01 m Sandy alluvium

Figure 4. Awkley Lane Wentlooge Formation profiles

13 WA Awkley Lane

110 3816+_ 50BP 3m O.D. 3816+_ 50BP Log 3991+_ 55BP 111 3991+_ 55BP 4044+_ 50BP 112 4044+_ 50BP 3885+_ 45BP 113 3885+_ 45BP

113

4286+_ 55BP 4286+_ 55BP 121

114 _ 4745+_ 50BP 4745+ 50BP 120 115 2m O.D.

_ 116 4683+ 55BP 4683+_ 55BP

118

5603+_ 50BP 5603+_ 50BP

Insect sample 119 Plant macro fossils sample C14 sample Alluvial silty clays (bluish grey hue) Peat (wood peat vs humified peat) Sandy alluvium 1m O.D. 01 m 5000BC 4000BC 3000BC 2000BC Figure 5. Radiocarbon determinations in relation to their OD and the Middle Wentlooge wood and fen peat 14 ecological stress during growth, and this is why the ring pattern does not fit in with any dated reference chronologies. It is likely that the distressed growth pattern occurred because local growing conditions were unfavourable, and this stress may be due to the proximity of trees to saline water and saline water ingression.

The inorganic fine-grained sediments contain no organic matter to provide a radiocarbon date. Even the possible stabilisation horizons do not contain sufficient carbon to provide a radiocarbon result. The sequence was, however, sampled for palaeo-magnetic dating (GeoQuest 1993). Sediments from this sequence provided a coherent record or geomagnetic secular variation which is particular well preserved below the gleyed zone (ie, in the weakly laminated silty clay (layer 110) between 4.10 and c. 3 m OD. Demagnetisation tests confirmed that the sediments contain a stable remanent magnetism and there is a good agreement between replicate samples. Correlation between the profile and master curve is complicated by changes in the rate of deposition and the possibility of hiatuses and removed sediment. Absolute dates originally provided (GeoQuest 1993) have been revised in the light of revision of the ‘master curve’ and radiocarbon determinations from the basal peats. As a result M. Noel (pers. comm.) is able to indicate that deposits near to top of the profile (base of layer 103 at 5.7 m OD) are not younger than c. AD 500. Two index points were indicated as i) about 5.1 m OD at the top of layer 105, c. 0 BC, and ii) about 4.5 m OD at the base of 106 and top of 107 c. 500 BC. This latter date occurs immediately above the series of stabilisation or flood event layers (107, 108, and 109), at the base of the alluvium of a reddish brown hue.

Table 1. Radiocarbon determinations and calibrated results from the Middle Wentlooge Formation at Awkley Lane context context type depth OD material lab no C13‰ result BP date Cal BC Awkley Lane 110 bog oak 3.00 wood NZA-12532 -25.7 3816±50 2460-2140 111 peat band 2.92 humic acids NZA-12590 -26.63 11,350±120 11,840-11,070 111 peat band 2.92 plant matter NZA-12754 -26.63 3991±55 2840-2310 GGAT top of peat 2.62 peat GU-3119 -28.3 4190±70 2910-2580* GGAT top of peat 2.62 wood alder GU-3120 -28.3 4190±60 2900-2590* 113 top peat 2.78 peat NZA-12533 -27.79 4044±50 2860-2470 113 peat 2.64 peat NZA-12534-27.48 3885±45 2470-2210 121 peat 2.42 wood alder NZA-12528-27.68 4286±55 3090-2700 120 band 2.14 wood peat NZA-12529-27.48 4745±50 3640-3380 117/8 top peat 1.82 twigs NZA-12530 -26.78 4683±55 3630-3360 elm decline 4800 BP 118 base of peat 1.44 female alder NZA-12774 -26.57 5603±50 4530-4350 cones

* GGAT 1992

Depth (m) OD Layer Description Comment 0–0.10 6.41.–6.31 100 Dark brown (10YR 3/3) stonefree clay loam Topsoil 0.10–0.25 6.31–6.16 101 Greyish brown (10YR 5/2) stonefree silty clay B horizon 0.25–0.55 6.16–5.86 102 Pale brown (10YR 6/3) stonefree clay, rare medium Upper Alluvium fleshy roots and vertical voids 0.55–0.70 5.86–5.71 103 Greyish brown (10YR 5/2) stonefree clay with rare Upper Alluvium mineral (?Fe) staining and gleying along vertical rootlet channels 3mm wide, diffuse/gradual boundary 0.70–1.20 5.71–5.21 104 Greyish brown (10YR 5/2) stonefree clay with Upper Alluvium frequent mineral staining, gleyed, no sedimentary structures evident, diffuse/gradual boundary 1.20–1.70 5.21–4.71 105 Brown (10YR 5/3) stonefree silty clay with patches Upper Alluvium of light grey material, gleyed, no sedimentary structures evident,. 1.70–1.80 4.71–4.61 106 Grey (10YR 5/1) stonefree silty clay, no sedimentary Upper Alluvium structures evident 1.80–1.95 4.61–4.46 107 Dark grey (10YR 4/1) stonefree clay with up to 10% ?Stabilisation flecks of black materail. horizon 1.95–2.10 4.46–4.31 108 Grey (10YR 5/1) stonefree clay with occasional (5%) Grey Alluvium flecks of black material 2.10–2.30 4.31–4.10 109 Dark grey (10YR 4/1) stonefree silty clay with up to ?Stabilisation 10% flecks of dark material. horizon 2.30–3.65 4.10–2.75max 110 Light grey (10YR 6/1) mixed clay with very Grey Alluvium occasional (2%) flecks of dark organic plant material (?Phragmites). Thinly (1-2mm) and weakly laminated with occasional sand and silt partings (flood couplets). A large log (oak) is preserved in layer near its base. 3.35–3.37 2.92–2.90 111 Black (10YR 2/1) peat. Thin layer of loose black Peat lens desiccated peat following contour of ridge of peat 113 below. 3.37–3.45 2.92–2.86 112 Light grey (10YR 6/1) stonefree clay. Layer of Grey Alluvium. alluvial deposit lying over ridge of peat 113 3.45–3.75 2.86–2.56 113 Black (10YR 2/1) humified peat, very little Peat waterlogged wood/twigs etc. Layer of peat with a ridged on the south side of the section 3.75–4.05 2.56–2.26 121 Black (10YR 2/1) peat with up to 20% wood Wood Peat inclusions 4.05–4.12 2.26–2.08 114 Dark reddish brown (5YR 2.5/2) peat with occasional Peat (5%) wood inclusions, clear smooth boundary. 4.12–4.14 2.08–2.06 120 Black (10YR 2/1) peat. Thin layer of dark peat Peat running across section, clear smooth boundary 4.14–4.22 2.06–1.98 115 Black (5YR 2.5/1) peat with occasional (up to 5%) Peat small pieces of wood. 4.22–4.38 1.98–182 116 Dark reddish brown (5YR 2.5/2) peat with occasional Peat (5%) wood inclusions, one moderately sized piece of wood. 4.38–4.40 1.82–1.80 117 Grey (10YR 5/1) stonefree clay. Thin alluvial layer. 4.40–4.80 1.80–1.40 118 Dark reddish brown (5YR 2.5/2) peat with occasional Peat wood inclusions. 4.80+ 1.40+ 119 Grey silty sand. Sandy

Sampling the sequence

The minerogenic and peat deposits below the first step (Plate 1) at c. 3.7 m OD were sampled in monolith tins to facilitate subsampling for pollen and diatoms and one portion of the upper sequence encompassing a putative buried soil (107) was also sampled in a monolith tin. The upper sediment sequence and the gap between the two sets of monoliths was filled with sub-samples from soil sequences of bulk samples removed for land snail and grain size analyses (see Fig. 4). The minerogenic deposits below 103 (5.28 m OD) were sampled as contiguous disturbed samples for

16 Plate 1. Sampling the minerogenic sediment Plate 2. Awkley Lane excavation showing the fully sequence in monoliths at Awkley Lane excavated and exposed upper and middle Wentlooge sequence

Awkley Lane Clay (100%)

1119 1125 1100 1118

50% 50%

Clay

Silt (100%) Sand 50% Sand (100%)

Figure 6. Awkley Lane; particle size triangle

17 snails and grain size analysis. In addition three groups of three bulk samples were taken from top, middle and bottom of the recovered peat for waterlogged plants and insects. Most of the minerogenic deposits were also sampled for archaeomagnetic dating.

Textural features of the minerogenic sequence by J.R.L. Allen

A series of 26 samples was analysed for grain-size distributions and the triangular diagram (Fig. 6) shows values of the clay-silt-sand ratio. The methods used are detailed in the Appendix. The triangular diagram (Fig. 6) show that the sediments range from clayey-sandy silts to clayey silts. Falling into a finer (top of layer 106–105; 4.68–5.28 m OD) overlying a coarser (layers 112–106; 2.68–4.68 m OD) subset, they form a distinct trend on the triangular diagram, similar to trends noted at Hallen Marsh and Vimpenny’s Lane (see below).

The silts of the stratigraphically lower subset are surprisingly coarse grained for such an inland site. The grain-size distributions show strongly defined modes which typically exceed the means in value, well-developed coarse tails, and noticeable proportions of sand. The resolution provided by the sampling interval is not great, but the presence of three textural trends may be suggested: fining- up (112 to middle of 110; 2.68–3.18 m OD), coarsening-up (middle 110; 3.18–3.58 m OD) and, finally, fining-up (top of layer 110 to 106; 3.58–4.59 m OD). There is nothing texturally to distinguish the two soils considered to be present in this part of the sequence.

A sharp and substantial reduction in grain size occurs between samples in 107 at 4.49–4.68 m OD, interpreted as a possible soil reminiscent of that reported from Vimpennys Lane. The overlying silts, belonging to the finer subset, display a strong coarsening-up trend, as is also seen at Vimpennys Lane. At the latter site, however, where a longer sequence terminating in a soil was sampled above the textural break, the coarsening-up pattern is followed by a short fining-up trend.

The unusually coarse texture of the lower silts suggests either that they were deposited close to a major tidal creek which reached deeply back into the marsh; that they formed near a no-longer recognisable stream issuing from the hinterland; or that they include hillwash from nearby slopes underlain by head/bedrock. Under the first possibility, the textural trends may suggest changes in either the position of the distant marsh edge from which the creek extended and/or the rate of sea- level change. Under the second and third possibilities, the trends could register variations in sediment yield from the nearby slopes induced by changes in climate and/or vegetation cover.

The dramatic reduction in grain size between samples at 4.59 m OD points to a sharp change in sedimentary conditions, and the sediments are not what would be expected at such an inland site. Either the postulated creek/stream suddenly ceased to be active, leaving only distant sources of sediment, or the site became isolated in some way from the neighbouring slopes. The subsequent coarsening-up trend suggests an episode of marsh- edge retreat and/or positive sea-level tendency.

Pollen analysis by Robert G. Scaife

The Awkley Lane profile is the deepest at c. 4 m and acts as a key to establishing a local and regional pattern of vegetation and environmental change. A total of 65 pollen samples, of which 62 contained pollen, was examined from the profile. Material was obtained in the field using

18 monolith profiles taken directly from the excavated section which were sub-sampled for analysis and stratigraphical description in the laboratory of Wessex Archaeology.

A total of five pollen assemblage zones (p.a.z.) and a number of pollen assemblage subzones (p.a.s.z.) has been recognised at Awkley (Fig. 7). These have been delimited and characterised from the base of the profile at 1.44m OD upwards as follows.

AWK:1 1.44–1.66 m OD, detrital peat/lower fen peat. Ulmus-Tilia-Quercus-Corylus avellana type-Alnus. APF are from 138,000 grains/ml at the base to 215,000 grains/ml at the top. Tree, shrub pollen are dominant (to 75% and 40 % resp.) along with fen/marsh taxa which are dominated by Alnus (to 65%). This basal zone is characterised by highest values of Tilia (to 33%), Ulmus (to 14%), and Fraxinus (4%). Other trees include sporadic occurrences of Betula, Pinus Populus, Fagus, and Taxus. Shrubs are dominated by Corylus avellana type (declining from 40%). There are few herbs in this zone compared with subsequent zones (10% NAP) with only small percentages of Poaceae (6%) being of note. A single cereal type grain is noted. Alnus is the dominant wetland taxon (to 65%) with Cyperaceae (to 20%). Spores comprise Dryopteris (monolete) type (c. 20–25%) and small numbers of Polypodium vulgare (<5%).

Sampling gap in profile from 1.54 to 1.78 m OD.

AWK:2 1.66–2.53 m OD, detrital lower fen &wood peat. Quercus-Corylus avellana type-Dryopteris type. APF values attain highest values of up to 1,730,000 grains/ml in the lowest level of this zone and subsequently decline upwards. In the intervening gap in the sample profile between 1.54 m and 1.78 m OD there are reductions in Ulmus, and Tilia and an increase in Corylus avellana type. Fraxinus (<5%) is present throughout. Three pollen assemblage sub zones have also been recognised. Sub-zone-b (1.86–2.21 m OD) shows further reduction in Ulmus (to absence), Tilia (to 5%) and Corylus type (from highest values of 70% to c. 25%). Sub-zone-c wood peat (2.21–2.53 m OD) shows some expansion/regeneration of the these elements; especially Ulmus to 10–12% and Corylus to c. 40%. There is a minor expansion of Fraxinus and Taxus is also noted. Herb pollen increases markedly with increased taxonomic diversity and particularly expansions of Poaceae (to 59%) peaking in sub-zone-a (to 59%), Asteraceae types-Aster type in sub-zone-a (peak to 17%) and Bidens type in sub-zone-b (6%). Plantago lanceolata (<5% in sub-zones a and b), Chenopodiaceae (8%) and Apiaceae type 3 are of note. Alnus values are highest in sub-zone-a (to 90%) but decline in sub-zone-b (av. 20–25%) with a peak of 65% at 2.38 m OD in subzone c). There is a minor expansion of Salix and also peaks of Cyperaceae in sub-zones a and c (40% and 30% respectively). Typha angustifolia/ Sparganium and T. latifolia become more important in this zone. Aquatic megaphytes include a peak of Lemna at the top of sub-zone-a. Spores of ferns are dominated by Dryopteris type (monolete spores) which increase to high values (80–90%) in this zone. Of note is a peak of Thelypteris palustris at the top of sub-zone-c (to 20%).

AWK:3 2.53–2.82 m OD, humified detrital peat (context 113). Betula-Corylus avellana type-Ericales. APF Values increase to between 220,000 and 855,000 gains/ml. This zone is characterised by a sharp decline in herb percentages (especially Poaceae) and a further decline of Tilia (to <5%) and a sharp reduction in Dryopteris (monolete) type. Quercus also declines to 10–15%. These are in response to a very marked expansion of Betula which is dominant peaking to 80%. Of note are peaks of Erica (5%) and Calluna (to 18%). Pinus, Ulmus, Tilia, and Fraxinus remain consistent with small values (<5%). Sporadic Fagus is noted. A possible sub-zone (AWK:3-b) change occurs between 2.64 and 2.58m OD where Quercus and Corylus type expand in contrast to a sharp declines in Betula, Erica, and Calluna. Betula re-expands to high values but less than earlier in this zone. NAP decline with sharp reduction in Poaceae, to low levels. There is, however, a minor peak of Poaceae between 2.64 m OD and 2.60 m OD (sub-zone 3-b). There are also declines in Plantago lanceolata, Apiaceae, and Asteraceae types noted in the preceding zone. Alnus is the principal marsh autochthonous taxon (20–30%) but reduced over the preceding zone AWK:3. Other wetland taxa are somewhat reduced over AWK:3 with only sporadic occurrences of Salix, Typha angustifolium/Sparganium type, and other aquatic megaphytes. Of note are peaks of Osmunda regalis and Sphagnum. Spores of Dryopteris type are markedly reduced over the high values present in AWK:2.

AWK:4 2.82–3.95 m OD, upper Wentlooge sediments. Quercus-Corylus avellana type -Chenopodiaceae - Pteridium aquilinum-Polypodium-Pre-Quaternary Palynomorphs. APF values decline with the stratigraphical change from peats to minerogenic sediments. This local pollen assemblage zone is defined by sharp reductions in Betula, and expansions of Cyperaceae, spores of Pteridium aquilinum, Monolete/Dryopteris type. Polypodium vulgare and miscellaneous microfossils (Dinoflagellates and derived, pre-Quaternary palynomorphs). Absolute pollen frequencies decline sharply with change to minerogenic sediments from underlying peats. High values of Betula in AWK 3 decline very sharply to <5% and subsequent absence. There are also substantial reductions in Erica and Calluna and Fraxinus. Quercus attains highest values (65%) with some expansion of Pinus (to 8%). Tilia remains (low values) but there is an increase in its degraded grains from the base of this zone associated with the

19 change to minerogenic sediments. Quercus is dominant (to 65%) with Corylus avellana type (20–30%). NAP expand after the low values of AWK:3 peaking to 75% of tdlp. This expansion comprises increases in Chenopodiaceae (15–20% with a peak to 68%), Poaceae (to 30%), Plantago lanceolata (<5%), Cereal type (<5%). Within the marsh category, Alnus (to c. 20%) is consistent throughout. Sedges show some increase along with Potamogeton type. Spores become important with substantial expansions of Pteridium aquilinum, Dryopteris type (monolete) and Polypodium vulgare. Pre-Quaternary palynomorphs are incoming and attain high values. Dinoflagellates of Holocene or possibly pre-Quaternary origin also become important.

AWK:5 3.95–5.24 m OD, upper Wentlooge sediments. Quercus-Corylus avellana type-Poaceae-Cyperaceae. APF values remain low (especially at the lower zone boundary where some levels did not produce pollen) but slightly increased over AWK:4 (ranging from c. 4000 to c. 40,000 grains/ml.). This zone is delimited by reduction in Pinus to sporadic occurrences and some expansions of Betula (including a peak at 4.66 m OD) and Cyperaceae (to 50%). Quercus (20–30%) and Corylus avellana type (to 30%) remain the dominant trees and shrubs with sporadic occurrences of Fraxinus, Fagus, Taxus and Ericales after absence in the preceding zone. Herbs remain important (to 70%). Chenopodiaceae remain consistent from the preceding zone. Poaceae becomes dominant (to 50%) with an increase in the diversity of types – Cereal type, Secale cereale (single grain), Fabaceae spp., Polygonaceae spp., Plantago lanceolata (to 10%), and Asteraceae types (Lactucoideae, Centaurea spp., Bidens, Anthemis type, Cirsium type. Of note are halophytes including P. maritima type (1%), Armeria ‘A’ line, Armeria ‘B’ line, and possibly Aster type and large (non-cereal Poaceae). Within the fen/aquatic category, there is a significant increase in Cyperaceae (to 50%). Alnus remains consistent with small values (average 10%). There are also sporadic occurrences of Salix, Caltha type, Littorella uniflora, cf. Butomus umbellatus, and Alisma type, Typha latifolia, Typha/Sparganium, and the spores of Osmunda regalis. Aquatics are present include-Myriophyllum, Callitriche, Lemna, Potamogeton type (possibly Triglochin) and occasional cysts of algal Pediastrum. Spores of Pteridium aquilinum, Dryopteris (monolete type), and Polypodium vulgare and Miscellaneous types (Palynomorphs and Dinoflagellates) remain as AWK:4. There is some increase in Sphagnum although percentages remain <5%.

AWK:5-a. At a depth of 4.66 m OD there is a shallow more organic horizon. This appears to be a stabilisation horizon (? soil) within the otherwise salt-marsh silts. Palynologically this is characterised by a small expansion of Alnus (30%) and slightly above by increase of Betula (15%).

Prehistoric vegetation and environmental changes Interpretation of the pollen data at Awkley must take into account the taphonomy of the pollen recovered from the lower organic peats which contrast with the largely inorganic and fluvially derived upper sediments. Furthermore, differentiation between the on-site (autochthonous) vegetation and the vegetation communities of the adjacent well drained terrestrial soils is important.

The Lower Fen Peat: In the basal and, therefore, earliest pollen zone (AWK:1), the peat accumulated under a typical alder fen carr woodland (Alnetum glutinosae). This somewhat dry fen carr supported a ground flora of ferns (eg, Dryopteris felix femmina and D. felix mas-female and male fern) and sedges such as Carex paniculata (tussock sedge). Taxus baccata, although only a single record, was typically a tree of such dry fen carr woodland (Godwin 1975a; Scaife 2000). This fen carr apparently developed on a previously wetter, fluvial or marine sandy deposit. This is comparable with the typical Somerset Levels hydroseral succession and is seen throughout low-lying regions of the Bristol Channel.

The location of the Awkley Lane site in relatively close proximity to higher, dry-land areas has resulted in a clear representation of the vegetation of these areas. In zone AWK:1, Tilia (Tilia cordata; small leaved lime) attains high pollen values which suggest local dominance of this largely under-represented pollen taxon (Andersen 1970; 1973). Its proximity and importance is also attested by the presence of fruits in the peat (see below). The importance of lindens in the landscape is discussed below. Other elements of the dry-land flora comprise especially Quercus (oak), Ulmus (elm), Fraxinus (ash), and Corylus avellana (hazel). Whilst Tilia was clearly the dominant woodland element, these tree and shrub taxa were also of importance in the local environment. It is possible that oak, ash, and hazel formed parts of the dry carr woodland or woodland fringing the edge of the mire – perhaps on the thicker, down-slope soils.

20 A

3816 + 50 BP 3991 + 55 BP 4044 + 50 BP 3885 + 45 BP

4285 + 55 BP

4745 + 50 BP

4683 + 55 BP

5603 + 50 BP

Figure 7a. Awkley Lane; pollen diagram.

21 B

Figure 7b. Awkley Lane; pollen diagram.

22 C

Figure 7c. Awkley Lane; pollen diagram.

23 D

Figure 7d. Awkley Lane; pollen diagram. Note that alder (Alnus sp) is included with fen species.

24 E

Figure 7e. Awkley Lane; pollen diagram.

25 F

Figure 7f. Awkley Lane; pollen diagram.

26 Ilex aquifolium (holly) is recorded. It is usually extremely poorly represented in pollen spectra and thus, as with lime, implies some importance. Alternatively, and very likely, these woodland trees formed lesser elements within the dominant lime woodland. Radiocarbon dating confirmed the age of this basal peat as 4530-4350 cal BC (NZA-12774, 5603±50 BP), ie, late Atlantic age, middle Holocene (Flandrian Chronozone II) or, at latest, very early Neolithic (Flandrian Chronozone III). In either case, the pollen is a representation of the dominant and maximum extension of middle Holocene woodland prior to any extensive human influence and deforestation. This is also reflected by the small representation of non-arboreal pollen in the Awkley profile as a whole.

In pollen zone AWK:2, the importance of lime and elm is reduced and there are increasing numbers of herbs. It is very unfortunate that there is a gap in the sample record, since evidence of a phase of important ecological change has been lost. However, it is suggested that this zone AWK:l/2 transition is the early part of the prehistoric elm decline at c. 5000 BP, or slightly later if dates for the Somerset Levels pertain. This is confirmed by the radiocarbon date of 3630-3360 cal BC (NZA-12530, 4683±55 BP) near the base of this zone. Further reduction in elm pollen at 1.86 m OD (sub-zone a and sub-zone b) and sharply expanding values of Poaceae (grasses) with sporadic occurrences of cereal type pollen, Plantago lanceolata (ribwort plantain), and peaks of other herbs including Asteraceae (Bidens type and Aster type), Urtica (nettle and pellitory), Chenopodiaceae (goosefoots and oraches) are all indicative of changes in the local dry-land vegetation. The reduction in lime and elm and the increase in herbs is strongly indicative of the first impact of human disturbance in the environment and of agriculture attributed to a local Neolithic economy.

Associated with this event are also changes in the status of the mire at the sample site. In sub- zone b, there is evidence of an ephemeral phase of increasing wetness as evidenced by increasing pollen frequencies of reed swamp taxa including Cyperaceae (sedges), Typha latifolia, and Typha angustifolia/ Sparganium type (bulrushes) and bur reed, and aquatic megaphytes such as Potamogeton (pond weed), Callitriche (water starwort), and Lemna (duckweed). Alnus (alder) declines sharply in sub-zone b in response to this increasing wetness. Local woodland clearance on the nearby interfluve may have caused a reduction in evapotranspiration, and increased surface run-off resulting in a higher ground water table. The fragile balance of alder in fen carr would clearly have been disturbed resulting in the demise of carr. Typically, alder will tolerate flooding of the peat surface around the root boles for some 3–4 months of the year during winter (Tansley 1949; McVean 1953; 1956). However, with the exception of a single peak of Alnus at 2.38 m OD, values of alder remain much reduced over its earlier dominance. Fluctuation of the reed swamp taxa (see Cyperaceae and Typha latifolia) may indicate a rather unstable/variable wetland habitat. Small increase in Salix (willow) pollen may be also significant since willows are largely under-represented in pollen spectra. Because of the very substantial numbers monolete Dryopteris spores (typical ferns) it is also likely that these are ferns growing within the wet, fen carr marsh habitat-although subsequently the diagnostic marsh fern (Thelypteris palustris is well represented.

By pollen assemblage sub-zone c, the numbers of grasses and weeds (see for example Plantago lanceolata) are much reduced. There is an accompanying increase in arboreal and shrub pollen. By the top of local pollen assemblage zone 2, there is some evidence of woodland regeneration with expansions of Tilia and Ulmus to c. 50% of previous values and some expansion of Fraxinus (ash), the latter being a typical secondary woodland element. These changes may suggest either local abandonment or change in agricultural or land use on the adjacent dry land. Although only a single grain/record, Fagus in this zone may be significant as this tree and also ash are very poorly represented in pollen spectra.

27 Pollen zone AWK:3 exhibits a very marked change in the character of local woodland. Betula (birch) becomes the dominant local vegetation. Alnus remains consistent at levels seen in subzone a and b of the preceding zone 2. Quercus, however, declines sharply in response to the increasing importance of birch. Corylus (hazel) remains broadly similar with some reduction. Clearly, there was a marked increase in growth of birch woodland. The decline in oak noted above may be both a real decrease in the area of growth and also a statistical function of the expansion of birch within the same pollen sum. The relatively small change of hazel implies that hazel (scrub?) also expanded significantly. Also associated with this change is the expansion of acidophilous elements including Ericales (ling and erica), a small peak of Osmunda regalis (royal fern), and Sphagnum (bog moss).

Birch has been recognised particularly as a component of fen carr woodland in the Somerset Levels (Beckett 1979) and these changes and high values present here may suggest a phase of drier fen wood or scrub development. Certainly, values of Cyperaceae are significantly reduced and the expansion of the acidophiles noted may represent the development of a different ground flora to that which existed in the alder carr with sedge and fern understorey. This does not, however, explain the continuance of alder at preceding levels unless differing areas of the mire/low lying ground were colonised rather than one community replacing another. A further alternative is the possibility that oak and lime woodland growing on the near terrestrial zone was cleared for its wood (for there is no real indication of agricultural expansion). The sandy character of the local geology would have been susceptible to leaching and creation of podzolic/acid soils suited to colonisation by ericaceous communities and in wetter areas on the fringes of the mire by Sphagnum. Plant macrofossil (seed and wood) data and Coleoptera indicate that the autochthonous vegetation communities of the fen were Sphagnum bog and densely vegetated marshes.

The Upper Clay-Silt: From 2.82 m OD there is a marked change in stratigraphy from the fen carr wood and detrital peat of pollen zone AWK:1, 2 and 3 into the more or less homogeneous blue- grey clay and silts of zone 4 upwards. Clearly, this change implies a major environmental change to alluvial and/or salt-marsh conditions. This changing environment and resultant lithostratigraphy importantly will have also changed the taphonomy of the pollen. Whilst the peat horizons discussed above (zones 1–3) will follow the typical pollen models for localised basins in a wooded environment (eg, Tauber 1965; 1967), interpretation of the minerogenic sequences must account for other taphonomic aspects such as fluvial transport and deposition. This includes the reworking of older soil/sediment reservoirs containing pollen and an extended pollen catchment through increased ‘openness’ as well as microfossil transport from wider parts of basin along with river-borne sediments. Furthermore, the sediments discussed here in pollen zones AWK:4 and AWK:5 are largely in part formed in marine/brackish water salt-marsh caused by regional positive RSL causing the widespread estuarine inundation of the fen peat/marsh zone.

In zone AWK:4, Quercus (oak) and Corylus avellana (hazel) are the dominant pollen taxa reflecting the broader importance of this woodland within the region. Tilia cf. cordata (not degraded) remains in small numbers throughout indicating continued growth within the catchment on drier better drained soils. Ash percentages decline although this does not necessarily imply that there was a real reduction in its growth/ importance since it is very likely that the pollen sources became farther away from the sample site. This frequently occurs with Tilia (Waller 1994; Scaife 2000). In zone 4, this is exemplified by the reduction in not only Fraxinus but also of the Ericales noted in zone AWK:3. In contrast to these declining responses to alluviation/flooding, the pollen spectra give a wider picture of the regional vegetation. In addition to the oak and hazel noted, which maintain importance due to their high pollen productivity and anemophily (wind pollination), evidence for regional agriculture is more pronounced. Opening of the environment allows increased airborne input of the airborne pollen 28 catchment as well as pollen fluvially derived from further afield. Here, there is an increase in cereal pollen especially in zone 5 (note Secale cereale – rye) along with typical weeds of cultivation such as Plantago lanceolata, Polygonaceae spp., and possibly other taxa which are not differentiable to species or even genus (eg, Fabaceae).

Marshland: The change to grey minerogenic sediments represents a shift from anaerobic peat formation to freshwater and/or more likely brackish estuarine water and salt-marsh conditions. From the base of pollen zone AWK:4, and the start of the grey sediments, there is an expansion of Chenopodiaceae which is a strong indication of salt-marsh development typified by Salicornia (glassworts) and Atriplex (oraches). Other halophytes are present and become increasingly important up the profile and especially in the uppermost zone (AWK:5). These taxa include sporadic but nevertheless significant presence of Armeria ‘A’ and ‘B’ line (thrift and sea lavender), Plantago maritima (sea plantain), Aster type (including sea aster), and large Poaceae, the latter being some halophytic grasses with large diameter pollen grains but not of cereal type. Potamogeton type may be Triglochin maritima but also includes Potamogeton (pond weed) which has similar pollen morphology preventing identification to a lower taxonomic level. These taxa are clear evidence for marine/brackish water incursion of the previously freshwater fen habitat. This is not unexpected since there is much evidence for this in other low lying areas along the Severn/Bristol Channel. In the Somerset levels this is evidenced clearly at Meare Heath (Hibbert 1980) where the raised Sphagnum ombrogenous mire was inundated at c. 910–760 cal BC (SRR-914, 2624±45 BP) lasting for c. 150 years.

Freshwater marsh/aquatic vegetation elements are present in the sediments since the out-flowing river catchment was transporting pollen from upstream sources. Cyperaceae are dominant with other fen marsh elements present including Caltha type (marsh marigold), Butomus umbellatus (flowering rush), Alisma plantago-aquatica (water plantain), Typha latifolia (bulrush), Typha/Sparganium (lesser bulrush and bur-reed), Myriophyllum spicatum (spiked water-milfoil), Callitriche (water starwort), Lemna (duckweed), and cysts of algal Pediastrum. Potamogeton type is also present but as noted above may also derive from Triglochin maritima a halophyte or from water pondweed (Potamogeton). Alder from fringing or valley fen carr is consistent throughout the two upper zones but values suggest that its importance in the region was diminished compared with the earlier and more extensive areas of fen peat carr in the Avon Levels.

Summary of environmental change i) Pollen zone AWK:1. Late Atlantic/middle Holocene (Flandrian Chronozone II; c. 4500 cal BC): Alder fen carr with a ground flora of sedges, grasses, and ferns occupied the sample site. Adjacent dry-land was dominated by lime woodland, with oak, elm, hazel, and ash. These latter may, however, have been constituents of drier parts of the fen carr and on heavier soils at the base of the valley sides. The pollen catchment is likely to have been of small extent being in the order of hundreds of metres surrounding the site. ii) Zone AWK:2a Lime, elm, and oak remain important but hazel expands sharply. In sub- zone b, lime and elm woodland is reduced and this phenomenon is thought to be the Neolithic ‘Primary Elm Decline’. Elm reduction was most possibly due to the spread of elm bark beetle and disease (Girling 1988) and the clearance of the lime woodland by the incoming of Neolithic/human activity. This is evidence by expansion of grasses, cereals and weeds of cultivation and human disturbance. Date c. 5000–4500 BP. Alder remains important on the site but there phases of wetter sedge fen and reduced alder. These phases may be due to the human activity increasing the ground water table locally (woodland clearance, increased run-off, decreased evapotranspiration). iii) Zone AWK:2c there was a reduction in human activity – at least cultivation, allowing secondary woodland regeneration of elm, lime, and possibly ash. Hazel remained 29 important and the carr community saw the stabilisation again of alder carr woodland with a ground flora dominated by marsh type ferns. iv) A substantial ecological change in zone AWK:3 (c. 4200 BP) may be due to clearance of oak, lime, and possibly elm woodland for timber on nearby drier areas, colonisation by pioneer birch scrub and possible extension or increase in flowering of local hazel. Sandy soils became degraded giving pedalogical conditions suited to the establishment of an acidophilous heathland community (Erica, Calluna, and Sphagnum moss) c. 2550 cal BC. An alternative interpretation is that drying out of the fen carr dominated by alder gave way to birch as the dominant carr woodland also with development of an acidophilous bog. The latter is in accord with the data from the Somerset Levels but is not at the same time. v) Erosion and disturbance of the upper peat occurred due to inundation by brackish water or freshwater ponded back in the river systems due to increasing RSL (relative sea-level) after c. 2550 cal BC. A date from the upper peat/sediment contact will produce an age for this event. This will however, be a terminus post-quem because of the erosion of the top of the peat and/or disturbance which has occurred. vi) Change to minerogenic sediment accumulation and presence of halophytic (salt tolerant) plants in pollen zone AWK:4 and 5 marks the inundation of the freshwater fen carr communities and replacement by progressively important salt-marsh habitats. The openness of the salt-marsh extended the pollen catchment which resulted in greater long distance and regional pollen input and also from freshwater fluvial discharges. The dating of the upper part of this profile is at present conjectural (Late Bronze Age–Iron Age– Romano-British?) but it seems that oak and hazel woodland remained the primary woodland elements throughout the late-prehistoric period. Secale cereale (rye) is present and is typically a Roman cereal crop but is now also occasionally found in Iron Age contexts (Chambers 1989) and even in Late Bronze Age ones (Grieg 1991). There are unfortunately no further pollen indicators. vii) Ephemeral phases of salt-marsh drying (for example AWK:5-a) possibly caused short lived phases of small negative sea-level tendency. This allowed partial pedogenesis and or organic formation in the grey salt-marsh sediments.

Diatom analysis by Nigel G. Cameron

The results of diatom analysis are presented in Figure 8. On the diagram the samples are identified by site name, AWK: Awkley Lane; HAL: Hallen Marsh; VIM: Vimpenny’s Lane. The subsequent numbers in the sample names refer to depths above Ordnance Datum. Diatom species salinity preferences are classified using the halobian groups of Hustedt (1957, 199):

1. Polyhalobous >30 g l-1 2. Mesohalobous 0.2-30g l-1 3. Halophilous – optimum in slightly brackish water 4. Oligohalobous indifferent – optimum in freshwater but tolerant of slightly brackish water 5. Halophobous – exclusively freshwater 6. Unknown – diatoms of unknown salinity preference

The principle source used for species salinity classification was Denys (1992).

30 Figure 8. Diatom analysis; all sites

31 +4.42m OD (context 107; within pollen zone AWK5): Of a series of samples assessed in 1994, and a further six examined in 2000, only one was countable. The diatom assemblage of this sample is composed of 55% marine species and almost 35% brackish water taxa. Freshwater diatoms comprise just over 5% of the assemblage and approximately 5% of the species were of unknown salinity preference. The marine planktonic diatom Paralia sulcata was dominant, comprising almost 40% of the assemblage and the mesohalobous diatom Nitzschia navicularis accounted for almost 30% of the assemblage. Both of these taxa are typical of estuarine sediment assemblages. The latter species is not common in the other samples analysed for diatoms, although it is common in estuarine sediments analysed in other geoarchaeological investigations in the Severn Estuary (eg. Cameron 1997). Vos & de Wolf (1993) consider Nitzschia navicularis to be part of the marine-brackish epipelic (motile mud-dwelling) diatom community and the species is an autochthonous component of the fossil assemblage of tidal mudflats. The context (107) description for this sample indicates a grey clay, with the comment that the sediment may be a buried topsoil. The diatom assemblage represents a highly saline, tidal, diatom habitat.

Molluscs by Michael J. Allen

Preservation of shells in these deposits is known to be poor and patchy (cf. Allen & Ritchie 2000; Bell & Johnson in Lawler et al. 1992). Nevertheless, the significance of determining brackish water, freshwater, or terrestrial facies from limited contexts was high. Contiguous samples were taken through the main sedimentary sequences (Awkley Lane, Vimpenny’s Lane, Awkley Interface), as well as from ditches on the occupation sites (Hallen and Northwick). In each case large samples (nominally 1750 g) were processed after the removal of a sub-sample suitable for the measurement of magnetic susceptibility. Samples were disaggregated in water and washed through a nest of sieves down to 0.5 mm (cf. Evans 1972). The results are presented in Table 2 and the nomenclature follows Kerney (1999).

Shells only survive in a limited portion of this 4 m section, essentially that coincidental with the two putative buried soils or temporary stabilisation horizons indicating and confirming temporary stasis. The assemblages are predominately dominated by fresh and brackish water species.

Insect remains by Mark Robinson

Insects were analysed from the Neolithic peats at the base of the alluvial sequence at Awkley Lane. They have been sampled in sequence, with three groups of three samples from within the Middle Wentlooge peat. Sub-samples from all of them were assessed for insect remains, which were found to be sparse and poorly preserved. It was therefore decided to limit analysis to 2 kg of the middle sample from each group. Each sample was subjected to paraffin flotation to recover insect remains. The flots were caught on a 0.2 mm sieve, washed in detergent and sorted under a binocular microscope. Identifiable insect remains were absent from context 114 (2.09 m OD). The results from the other two samples are given in Table 3. Nomenclature for Coleoptera follows Kloet and Hincks (1977).

The low concentration of insect remains makes detailed interpretation difficult, although the majority of the taxa are characteristic of wet habitats. Just under half the beetles in context 118 (1.49 m OD), the lowest peat layer, were aquatic, Ochthebius minimus being the most numerous. Aquatic larvae of Trichoptera (caddis fly) and Chironomidae (midges) were also present. There were fewer aquatic insects in context 121 (2.49 m OD), the top peat layer, but they included the water beetle Hydrobius fuscipes. The aquatic insects are more suggestive of the faunas of small pools and swamp areas on the peat rather than insects introduced in floodwater.

Table 2. Mollusc data from Awkley Lane (note: only samples with shells tabulated: totals exclude C. acicula)

Sample 2 10 3 4 5 6 7 9 11 12 14 15 17 18 20 21 23 25 8 1 16 22 13 26 19 24 Sample 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 Context 112 110 110 110 110 110 110 110 110 110 110 110 110 110 109 108 108 107 106 105 105 105 105 105 105 104 Depth (mOD) 2.78- 2.88- 2.98- 3.08- 3.18- 3.28- 3.38- 3.48- 3.58- 3.69- 3.79- 3.89- 3.99- 4.09- 4.19- 4.29- 4.39- 4.49- 4.59- 4.68- 4.78- 4.88- 4.98- 5.08- 5.18- 5.28- 2.68 2.78 2.88 2.98 3.08 2.18 3.28 3.38 3.48 3.59 3.69 3.79 3.89 3.99 4.09 4.19 4.29 4.39 4.49 4.58 4.68 4.78 4.88 4.90 5.08 5.18 Wt (g) 1750 1750 1750 1750 1750 1750 1750 1750 1750 1750 1750 1750 1750 1750 1750 1750 1750 1750 1750 1500 1400 1750 1750 1750 1750 1500 Succinea putris (Linnaeus) ------4 ------Succinea cf putris/Oxyloma - 1 ------5 - - 1 - - - Oxyloma pfeifferi (Rossmässler). ------1 - - - 5 ------cf Catinaella arenaria/S. oblonga ------1 ------Vallonia costata (Müller) ------1 (b) ------Cepaea/Arianta spp. ------1 ------Land mollusc Taxa 0 1 0 0 0 0 0 0 0 0 0 0 0 0 3 0 0 0 3 1 0 0 1 0 0 0 LAND MOLLUSC TOTAL 0 1 0 0 0 0 0 0 0 0 0 0 0 0 3 0 0 0 10 5 0 0 1 0 0 0 Bithynia tentaculata (Linnaeus) ------15 12 3 - - 2 ------Bithynia tentaculata operculum ------(16) (26) (7) - (6) (1) - - (2) - - - Aplexa hypnorum (Linnaeus) ------1 ------Lymnaea truncatula (Müller) ------9 13 - 17 5 ------Lymnaea peregra (Müller) ------11 6 - - 1 ------Lymnaea sp. ------4 ------Anisus leucostoma (Millet) ------10 5 - 1 2 1 ------Anisus vortex (Linnaeus) ------17 1 ------Gyraulus albus (Müller) ------5 1 - - - - - 1 - - - - Hippeutis complanatus (Linnaeus) ------7 1 ------Pisidium (valves) 2 Pisidium personatum Malm ------1 ------Pisidium obtusale (Lamarck) ------2 ------Pisidium pulchellum Jenyns ------3 1 ------Hydrobia ventrosa (Montagu) ------1 - - - 2 1 3 - - - - - Hydrobia ulvae (Pennant) ------1 1 1 - - - - - Aquatic Taxa 0 0 0 0 0 0 0 0 0 0 0 0 0 0 11 9 1 2 4 4 2 1 1 0 0 0 FRESHWATER TOTAL 0 0 0 0 0 0 0 0 0 0 0 0 0 0 79 58 7 18 14 3 0 1 2 0 0 0 MARINE TOTAL 0 0 0 0 0 0 0 0 0 0 0 0 0 0 4 1 0 0 3 2 4 0 0 0 0 0 Ostracods (valves) ------54 - - 19 17 ------Marine Gastropod ------1 ------TOTAL 0 1 0 0 0 0 0 0 0 0 0 0 0 0 86 59 7 18 27 10 0 1 3 0 0 0

Both samples contained ground beetles of marsh or peaty habitats, for example Elephrus cf. cupreus, which occurs at the edge of standing water where some vegetation is present and in forest swamps, and Agonum obscurum, which occurs in damp deciduous forests and densely vegetated marshes (Lindroth 1974, 33, 82). Evidence for carr (woodland growing on peat) was given by a scolytid beetle, possibly Lymantor coryli which lives under the bark of Alnus glutinosa (alder) and Corylus avellana (hazel), in context 118 and Rhynchaenus sp. (not pratensis), a weevil which feeds on tree leaves, in context 121. However, some open meadow-like areas were suggested by a couple of examples of the weevil Apion sp., which mostly feed in clovers and vetches, context 118.

Table 3. Coleoptera and other insects from Awkley Lane

Min. no. indiv Context 118 114 121 Sample 1036 1032 1030 OD 1.49 2.09 2.49 Coleoptra Elephrus cf. cupreus 1 - - Pterostichus diligens - - 1 Agonum obscurum - - 1 Hydrobius fuscipes - - 1 Helochares or Enochrus sp. - - 1 Ochthebius minimus 1 - 1 O. cf. minimus 5 - - Hydraena sp. (not testacea) 1 - - Ptenidium sp. 1 - 2 Olophrum cf. fuscum - - 1 Stenus sp. 1 - - Lathrobium sp. (not rufipenne) 1 - 2 Medon sp. - - 1 Philonthus sp. - - 1 Aleocharinac indet 1 - - Pselaphidae indet. 1 - - Corylophus cassidoides 1 - - Apion sp. (not Malvaceae, Urtica or Salix-dependent) - - 2 Rhynchaenus sp.(not pratensis) - - 1 cf. Lymantor coryli 1 - - Other insects Trichoptera indet. – larva 1 - - Chironomidae indet – larva 1 - - Diptera indet – puparium - - 1

Waterlogged Plant Macrofossils by Alan J. Clapham

Seven samples were examined from the Middle Wentlooge peats; three contiguous samples from the dark reddish brown basal wood peat (context 118), three contiguous samples from peat contexts 115 (1.99 m OD), 120 (2.04 m OD), 114 (2.09 m OD), and a final sample from wood peat context 121 (2.49 m OD). Waterlogged plant macrofossils were, in most cases, well preserved (Table 4)

Samples from the three zones of peat were analysed. As the volume of each sample was the same it is possible to produce an equivalent diagram for seeds as for pollen, this is presented in Figure 7. In order to aid the discussion of the waterlogged plant remains the results are presented as groups corresponding to the peat horizons which they represent.

Context 118 (1.44m OD, 1.48m OD and 1.54m OD; polen zone AWK:1) The dominant tree remains within the samples were of alder (Alnus glutinosa), and included fruits, female cones, and male catkin fragments and scales. In some cases it was possible to detect the presence of anthers within the latter. Alder buds and budscales were also identified from

34 1.44 m OD. Hazel (Corylus avellana) nutshell fragments were identified from the lower two samples, including one charred. Two immature small-leaved lime (Tilia cordata) fruits were identified.

Other species identified included those which indicate wet/marshy conditions and prefer to have their feet in water. These include celery-leaved buttercup (Ranunculus sceleratus), creeping yellow-cress (Rorippa sylvestris), alternate- leaved golden saxifrage (Chrysosplenium alternifolium), gipsywort (Lycopus europaeus), marestail (Hippuris vulgaris), hemp agrimony (Eupatorium cannabinum), water-plantain, rush (Juncus sp.), great fen-sedge (Cladium mariscus), sedges (Carex spp.), and reedmace (Typha sp.).

A drier habitat is indicated by the presence of buttercups (Ranunculus subgenus Ranunculus), whilst a scrubbier element is represented by the presence of stinging nettle (Urtica dioica), bramble (Rubus Section 2 Glandulosus) and woody nightshade (Solanum dulcamara). Charcoal fragments were found in each of the samples.

As the samples contain the same plant taxa representing similar habitats it is possible to treat the three samples as one with regards to environmental reconstruction. The dominance of alder confirms the pollen evidence for the dominance on-site of alder carr within which, on the drier areas, hazel was present. The undergrowth consists of a wetland/marshy flora with tall herbs such as reedmace, hemp agrimony, and great fen sedge associated with stinging nettle. Bramble and woody nightshade can be found scrambling over this vegetation. Wet/marshy species grew beneath while, in the open water, which may be present as pools within the alder carr, marestail can be found growing along with waterfleas, the eggs of which were common in sample 1.54 m OD. The occurrence of fruits of small-leaved lime suggests that drier land is not too far away. The presence of the charcoal is difficult to interpret, it may represent in situ burning, whether natural or anthropogenic, or may represent an allocthanous element. Caddis fly larvae were also recorded from sample at 1.44 m OD indicating an aquatic environment.

Context 115 (2.06 m OD; pollen zone AWK:2) Context 115 was a woody peat with common monocotyledonous roots. There were no tree remains identified from this sample, most of the species recovered are representative of wetland habitats and in this case it most likely indicates the presence of reedswamp/fen. The sample was dominated by seeds of the square-stalked St John’s-wort (Hypericum tetrapterum), watermint (Mentha aquatica), and hemp agrimony cypsela fragments. Rhizome fragments of reed were also very common as were small grass caryopses. Other species include buttercup, stinging nettle, orache (Atriplex sp.), bramble, woody nightshade, rush (Juncus sp.), and sedge (Carex sp.) along with reedmace (Typha sp.). Charcoal fragments were also recorded, the origin of which is difficult to determine. The presence of one seed of elder (Sambucus nigra) was also identified. Water flea eggs were present.

Context 120 (2.14 m OD; pollen zone AWK:2) Context 120 was described in the field as a black peat (most likely detrital in origin). This sample produced a radiocarbon date of 3640–3380 cal BC. Again there were no tree species present. The dominant species in this sample were, stinging nettle, fragments of hemp agrimony fruits, and fragments of Phragmites rhizomes. Other species included celery-leaved buttercup, square-stalked St John’s-wort (although not as common as in context 115), bramble, woody nightshade, gipsywort, watermint, sedges, small grasses, and reedmace. waterflea eggs were also present which suggests some pools of open water. This represents a tall herb fen/reedswamp.

Context 114 (2.09m OD, pollen zone AWK:2) Of the three samples from this peaty horizon, this was the richest sample in terms of numbers and taxa of waterlogged plant remains and again was dominated by taxa which are found growing in reedswamps. The sample was dominated by the presence of fragments of hemp agrimony fruits and rhizome fragments of Phragmites. Other common species included celery-leaved buttercup and watermint. A species which is often found in fens, common meadow-rue (Thalictrum flavum) was also recorded. Other species included, stinging nettle, red goosefoot (Chenopodium rubrum), orache (Atriplex sp.), golden dock (Rumex maritimus), square-stalked St John’s-wort, bramble, hairy willowherb (Epilobium hirsutum), woody nightshade, gipsywort, sedges, small grasses, and reedmace.

In this sample the aquatic habitat was represented by the presence of rigid hornwort (Ceratophyllum demersum var apiculatum), fool’s watercress (Apium nodiflorum), and narrow-leaved water-plantain (Alisma lanceolatum). Again this sample can be said to indicate the presence of reedswamp/tall herb fen. The presence of a more aquatic environment is more prevalent in this sample than in those from contexts 1145 and 115.

These three samples discussed indicate the presence of a tall herb fen/marginal aquatics/reedswamp, with no arboreal elements apart from the find of one elder seed. The habitat

35 Table 4. The waterlogged plant remains from the wood and fen peat of the Middle Wentlooge Formation at Awkley Lane

Context 118 118 118 115 120 114 121 Sample 1037 1036 1035 1034 1033 1032 1030 O.D. 1.46-1.41 1.51-1.46 1.56-1.51 2.01-1.96 2.06-2.01 2.11-2.06 2.51-2.46 (1.44) (1.49) (1.54) (1.99) (2.04) (2.09) (2.49) 14C yrs BP(uncalibrated) 5603±50 4745±50 4286±55 3 Volume processed cm 300 300 300 300 300 300 300 Chara oogonia - - - 2 - - - Cenococcum geophilum 41 15 33 - - - 133 Musci common common occasional - - occasional - Ceratophyllum demersum var apiculatum - - - - - 2+14f - Ranunculus sceleratus - 1 - - 7 63+60(½s)+1 - 8f Ranunculus subgenus Ranunculus 2+7f 21+35f 4+8f 2f - - - Thalictrum flavum - - - - - 3f - Urtica dioica 2 - 1+2f 18 301 18 - Alnus glutinosa fruits 33+68f 35+8f 63+42f - - - - Alnus glutinosa female cones 14+9f 18+19f 15 - - - Alnus glutinosa male catkin frags (stem) 28 100+ 26 - - - - Alnus glutinosa male catkin scales 285 ------Alnus glutinosa budscales +buds 5+2 ------Corylus avellana nutshell fragments 19+1char 3 - - - - - Tilia cordata immature fruits - 2 - - - - - Rorippa sylvestris 1 ------Chenopodium rubrum ------1 - Atriplex sp. - - - 1 - - - Rumex maritimus valves - - - - - 7 - Rumex maritimus nutlets - - - - - 1+1f - Hypericum tetrapterum - - - 335 12 2 - Chyrsosplenium alternifolium 144+20f 54 - - - - - Rubus Section 2 Glandulosus - 2f 3+1f 5f 2+5f 5f - Epilobium hirsutum - - - - - 2f - Apium nodiflorum - - - - - 1 - Solanum dulcamara - 2f - 1+8f 4f 3+11f - Lycopus europaeus 9 - - - 1 14 - Hippuris vulgaris 1 - -- 149+6f 4 145+46f - Mentha aquatica/arvensis 1 - - 1 - - - Sambucus nigra 1 ------Eupatorium cannabinum 8f 3f - 257f 1000f+ 1000f+ - Alisma lanceolatum seedcase - - - - - 1+1f - Alisma sp. embryo - - - - - 1 - Alisma sp. seedcase 1 ------Juncus sp. 2 - - 1 - - - Cladium mariscus 1 ------Carex sp. (lenticular) 105+127f 75+21f 1+6f 1 - - - Carex sp. (trigonous) 16+17f 7+10f 2f - 13f 36f - Carex sp. utricule fragments - - - - - 16 - Typha latifolia/angustifolia 6 1 - - - - - Sphagnum spp. leaves ------1000+ Betula sp. seed ------5+2f Phragmites australis rhizome fragments - - - 100+ 100+ 100+ - Small Poaceae - - - 170 13 12 - Typha latifolia/angustifolia - - - 22 7 55 - Leaf fragments 2 ------Charcoal 83f 14f 2f 36f - - - Culm node - 1 - - - - - Budscales 1 ------Bud bases - 1 8 - - - - Leaf abscission pads 39 - - - - - 2 Miscellaneous anther 1 1 - - - - - Worm cocoons ------16 was dominated by the tall herbs such as reeds, reedmace, stinging nettle and hemp agrimony, which supported the scrambling woody nightshade and bramble. The undergrowth supported plants such as gipsywort, celery-leaved buttercup, square-stalked St John’s-wort, watermint, and golden dock. It can perhaps be said that the samples show a steady progression from carr to fen/reedswamp. In context 114, the presence of aquatic species such as rigid hornwort, narrow-leaved water-plantain and fools watercress may indicate a rising in the watertable creating more pools of open water.

36 Context 121 (2.41 m OD, pollen zone AWK2-c) This sample contained large chunks of wood and was very humified, in the field it was described as a black peat with up to 20% wood inclusions. This context produced a radiocarbon date of 3090–2700 cal BC (NZA-1258, 4286±55 BP). This sample was the middle sample from the uppermost peat horizon.

There were very few plant remains, the sample being dominated by large quantities of Sphagnum moss leaves (Sphagnum spp.) The only other remains were birch (Betula sp.) fruits. Birch bark and moss stems, presumably of Sphagnum were also present in large quantities. One noticeable feature of this sample was the sclerotia of Cenococcum geophilum, although numerous the striking feature about them was that some were up to 3 mm in diameter which is large for fungal sclerotia.

This sample represents a Sphagnum bog, on which birch trees were growing. The small numbers of birch fruits recovered suggest that the trees were growing some distance from the sample, but, conversely, the presence of the birch bark suggests that they may be more local.

Overall, there seems to be an agreement between the different sources of environmental evidence. The lack of beetle and other insect remains with the samples makes it difficult to carry out an accurate comparison, which is not so with the pollen record. Altogether the different environmental evidence suggests that there is a change in habitat through time, with alder carr giving way to tall- herb fen/reedswamp, which finally gives way to a birch, Sphagnum dominated woodland.The lack of archaeological evidence from the area means that it is difficult to determine if these changes were entirely natural, anthropogenic, or a combination of both.

2. Vimpennys Lane

The sequence

Two trenches were excavated off Vimpenny’s Lane (ST 5561 8211) approximately 1.5 km to the west of Easter Compton (Fig. 9). Apart from providing a sedimentary sequence these trenches were excavated specifically to investigate a lens of freshwater peat at 3.5 m OD (2.7 m below ground level) previously identified by Lawler et al. (1992, 72). The objective was to provide a full analysis of the sedimentary sequence including palaeo-magnetic dating. A full range of samples including mollusc columns, pollen monoliths, samples for particle size analysis, was taken from trench 2.

Both trenches were c.10 10 m, and were excavated to a maximum depth of 3 m. The north-east side of each trench was stepped to give a running section. The profile in trench 2 was recorded and sampled (Fig. 9) and an attempt made to obtain palaeo-magnetic dates from the fine-grained sequence. An auger hole in the base of trench 1, using a dutch auger, extended the profile to 5.8 m depth (-0.15 m OD), but did not enable sample retrieval. The sediments were described by the archaeologists in the field and these subsequent descriptions augmented by examination of monoliths in the laboratory (see descriptions).

The sequences in trenches 1 and 2 were virtually identical. A series of low-energy greyish brown silty clay alluvial deposits 1.25 m thick (to 4.2 m OD) with low or suppressed magnetic susceptibility (11 to14 SI10-8Kg-1) rising in the lower 0.5 m (to 16 SI10-8Kg-1), were present below the present day soil. These overlay a 0.2 m thick, very dark grey silty clay putative stabilisation horizon (layer 207, 4.19 m OD), with slightly suppressed rather than enhanced magnetic susceptibility, developed over a greyish brown silty clay similar to that above. The undulating disturbance patterns seen in section and irregular nature of this layer as seen elsewhere (eg, GGAT layer 223, see Fig. 9), may suggest turbulent water erosion, flooding, and mixing. Below c. 3.90m are a further 0.95m of grey and dark grey silty clay alluvial deposits which are distinctly laminated towards the base. This zone of very fine slightly siltier laminations appear to be remnants of depositional strata, probably flood couplets (very fine layers of alternating silt and silty 37 clay, each pair relating to a single flood event), resulting from over-bank deposition during periods of flood of MHW, or possibly tidal movement. This unit overlies a narrow band of dark grey/black, highly humified, silty peat (layer 211 at 3.04 m O.D.), 0.05m thick which overlies a heavily rooted silty sand/silt loam grading into 1.15m of light grey silty clay alluvium. This in turn overlies a second band of peat (1.74 m O.D.), 0.4 m, thick and only recorded in a auger hole through the base of Trench 1. The final 1.35 m at the base of the auger hole comprise a series of grey silty clay and grey sands.

Depth (m) OD Layer Description Comment 0-0.15 5.80-5.65 200 Dark greyish brown (10YR 4/2) silty loam, occasional pebbles, rare small pebbles. 0.15-0.25 5.65-5.55 201 Brown (7.5YR 5/6) stonefree silty clay, with thin lenses of lighter grey matrix. 0.25-0.55 5.55-5.25 202 Light brown (7.5YR 6/4) stonefree masive (unstructured) Upper Alluvium silty clay, common mineral (iron and mangenese) staining, occasional gritiness caused by concretions of iron and oxyhydroxides, substantial rootlets (Phragmites), and stems marked with orange traces, diffuse boundary. 0.55-0.80 5.25-5.00 203 Greyish brown (10YR 5/2) stonefree silty clay loam, Upper Alluvium occasional mineral staining. 0.80-0.90 5.00-4.90 204 Grey (5Y 6/1) stonefree silty clay loam; 30mm zone of grey Upper Alluvium silty clay at top of band of fine gleyed material. 0.90-1.30 4.90-4.50 205 Brown (10YR 5/3) stonefree silty clay, occasional mineral Upper Alluvium (iron) staining. 1.30-1.60 4.50-4.20 206 Dark greyish brown (2.5Y 4/2) stonefree silty clay, common Upper Alluvium mineral (inc manganese) staining, very rare charcoal, some relict fine (1mm) laminations towards the base of this unit. 1.60-1.75 4.20-4.05 207 Very dark grey (10YR 3/1) stonefree slightly silty clay, humic/ ?Buried soil/ organic layer [but not so in sampled core] frequent mineral stabilisation horizon (iron) staining, coarse matrix, slightly gritty due to the presence of sand element and a number of shells (Hydrobia). Some ‘swirling’ or undulating disturbance pattern seen in section. Distinctly less organic than the same horizon recorded in evaluation trench, described as with rare snail shell fragments and partly decayed reed/grass stems and rootlets. 1.75-1.90 4.05-3.90 208 Grey (2.5Y N/5) stonefree silty clay, mottled, common Grey Alluvium mineral staining and occasional vegetable matter and r rootleting indicated by orange-brown staining, some ubvertical zones and localised patches. 1.90-2.05 3.90-3.75 209 Grey (10YR 5/1) stonefree silty clay, occasional mineral Grey Alluvium staining and very occasional vegetable matter marked by orange brown staining. 2.05-2.65 3.75-3.15 210 Grey (2.5Y 5/0) silty clay, occasional vegetable matter, Grey Alluvium eroded peat and some intact root stems pass through this unit, diffuse silty rhythmical laminations towards base with alternating very fine laminae of silt and silty clay – distinct flood couplets in lower 0.2m (2.45-2.65m OD), clear to gradual wavy boundary 2.65-2.75 3.15-3.05 211 Dark grey/black (10YR 3/1–2/1) highly humified silty detrital Phragmites peat with Phragmites, mottled with light grey silty clay ?backfen peat patches, gradual smooth boundary and peat grades downwards into heavily rootleted, light grey silty sand 2.75+ 3.05+ 212 Grey (10YR 5/1) stonefree silty sand/silt loam, heavily Grey Alluvium rootleted, common vegetable matter. Auger hole through trench 1 3.10-4.10 2.60-1.50 Light grey (7.5YR N7/0) silty clay, common plant matter. Probably the same layer as 210. 4.10-4.25 1.50-1.45 Very dark brown (10YR 2/2) peat. Peat 4.25-4.50 1.45-1.20 Black (10YR 2/1) peat. Peat 4.50-4.55 1.20-1.15 Dark grey ((10YR 4/1) silty clay, common plant matter. Alluvium 4.55-4.90 1.15-0.80 Grey (2.5Y N5/0) sand, occasional vegetable matter. Alluvium 4.90-5.80 0.80- -0.10 Grey (10YR 5/1) silty sand. Alluvium 5.80-5.85 -0.10- -0.15 Grey (7.5YR N5/0) silty sand clay. Alluvium 5.85+ -0.15+ Grey (2.5Y N5/0) sand. Alluvium

In general, the sequence was the same as that identified by Lawler et al. (1992, 72–3). The upper band of silty peat at 3.14 m OD was dated to 2900-2620 cal BC (4182±55 BP, NZA 12527) and appears to correspond with that found in evaluation (trial pit G009, context 227 at 2.8 m OD), dated

38 WA Vimpenny Tr.1 WA Vimpenny Tr.2 GGAT G009

Mollusc & soil column

201

202 Pollen Zones

203 5 m O.D. 2b 204

205 2

206

2a 207 4 m O.D. 208 209 1b 0 10 20 30 x10-8 SI/kg 210 1 4182+_ 55BP 211 3 m O.D. 1a 212 4420+_ 90BP

2 m O.D.

Insect sample 1 m O.D. Plant macro fossils sample C14 sample Alluvial silty clays (brownish red hue) Alluvial silty clays (bluish grey hue) ?Stabilisation horizon Peat (wood peat vs humified peat) Sandy alluvium 0 m O.D.

01 m

Figure 9. Vimpenny's Lane Wentlooge Formation profiles

39 to 3350–2900 (4420±90 BP, GU3121) (Lawler et al. 1992, 115). This would categorise it as a fourth peat as defined by Lawler et al. (1992, 8). The lower band of peat would also, by virtue of its relative height OD, fall into this fourth peat category and may represent an area where the fourth peat development was interrupted by inundation, with further development resuming at a later date.

The sequence represents low-energy estuarine deposition with two obvious episodes of stasis and peat formation and one possible non-organic stasis or stabilisation horizon. The base of the sampled sequence is dated to about 3000 cal BC, and no other chronological references can be provided for the overlying 2.7 m of minerogenic sediments.

Chronology

As most of the exposed sequence was minerogenic there was little opportunity for absolute dating. Palaeo-magnetic intensities and directions were very scattered within the minerogenic sequence. Demagnetistaion revealed stable remnance, suggesting that the scatter in vectors was due to inhomogeneous magenetisation composed of two stable components of differing age: the primary depositional vector overprinted by secondary chemical magnetisation arising during diagenesis. The palaeo-magnetic record was, therefore, unsuitable for absolute dating.

Only one horizon, a well humified Phragmites peat band 0.1 m thick, was suitable for radiocarbon dating. A result of 4420±90 BP was obtained from the evaluation sequence, and a slightly latter determination of 4182±55 BP from the analysed sequence (Table 5). These give calibrated results at the end of the 4th–beginning of the 3rd millennium BC.

Table 5. Radiocarbon determinations and calibrated results from the detrital peat at Vimpenny’s Lane context context type depth OD material lab no C13‰ result BP calib Vimpenny’s Lane 211 ?OLS 3.14m plant matter NZA 12527 -25.98 4182± 55 2900-2620 GGAT 2.80m GU-3121 -27.2 4420± 90 3350-2900*

* GGAT 1992

Sampling the sequence

As with the Awkley Lane sequence the full minerogenic sequence, excepting the upper 0.5 m, was sampled in monolith tins to facilitate subsampling for pollen and diatoms. Again gaps in this sequence were filled by sub-sampling disturbed samples taken for land snails (Fig. 9). A series of 19 contiguous samples was taken for snails from which sub-samples were removed for grain size analysis. Bulk samples were was taken for archaeo-magnetic dating through the minerogenic deposits, but these failed to produce consistent results.

Textural features of the upper sedimentary sequence by J.R.L. Allen

The 19 samples from the upper minerogenic sequence were analysed and described. The triangular graph (Fig. 10) shows the scatter of values of the clay-silt-sand ratio. Mean values for some of the finer-grained samples should be treated with caution. Although very small in quantity,

40 Plate 3. Vimpenny's Lane excavation showing the fully excavated and exposed upper and middle Wentlooge sequence

Vimpenny's Lane Clay (100%)

1108 1117 1100 1107

50% 50%

Clay

Silt (100%) Sand 50% Sand (100%)

Figure 10. Vimpenny's Lane; particle size triangle

41 concretionary material was present which could not be dispersed. This has exaggerated the coarse tails of the distributions, inflating values for the mean.

The triangular graph, in addition to the partical size proportions, shows that the samples divide between a lower, coarser (layers 209–206 at 3.81–4.61m OD) and an upper, finer (layers 205–200, 4.61–5.75m OD) sub-set. Taken together, however, the two subsets define a single trend similar to that recognised at Hallen Marsh (below).

The sediments of the coarser sub-set are texturally similar to all but the finer deposits at Hallen Marsh and similar to deposits typical of salt-marshes in the Severn Estuary Levels. Given the resolution permitted by the chosen sampling interval, the particle size proportions reveals no clear stratigraphic trends in this sub-set. One sample from the middle of layer 206 at 4.45 m OD is, however, noteworthy for its unusually coarse mode.

There is a major textural change from the coarser to the finer sub-set between samples within layer 206 at 4.51 m OD. The latter sub-set is exceptionally fine-grained for Severn Estuary Levels salt- marshes, although not inconsistent with the interior position of the site. No stratigraphic trends within it can be recognised with any great confidence, but it may be noted that samples from the top of 203–201 at the top of the sequence (5.25–5.55 m OD) yield modes and curve shapes suggesting that these sediments are coarser than those below and above. It is possible that a weak coarsening- up trend is followed by a weak fining-up sequence.

The stratigraphically lower, coarser sub-set suggests a stable marsh that is comparatively near the marsh edge and/or major creeks. A sudden, major change in conditions is suggested by the sharp textural break within layer 206 above sample at 4.41–4.51 m OD and the appearance of the finer- grained subset. The new environment appears to involve a distant marsh edge and/or remote major creeks (?sudden diminution in creek density). These conditions were either stable or, possibly, included an earlier episode of slight marsh-edge retreat and/or positive sea-level tendency, followed by a later episode of slight marsh-edge advance and/or negative sea-level tendency.

Pollen analysis by Robert G. Scaife

Pollen extraction procedures used at Vimpenny’s Lane followed those detailed for the analysis of Awkley Lane. Samples of 5 ml were used and absolute pollen frequencies were calculated.

The Vimpenny’s profile can be divided into two principal pollen assemblage zones (Fig. 11) which relate to regional vegetation. A number of local pollen assemblage sub-zones are also defined which relate to changes in the on-site vegetation communities. These are also manifested in the changing stratigraphy of the site. These pollen zones are defined and characterised as follows.

VIMP:1 (212, 211, 210, 209) 2.90–3.98 m OD, Pinus-Quercus-Tilia-Corylus avellana type This zone is characterised by highest values of Quercus (to 60%), Tilia (to 6%), and Corylus avellana type (to 28%). Also present are Pinus (to 10%) and sporadic/non-continuous occurrences of Betula, Picea, Ulmus, and Fraxinus. Herbs (30–40%) are dominated by Chenopodiaceae (to 24% at the top of VIMP 1.a.) and Poaceae (peak to 30%) with sporadic occurrences of other taxa including halophytes – Armeria ‘A’ and ‘B’ line. Marsh taxa include Alnus (10%), Cyperaceae (to 18%), and some aquatic types (?Potamogeton) and reed swamp taxa (Typha and Sparganium). There are substantial numbers of spores comprising monolete Dryopteris type (20%), Pteridium aquilinum (peaking to 40%), and Polypodium vulgare (basal level = 22%). Of note are higher values/numbers of pre-Quaternary palynomorphs in VIMP 1.a and Sphagnum incoming at the top of the zone from 3.58 m OD.

42 Two local pollen assemblage sub-zones have been recognised. These are characterised as follows.

VIM:1a, (212, 211) 2.90–3.16 m OD: A radiocarbon date of 2900–2620 cal BC (NZA-12527, 4182± 55BP) from a detrital organic horizon (211) delimits the top of this sub-zone. Arboreal pollen is defined as above but with some reductions in Quercus and Pinus relating to the thin peat horizon, that is, reflecting increased percentages of Chenopodiaceae (expanding to 26%). Derived pre-Quaternary palynomorphs are more important. Absolute pollen frequencies increase from typically low values in these sediments to >100,000 grains/ml.

VIMP:1b, (210, 210) 3.62–3.98 m OD: In this sub-zone, there is an expansion of acidophilous taxa including Ericaceae (Erica and Calluna) and Sphagnum (to 15%). Some herbs are also incoming in this sub-zone-Plantago lanceolata, Lactucoideae, and Plantago coronopus type.

VIMP:2: (108, 207, 205, 204, 203, 202) 3.98-5.42 m OD, Chenopodiaceae-Herbs-Pteridium aquilinum. This zone is delimited by a sharp reduction in tree pollen at 184cm with marked declines in Quercus (down to 20– 25%), Pinus (2–3%) and Tilia (to occasional occurrences). Corylus avellana type remains consistent from the middle of zone 1 (zone 1b) at c. 20%. Herbs become dominant (to 75–80%) with Poaceae (to 50%), Chenopodiaceae (25%) and increased herb diversity, including Plantago lanceolata, Cereal type and Asteraceae types. Incoming of Ericaceae (Erica and Calluna) in zone 1b continues throughout zone 2. Within the marsh/aquatic category, Alnus remains but with decrease in diversity. Ruppia maritima is present. Of note is a peak of Potamogeton type (cf. Triglochin) along with the halophytes Plantago maritima, Aster type and Chenopodiaceae (zone 2b). Spores become more important with increasing percentages of Pteridium aquilinum (to 55%), Polypodium vulgare declining from values of zone along with other spores showing some further expansion after zone 2a.

Two local pollen assemblage subzones have been recognised.

VIMP:2a, (208, 207 lower 206) 3.98–4.26 m OD: Immediately above the lower zone boundary (VIMP:1/2) at 3.94 m OD absolute pollen frequencies attain highest values (to c. 200,000 grains/ml) after which values decrease sharply to typical site values of 5–10,000 grains/ml. In this sub-zone there is a peak of Poaceae (highest values of 50%) and Chenopodiaceae (27%). There are, however, reductions in Pinus (to absence), Quercus (to 15%) and especially spores – Pteridium aquilinum, Dryopteris type, Polypodium vulgare, and miscellaneous pre-Quaternary palynomorphs (to absence) and Dinoflagellates.

VIMP:2b, (205, 204, 203) c. 4.70–5.08 m OD: Defined by a marked peak of Potamogeton type (38%) (includes Potamogeton and Triglochin maritima). This is associated with expansions of Plantago maritima (17%), Plantago media/major type, Aster type ( 5%), and Chenopodiaceae (to 28%). There are corresponding reductions in dinoflagellates and pre-Quaternary palynomorphs.

As discussed above the lithostratigraphical sequence at Vimpenny’s Lane comprises largely minerogenic, grey silt, and silty clay. This has a texture and appearance typical of salt-marsh sediments, a fact largely confirmed by the pollen data obtained for this site. Although salt-marsh was important throughout the time-span of this sequence, changes in relative sea-level have given subtle changes in the on-site vegetation and consequent stratigraphy. Of particular note is a thin but highly organic detrital peat at 3.18 m OD which has furbished a radiocarbon date demonstrating that the base of the pollen profile is of Middle–Late Neolithic age 2900–2620 cal BC (NZA- 12527,4182±55 BP). The regional vegetation of the late prehistoric period was dominated by oak, lime, and hazel (seen in the lower pollen zone). A cessation in sedimentation at a depth of 3.98 m OD (184 cm in the pollen profile) possibly delimits the later prehistoric sediments from sediments of later historic (?post-Roman–medieval) age. During the interval of this hiatus, woodland clearance had taken place. The pollen data overall can be correlated with the analysis of Awkley Lane and Hallen and from further afield from the Somerset Levels which provide a detailed and established framework for this geographical region.

43 A: Vimpenny's Lane

TREES & SHRUBS HERBS

F. DIF

TYPE

YPE YPE E M L IACE E UN P S P NDIFF.YPEYE VI L RET Y R YE T T LIX PE IA TYPE U AYU E IA M O US CO MUNISSV TY LARIA T ILL T PE H egrdd) RUM NG U IU G BAA I LA L NDULA X T ONET RICUMACEAE S IUM TYPE E S PHUL RCU L N THAT NODIYA IA PENT YONEAE EALINDIF. U MUS ES RYLU AC E LANAPE NU V NUSTYER SACEAE UNDIFF.IA E L ETULA I U RIUOLIIAL IIAAXUS ae ACC TRA M N NA ALV A TLPRH E R EDC ILIP O OG UACUAR RECAA LPE Strat. B PINPC S EAUL Q F AXNPPUUST S TLJUNIPER (d T CO EICALE UNAAEMRASI NCULUHOR PS cf. UT IOHYPEM DI CE S ESEG LARIAC TYP E OPFAB UMT CEAE T IFOM PE RO F PTAPHEDERA C AEPO NDIFF.PLYONM GRUM ACAMRIA EUA N MCALY '' RIAF SRO 'B' TEG INEZone 32

96 VIMP 2.b.

112

128 VIMP 2 144 Depth cm 160 VIMP 2.a 4182+55 BP 176

192 VIMP 1.b 208

224

240 VIMP 1 256

272 VIMP 1.a 288

20 40 60 20 20

For Key See Figure 7a % = Sum Total Dry Land Pollen

Figure 11a. Vimpenny's Lane pollen diagram.

44 B: Vimpenny's Lane cont.

HERBS Cont. MARSH SPORES

UM OR L E TYPE F P S U R A TY A TYPE AED E TE P UM O IM ELATA NOP GD A OIA LG R OR YPE O NIFLO A ER LG R UAE UNCTATUM R IGRA T P IMA ALI UILIN YE ARI FICINALS AE UTUM V A U L V A IS EG S LANC M OFTY E I TYPE R ST P A YPE D TOIA E S M RA P O OCO A TYPE YEIS T T YL UM L RNIGETON TY A M Q RTSE S U MARIT D D O ATAG CIE A ANGUSTI OIUM C T T R TE P AURE NE Y E E POS E MO IDIU O NT GO IUMTA CAE GACEAU NA UN OTERI ETTM CT O REALR YPEDE T L RI PLPFLM L L LA AL LIT TAIU ORELALE CU ID P NSS YPENHR O E A AIXALTHA TYPEYPY. H O YPERAC AETE A RY P OLYP VRN P ANTPANTAG GOP MAJPANcfLGLVRIAN MSABIOSCCISAB ATEA ATEMISIACIRSCEN LAPA U C L UNIALN N IFIED/DS C MYM OPH RIOcf H IPPPTA LL RIS M SPIC ULEMRUPPITY HATY HAAC IF LOS TYPEPR D P SPHAGNULI EATH W Zone 32

96 VIMP 2.b.

112

128 VIMP 2 144

160 VIMP 2.a Depth cm 176

192 VIMP 1.b 208

224

240 VIMP 1

256

272 VIMP 1.a 288

20 20 40 20 20 40 20 20 40 60 20 20 20 % Sum t.d.l.p. % sum tdlp + Marsh % sum tdlp + Spores

Figure 11b. Vimpenny's Lane pollen diagram. Note that alder (Alnus sp) is included with fen species in Fig 11b.

45 C: Vimpenny's Lane cont.

MISC. TOTALS

S E R AE ml E NARY / C R ins UM E a ATA S SUM CHOSPHE S gr UAT B 0 IASTRNEM Q U ALEBS SH RES EN x1 D E- EES R IC R O C LL F R ER A IS O PE ZYG HYSTRIP TR SH ER H M SP M P AP Zone 32

96 VIMP 2.b. 112

128 VIMP 2 144

160 VIMP 2.a

Depth cm 176

192 VIMP 1.b 208

224

240 VIMP 1 256

272 VIMP 1.a 288

20 40 20 40 60 20 20 40 60 80 20 40 20 40 60 80 20 40 200400600800 500 100015002000 % = Sum + Misc. Rob Scaife

Figure 11c. Vimpenny's Lane pollen diagram.

46 Vimpenny’s Lane: the environment Peat and mineral sediments are present at a depths below 2.90 m OD, underlying the base of this pollen profile which starts within the grey minerogenic sediment of layer 212. Sampling was not undertaken from deposits below the exposed section, and recovery of suitable samples from the auger hole (Fig. 9) was not feasible. Pollen was not present in the mineral sediments of unit 212 (below 2.90 m OD.). It seems likely, however, that the lower peats are of similar age and character to those described for Awkley Lane (above). Given the broad topographical extent of the Avon Levels, formation of the principal stratigraphical units is likely to have been autogenically controlled. The overall character of the levels is undoubtedly due to the status of, and changes in, the relative sea-level which affected the base level and extent of estuarine, salt-marsh, and freshwater ecosystems.

The lowest pollen spectra at the base of zone VIMP:1 suggests that the sediment (context 212) was deposited in a salt-marsh (possibly a mudflat, environment). This is evidenced by halophytes, especially the Chenopodiaceae which include goosefoots, oraches, and glassworts. Values of these increase substantially in zone 1a along with pre-Quaternary palynomorphs. This phase culminated in accumulation of detrital peat (context 211) possibly caused by a negative change in relative sea- level allowing the colonisation of marsh vegetation. Clapham suggests (below) that this was Phragmites autralis fen. However, the numbers of halophytic pollen taxa also suggests periodic estuarine incursions. This organic horizon has provided a valuable, later Neolithic, date for this phase of 2900–2620 cal BC (NZA-12527, 4182± 55BP). Subsequently (from 3.22 m OD), there is a change back to salt-marsh and/or mudflat sedimentation (context 210). Whilst there remain relatively high values of derived geological palynomorphs, these are in reduced numbers which may indicate a different sediment source to that of layer 212/211 (zone VIMP:1a). There is, however, a return to the higher values of tree pollen possibly derived from the fluvial catchment with Quercus (oak), Tilia (lime), Ulmus (elm), and Corylus avellana type (hazel and possibly sweet-gale) being the most important tree types. Pinus (pine) values (to 10%) typically reflects the over- representation of its saccate pollen grains in fluvial environments (Groot 1966) and are not considered important. This similarly applies to the occasional occurrences of Picea (spruce) for which there is an increasing number of records for long distance marine transport (Scaife 2000). Both Tilia and Fraxinus (ash) may be markedly under represented in pollen spectra (Andersen 1970; 1973). Values of Tilia, especially, are higher in sediments deposited during the Neolithic at Awkley Lane and this undoubtedly reflects the closer proximity to the site of drier and more favourable soils. At Vimpenny’s Lane, such land was over 0.5 km away, probably reflected in the smaller pollen frequencies of lime and ash. The pollen here was most likely fluvially transported and the occurrence of degraded pollen is a further indication of this and perhaps of sediment reworking.

At the top of VIMP:1 (at 3.74 m OD) there is a clear reduction in Tilia to low levels. This reduction falls below (by 20 cm) the zone VIMP:1/2 boundary which is delimited by a major reduction of Quercus and a possible hiatus in sedimentation. This reduction in Tilia pollen is also associated with an increase of Ericaceae (Calluna and Erica) and also herbs including Plantago lanceolata (ribwort plantain). It seems plausible that this is the often described ‘lime decline’ which has been similarly evidenced at Awkley Lane (above) and Hallen (below) in the Avon Levels and from the Somerset Levels (see discussion section). This event may provide a useful datum since the lime decline although an asynchronous phenomenon, due to its usual anthropogenic causation (Turner 1962), has been dated to the Middle–Late Bronze Age in many areas as discussed below. At Vimpenny’s Lane there is clear evidence of landscape changes associated with human activity at this time. Whilst waterlogging of the landscape may be responsible for reductions in lime pollen in certain circumstances (Waller 1994), here, the decline of lime seems to be associated with increasing evidence of human activity. Apart from the reduction of trees and increase in grassland, 47 there is some evidence of soil deterioration with increasing numbers of acidophiles (Erica, Calluna, and Sphagnum) suggesting woodland clearance (for agriculture?) and subsequent deterioration of the more sandy soils, possibly glacial sands of the lower Wentlooge Formation.

At 4.02 m OD there is significant evidence of environmental change (contexts 209/208) with marked reductions in trees (especially oak) and expansion of herbs (esp. Poaceae). There are also high absolute pollen frequencies at this horizon (to 200,000 grains/ml in context 208/207) but which subsequently decline sharply. This horizon is somewhat enigmatic and was not recognised in the field as a stabilistaion horizon. Two possible explanations may be suggested, first that the sediment sequence is broadly continuous and that there is a ‘real’ decline in trees at this time (that is, forest clearance) or second, that there is in fact a hiatus in the sediments so that these changes in woodland may have occurred during the intervening period. If the underlying lime decline was of Late Bronze Age date this cessation of sedimentation would apparently be of Late Bronze Age or Iron Age date. The reason for the highest absolute pollen frequencies/values at this event may be significant. If there was a palaeosol developed in this horizon (context 208), APF values would be expected to decline down the profile (Dimbleby 1988) rather than upwards as happens here. Interestingly, there is a reduction and absence of derived geological microfossils which suggests less sediment derived from marine sources and/or from bedrock erosion. It is possible that there was a change in taphonomy with perhaps freshwater sources introducing the Poaceae (and other herbs) from the inland zone.

Subsequently, the sediments and the pollen spectra continue to show that the on-site environment of deposition was one of salt-marsh and mudflat as demonstrated by the abundance of halophytes of which Chenopodiaceae are the most important. Between 4.80 and 5.04 m OD (zone 2b; context 205) there is a significant increase in Potamogeton type pollen. This taxon includes both Triglochin maritima (sea arrow grass) and Potamogeton (pond weed). In this sub-zone, there are also expansions of Plantago maritima (sea plantain) and Aster (various pollen taxa but also including sea aster) which suggests that the former (Triglochin) was most the likely taxon forming part of a middle–upper salt-marsh community. Thus, it is suggested that this sub-zone represents a partial drying out of the marsh with change from lower salt-marsh and mudflat to the middle and upper salt-marsh suggested.

After this transient drier phase, there was a return to mudflat conditions (contexts 203/202) and deposition of grey sediments containing much reworked geological pollen, dinoflagellates and increases of Pinus and Betula from fluvial sources. It is thought that these upper levels are of medieval or post medieval age.

A summary of the ecological and environmental changes Peat and sediment underlying the lowest levels of this profile were not available for analysis. It seems plausible that these underlying peats and mineral sediments are of similar age and character to those described and dated from Awkley Lane (above) forming in carr and/or sedge fen. At Vimpenny’s Lane the first pollen data from 2.90 m OD (base of VIMP:1; context 212/211) is of middle Neolithic age. From this point in the stratigraphy, the ecology and environmental changes are summarised as follows. i) Salt-marsh (VIMP:1a, 2.90–3.16 m OD.) (context 212) sedimentation as demonstrated by high Chenopodiaceae values and sediment character. This context also contains many reworked pre-Quaternary pollen/spores and dinoflagellates. The background, terrestrial vegetation comprised oak, lime and hazel woodland but possibly at some distance. Alder and sedges were fluvially transported from river valleys flowing into the estuary. ii) Stabilisation of the sediments caused by negative eustatic change (top of VIMP:1a, 3.14– 3.20 m OD) at c. 2750 cal BC. Middle salt-marsh habitat and formation of detrital peat from 48 marsh and fen vegetation (Chenopodiaceae, grasses, sedges)(context 211). Background vegetation remained one of oak, lime, and hazel – at some distance along with alder and Phragmites with occasional freshwater aquatic and marsh taxa (possibly sedges and bulrush and bur reed). iii) Return to lower salt-marsh, ?mudflat conditions and deposition of grey estuarine sediments (3.20 m OD; context 210). Evidence of minor freshwater input from in-flowing rivers comes from occasional pollen of aquatic macrophytes. Background woodland remained one of oak, ash, lime, and hazel. There is, however, evidence of reduction in lime woodland (possible the lime decline) and increasing acidification of local acid sub-strates with increases in Ericaceae (Calluna and Erica) and expansion of Sphagnum. Whilst this may suggest changing taphonomy, ie, pollen transported from different sources, it is also plausible that this is evidence of increasing human activity and consequent soil degradation – possibly due to increased human activity in? the Bronze Age. This is to some extent evidenced by the expansion of ‘typical;’ anthropogenic herbs such as ribwort plantain (Plantago lanceolata) iv) Possible Hiatus and/or truncation of the sediments at 3.98 m OD (context 208/209). v) Woodland clearance: At this point in the pollen profile there is a marked reduction in tree (oak) and shrub (hazel) pollen. vi) Continuation of salt-marsh (above 3.98 m OD; contexts 207–202). Possible changes in taphonomy give, initially, markedly high APF values above this interface but declining to the low values which are typical of these salt-marsh sediments. The background vegetation shows a general reduction in trees and in particular oak and hazel (lime had been removed prior to hiatus). There is an increase in herb diversity including evidence of human activity. The acid loving taxa (Ericaceae and Sphagnum) remain. The high APF values (context 208/207) also correspond to a reduction in oak percentages and especially low levels and/or absence of pre-quaternary palynomorphs and dinoflagellates. This suggests changes in sedimentary regime and taphonomy. It is possible that this may be a stabilisation phase (although not considered to be a soil) with build up pollen on the surface whilst there was less input from marine/estuarine sources with contributory derived geological microfossils and oak pollen from the regional vegetation. vii) Drier (?middle-upper) salt-marsh phase. 4.80–5.04 m OD (context 205). A peak of Potamogeton type is most probably Triglochin (sea arrow grass) since there are also expansion of allied taxa such as the Chenopodiaceae (oraches, glassworts etc.) sea plantain (Plantago maritima type) and sea aster (Aster type.). viii) Return to lower salt-marsh/mudflat conditions. Increase in pine, birch, and fern spores from increase fluvial sediment input/transportation.

Diatom analysis by Nigel G. Cameron

The two diatomaceous samples counted from Vimpenny’s Lane, the detrital humic peaty lens (211) at 3.18 m OD and laminated grey alluvium (210) at 3.26 m OD respectively, have similar diatom assemblages. Marine diatoms are dominant, comprising over 65% of the assemblage in the lower sample of the detrital humic peat (211) and almost 85% in the laminated grey alluvium (at 3.26 m OD). In both samples marine-brackish (polyhalobous to mesohalobous) and brackish (mesohalobous) groups each account for approximately 5% of the diatom assemblages. There is, however, a significant difference in the size of the freshwater (oligohalobous indifferent) component of these samples. In the detrital peat (210) layer, freshwater diatoms comprise almost 20% of the total, whereas in the laminated grey alluvium (210) freshwater diatoms account for only about 2% of the assemblage. Fragilaria pinnata, Fragilaria construens var. venter, and Fragilaria brevistriata are the main freshwater components at +3.18 m OD. Paralia sulcata and the marine

49 tychoplanktonic (the tychoplankton spend part of their lifecycle associated with non-planktonic or bethic habitats, in contrast to the plankton which complete their entire life-cycle in the open water) diatom Cymatosira belgica are the dominant marine species in both samples.

Diatom assemblages like that of Vimpenny’s Lane in the humic peat, which are composed of disparate marine and freshwater elements, and with a relatively small brackish water element, have been observed elsewhere in the Severn Estuary (Cameron 1997). In this case tidal waters are the most important source of diatoms. However, the appearance of a significant percentage of freshwater species can be attributed to a number of factors. For example, freshwater diatoms can be transported from their life habitats a short distance inland. The reason for the poor representation of mesohalobous taxa is unclear but the hyper-tidal nature of the Severn may have resulted in erosion and loss of brackish water habitats during transgressive and regressive phases. Further, considering the ecology of the dominant freshwater diatoms, those Fragilaria species present are opportunistic taxa which, despite their optimal growth in freshwater, appear to have wide salinity tolerances and can be found in significant numbers in habitats periodically exposed to higher salinities. However, consistent with its estuarine flora, the sample 3.28 m OD is a grey laminated alluvial clay (210). The sample from +3.18 m OD is a humic/peaty stratum (211), which is consistent with the significant component of freshwater diatoms in the assemblage. The dominant environmental factor in both cases is the influence of water from the estuary.

The diatom assemblages examined are predominantly estuarine with a large marine planktonic component typical of the Severn Estuary. Marine diatom plankton is commonly transported by tides to sites of lower salinity than that for the species optimal growth. The groups are nevertheless useful in indicating the predominance of tidal conditions.

The diatom assemblages show that detrital peat layer (context 211) has a mixed marine to freshwater diatom assemblage. Although it is dominated by marine plankton, this sample has a significantly greater percentage of freshwater taxa when it is compared with the other diatom assemblages that have been analysed from the minerogenic sediments. The freshwater component of this sample matches the context description which shows that the sediments have a high organic content. Diatom assemblages composed of taxa with widely different salinity preferences are common in coastal and estuarine sediments in general and in the hypertidal Severn Estuary in particular (Cameron 1993; 1997; Cameron & Dobinson 1998; 2000; Walker et al. 1998a; 1998b) and this mixing may be particularly marked at zones of transition between marine and freshwater sediments.

In summary, the diatom assemblages of all four samples are dominated by a flora common elsewhere in the Severn Estuary. The marine planktonic diatoms Paralia sulcata and Cymatosira belgica are dominant, but a large number of other marine and estuarine taxa are common, for example: Rhaphoneis spp., Podosira stelligera, Thalassiosira decipiens, and Pseudopodossira westii. The sample from Awkley Lane, 4.42 m OD has a greater percentage of the mesohalobous (brackish) non-planktonic taxon, Nitzschia navicularis. Whilst this species is rare at the other sites, where the marine tychoplanktonic species Cymatosira belgica is more common, it is an estuarine diatom associated with high salinity levels. This difference may be a reflection of slightly less saline conditions compared with the other samples. Alternatively the difference in diatom assemblages may also reflect a difference in depositional environment or suitable substrate for diatom growth as much as a reduced salinity.

It should also be noted that the freshwater diatoms that are present in these samples, for example several oligohalobous indifferent Fragilaria spp. and Navicula spp., are species tolerant of high

50 Table 6. Mollusc data from Vimpenny’s Lane

Sample 1000 1001 1002 1003 1004 1005 1006 1007 1008 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 b Context 209 208 208 207 206 206 206 206 206 205 205 204 203 203 203 202 201 201 200 Depth (mOD) 3.91- 4.01- 4.11- 4.21- 4.31- 4.41- 4.51- 5.61- 5.71- 6.85- 4.95- 5.05- 5.15- 5.25- 5.35- 5.45- 5.55. 5.65- 5.75- 3.81 3.91 4.01 4.11 4.21 4.31 4.41 5.51 5.61 5.75 4.85 4.95 5.05 5.15 5.25 5.35 5.45 5.55 5.65 Wt (g) 1750 1750 1750 1750 1750 1750 1750 1750 1750 1700 1750 1750 1750 1750 1750 1750 1750 1750 1750 Vallonia pulchella (Müller) - - 1 - - - 1 2 1 ------Land mollusc Taxa 0 0 1 0 0 0 1 1 1 0 0 0 0 0 0 0 0 0 0 LAND MOLLUSC TOTAL 0 0 1 0 0 0 1 2 1 0 0 0 0 0 0 0 0 0 0 Bithynia tentaculata (Linnaeus) Lymnaea truncatula (Müller) - - - - - 1 ------Lymnaea palustris (Müller) - - - - - 1 - - - 1 ------Lymnaea spp. ------1 - - - - 1 - - - - - Gyraulus albus (Müller) ------3 - - Hydrobia ventrosa (Montagu) - 1 - 106 117 - - 21 6 36 31 4 11 7 - - - - - Hydrobia ulvae (Pennant) - 2 - 2 6 - - 1 - - 5 5 - 3 - - - - - Hydrobia spp. - - - 7 5 - 2 3 - - - 4 ------Ovatella myosotis (Draparnaud) ------1 ------Aquatic Taxa 0 2 0 2 2 2 1 2 2 2 3 2 1 3 0 0 1 0 0 FRESHWATER TOTAL 0 0 0 0 0 2 0 0 1 1 0 0 0 1 0 0 3 0 0 MARINE TOTAL 0 3 0 115 128 0 2 25 6 36 37 13 11 10 0 0 0 0 0 Ostracod (valves) - - - 17 46 - 2 3 23 17 2 - - - - - 28 - - TOTAL 0 3 1 115 128 3 3 27 8 37 37 13 11 11 0 0 3 0 0

(note: only samples with shells tabulated: totals exclude C. acicula)

51 salinities and are often associated with salt-marsh and high tidal mudflats (especially in areas with a strong freshwater influence) despite their optimal growth in freshwater (Denys 1988).

Molluscs by Michael J. Allen

Only the upper 2 m of the alluvial profile (above 3.8 m OD) was sampled for molluscs in a series of 19 contiguous samples (Table 6). Shells below this level seemed absent in the field, but a low presence of Hydrobia ventrosa and of the estuarine species Phytia mytosis were present in the evaluation trench at this level (Bell & Johnson, in Lawler et al. 1992, 109). The putative stabilisation horizon at 3.9–3.8 m OD is devoid of shells, but again Hydrobia ventrosa and cf. Macoma were present in evaluation trench (op. cit.). The only point in the sequence where shell numbers are moderate is at 4.3–4.1 m OD, at the interface of two visually different layers (207, very dark grey (10YR 3/1) slightly humic silty clay and, 206, a grey (2.5Y N/5) silty clay with some organic matter). Only Hydrobia species are present, mostly H. ventrosa. Although not common today, there are old records of its existence in drainage ditches, rhines, and pills (Boyden et al. 1977, 524). This suggests a quite brackish creek or lagoon environment, and the higher numbers may indicate ponding of salty brackish water, perhaps in a local backfen environment, or washed in by flood events. Notably, this high occurrence of solely Hydrobia species is precisely coincident with the only occurrence of the rare salt-marsh pool plant Ruppia maritima in the two levels of the pollen diagram.

Immediately above this level (layer 206) shell survival is poor but the presence of Lymnaea species indicates the presence of freshwater discharging into this area and may suggest a changing environment. The presence of a few terrestrial species might also indicate drier salt-marsh conditions. Higher up the profile again (4.85–570 m OD; essentially layer 205) shell numbers increase, dominated by Hydrobia ventrosa indicating wetter conditions and probably mudflats. Rare instances of Lymnaea species indicate some freshwater conditons and Ovatella myosotis may suggest rotting debris on mudflats (Macan 1977; Janus 1979, 66). Above 5.15 m shells are largely absent. Three freshwater specimens of Gyraulus albus just below the present day soil profile are probably washed in, though significantly, the occurrence of ostracod valves increase at this point too (Table 6).

Insect remains by Mark Robinson

A sample of 2 kg from the detrial peat deposit (211) within the Vimpennys Lane alluvial sequence was subjected to paraffin flotation to recover insect remains. The flot was caught on a 0.2 mm sieve, washed in detergent and sorted under a binocular microscope. The results are listed below:

Coleoptera Min. no. indiv. Megasternum obscurum 1 Ochthebius minimus 1 Micropeplus caelatus 1 Stenus sp. 1 Pselaphidae indet. 1

The concentration of insect remains was very low and their preservation poor so no more of the sample was analysed. However, one of the species of Coleoptera (beetles) is of particular interest. Micropeplus caelatus is now extinct in England although it is still to be found in the British Isles in the south-west of Ireland. It inhabits swampy or marshy ground, where it occurs in decaying plant

52 refuse. Several examples of M. caelatus were, moreover, found in Neolithic reedswamp deposits at the base of the peat sequence of the Somerset Levels at the Rowland’s Track site (Girling 1977, 53). The sparse insect assemblage from Vimpenny’s Lane would be appropriate to vegetation debris in a reedswamp.

Waterlogged plant macrofossils by Alan J. Clapham

Two sub-samples of 300 and 250 cm3 from the organic detrial humified peat horizon (211), were analysed for waterlogged plant macrofossils. This layer has been dated to 2900-2620 cal BC (4182±55BP, NZA 12527). The sample was found to contain only a limited number waterlogged plant macrofossils (Table 7).

The waterlogged plant macrofossils were well preserved and allowed positive identifications in most cases. A total of 22 taxa was recovered; there were very few large plant macrofossil remains. The largest were fragments of reed (Phragmites australis). Charcoal fragments were also recorded with the majority being pithy in nature suggesting that some stems of reed had been burned. Charred culm nodes which can be assumed not to be of a cereal were also identified supporting this interpretation.

Tree species were represented by a single find of a birch (Betula) fruit. As the fruits of this tree have wings and only one example was recorded it may be that the origin of this fruit were extra- local, although the majority of birch seeds are known to fall within a short distance of the parent tree (Gimingham 1984). The dominant habitat was that of wetland/fen/reedswamp – taxa recorded include celery-leaved buttercup (Ranunculus sceleratus – which is also common in shallow brackish water in coastal marshes), stinging nettle (Urtica dioica), woody nightshade (Solanum dulcamara), gipsywort (Lycopus europaeus), watermint (Mentha aquatica), reed, and reedmace (Typha sp.). The presence of mosses also suggests damp conditions. Oak-leaved goosefoot (Chenopodium cf glaucum), common orache (Atriplex patula), and the Chenopodiaceae in general, along with docks (Rumex sp.) and cleavers (Galium aparine) are most often associated with disturbed, arable, or waste ground, but their presence here may be due to areas of the marshy/fen ground having been disturbed by some activity; whether this was natural or anthropogenic in nature is difficult to determine. The same can be said for the grasses (bent (Agrostis type) and couch (Elytrigia type) although, again, these grasses can be found in natural habitats. A scrub type habitat may be indicated by the presence of bramble (Rubus Section 2 Glandulosus) although, like the cleavers, these can be found in fen/reedswamp conditions growing over the other vegetation along with woody nightshade.

53 Table 7. The waterlogged plant remains from the detrital peat (context 211) from Vimpennys Lane

Context no. 211 211 Overall, the type of habitat indicated Sample no. 1019 1019 is tall fen/reedswamp vegetation O.D. 3.14 3.14 dominated by reedmace and reeds, 14C yrs BP(uncalibrated) 4182 ± 55 4182 ± 55 3 with some stinging nettle over which Volume processed cm 300 250 the bramble, woody nightshade, and

Musci common - cleavers are scrambling. Below this Ranunculus sceleratus 99+99f 46+69f tangled mass of vegetation, the celery- Urtica dioica 5 - leaved buttercup, gipsywort, and Betula sp. seed 1 - watermint can be found growing. In Chenopodium cf glaucum - 5 some areas where there has been some Atriplex patula 5+18f 8+19f disturbance oak-leaved goosefoot, Chenopodiaceae 5 61 common orache, and the grasses Rumex sp. 1 - occur. This type of reedswamp Rubus Section 2 Glandulosus 1f - Rosaceae thorn/prickle 1 - vegetation is nowadays often found Solanum dulcamara 36f 23f around ponds, lakes, and at the sides Lycopus europaeus - 3 of rivers where the water-flow is slow Mentha aquatica/arvensis 43+32f 52+47f or standing and in low lying areas Galium aparine 14+7f 7+19f where the watertable is high. This Asteraceae indet 14 3 type of vegetation can cover large Agrostis type caryopsis - 10 areas. The presence of the pithy Elytrigia sp. caryopsis 1 100+ charcoal and the charred culm nodes Phragmites australis rhizome fragments - 2 suggests that some of the taller Typha latifolia/angustifolia 21 30 Charcoal 80f - vegetation had been burnt at some Charred culm nodes - 26 stage. Whether this is a natural or Miscellaneous 17 - anthropogenic event is difficult to say, although the presence of the disturbed habitat plant species and the grasses may suggest that there was some limited human activity in the area, possibly to clear some of the tall, tangled vegetation to gain access to more open water for fishing or other hunting activities.

3. Hallen Marsh

The sequence

Several sherds of Iron Age pottery and other occupation remains were found in a ‘humic’ layer 0.9 m beneath the surface during observations of a geotechnical pit at Hallen Marsh (ST 3543 1804 and 6.4 m OD) (Russett 1990, 5). A number of machine excavated trial trenches were cut to clarify the nature of the remains (Lawler et al. 1992, 79–84). This evaluation identified a series of archaeological features comprising a rubble spread and deposits rich in artefacts, in particular Iron Age pottery, sealed beneath the ‘humic’ layer. Given the undoubted potential of this, one of the first prehistoric occupation site identified on the Avon Levels, a full archaeological excavation was conducted clearing an area of 1600 m² for examination (Fig. 12).

Within this cutting, detailed archaeological excavation was restricted to two areas (Fig. 12) where remains of Iron Age occupation had been identified, including two round-houses with surrounding drip-gullies that had been recut/re-established on several occasions (Plates 4 & 5); a third similar

54 area was recorded but not investigated. The archaeological remains and associated cultural debris are reported elsewhere (Barnes 1993; Gardiner et al. 2002).

The sequence at Hallen Marsh was exposed to a depth of only 2.5 m max (4.10 m OD; Fig. 13) but, by virtue of the archaeological excavations, a long section of some 40 m+ was exposed (Fig. 14). This provided an indication of lateral variation in the sediment sequence. Over much of the site only c. 1.4 m (to 4.36 m) of the deposits were exposed. Although the present surface topography is essential flat and level, the occupation deposits occurred under the upper reddish-brown alluvium on slight natural rises separated by stretches of alluvial infill. Because of this variation, two main sequences were described (Figs 13 & 14); one where occupation deposits were encountered and a second through the stratigraphy between two major foci of Iron Age occupation.

Non-occupation sequence (Fig. 14) Beneath the topsoil, uniform deposits of alluvial clay loam (100 and 101) with frequent ferric and manganese flecking, which providing the reddish brown hue, occurred across the entire area regardless of the occurrence of archaeological deposits. This alluvium was up to 1.1 m thick, but only 0.5 m thick over the archaeological deposits. At the base of this ‘upper alluvium’ a zone of more intense mottling occurred (102 and 151) in a greyish brown (10YR 5/2) silty clay. Where this overlay the archaeological deposits it was only 0.1 m thick (102) but was appreciably thicker where no archaeological deposits were encountered. Between the ‘islands’ of archaeology, thicknesses of up to 0.7 m of gleyed grey (2.5Y N7) and pale brown (10YR 6/3) alluvium (152 and 186 respectively) with lighter grey veins occurred (151). Between the ‘islands’ of archaeology, this local ‘channel-fill’ alluvium was up to 1.2 m deep.

Beneath both the archaeological deposits and the local channel-fill alluvium was a dark grey gleyed mottled and veined clay (184) which was at least 0.75 m thick (exposed to c. 4.1 m OD). It was laminated in some places and blocky elsewhere and, was typically mottled with diffuse orange, ferruginous colours, and streaked and veined with predominantly light grey colours. This alluvium (183 and 184) occurred at c. 4.65 m OD between the occupation areas, but was elevated and nearly 1 m higher (c. 5.4 m OD) where the occupation occurred. This was the ‘grey alluvium’ recorded elsewhere. The upper 0.2 m of this (183) was sandier (sandy clay) and a distinctly browner hue was probably the oxidised surface of the upper Wentlooge Formation hue.

Occupation sequence The occupation deposits were identified beneath the upper reddish brown alluvium as three areas of darker, greyer deposits rich in fragments of pottery, burnt clay, and animal bone. These were light grey (5Y 7/1) to grey (5Y 6/1) and dark grey (5Y 4/1) silty clay up to c. 0.15 m thick, with iron and manganese staining and weak medium blocky structure indicating a post-occupation soil formation. This layer sealed the surfaces, structures and ditches of the settlement and is thought to represent the development of a soil above the abandoned site. The ‘occupation’ deposit, as defined by the field archaeologists, was a light grey (10YR 7/2) clay up to 0.12 m thick with a compact surface (trampling and ?crusting) and distinctly greenish and olive (5Y 6/4) hues and was restricted to the areas of settlement features.

The surface beneath the occupation deposits was not obviously eroded and it is assumed that the lower levels between the occupation foci, although possibly exacerbated by later erosive and flooding events, are essentially the true form of the Iron Age topography. The occupation foci were located on slightly higher rises within an essentially flat and level alluvial plain. As there is no stratigraphic relationship between the channel and the settlement area, and a lack of artefacts within the channel the channel fills there is no way of telling whether the form of the ‘channel’ between the settlement areas has been deepened subsequent to settlelment and prior to infill by alluviation.

55 Occupation sequence Depth (m) OD Layer Description Comment 0-0.14 6.43-6.29 Topsoil 0.14-0.30 6.29-6.13 100 Greyish brown (10YR 5/2) stonefree clay loam, common Upper alluvium fine and medium fleshy rots, rare charcoal flecks, and rare small limestone fragments 0.30-0.59 6.29-5.84 101 Light yellowish brown (10YR 6/4) clay loam with rare fine Upper alluvium fleshy roots, - a reddish appearance is given to these layers by the frequent ferric and manganese flecking (gleying), gradual to diffuse smooth boundary to layer below (102) which is distinctly greyer. 0.59-0.66 5.84-5.77 102/ Light brownish grey (10YR 6/2) silty clay with rare small post occupation 151 limestone fragments, flecked with Fe ad Mn staining. Seals alluviation the archaeological occupation deposits. 0.66-0.85 5.77-5.58 105 Light grey (5Y 7/1) to grey (5Y 6/1) silty clay, weak Occupation medium to large blocky peds with moderate Fe and Mn staining, becoming darker in colour where highest concentration of artefacts occur and conversely distinctly lighter away from these three defined foci (buildings). It contains patches and spreads or lumps of buff brown clay indicating mixing and post possibly deposition colour changes. Occupation layer 0.85-0.91 5.58-5.52 321 Light grey (10YR 7/2) clay with greenish hue and compact Occupation surface surface 0.91+ 6.52+ 183 Brown (10YR 5/3) gritty gleyed clay, mottled with grey and Grey alluvium orange streaks. It occurs as a thick band which is laminated in some laces and blocky elsewhere; pre-dates the occupation, diffuse boundary

Non-occupation sequence Depth (m) OD Layer Description Comment 0-0.11 6.56-6.45 0.11-0.51 6.45-6.05 100 Greyish brown (10YR 5/2) stonefree clay loam, common Upper alluvium fine and medium fleshy rots, rare charcoal flecks, and rare small limestone fragments 0.51-1.15 6.05-5.41 101 Light yellowish brown (10YR 6/4) clay loam with rare fine Upper alluvium fleshy roots, - a reddish appearance is given to these layers by the frequent ferric and manganese flecking (gleying), gradual to diffuse smooth boundary. 1.15-1.33 5.41-5.23 151 Greyish brown (10YR 5/2) silty clay with lighter grey veins post occupation and mottling, diffuse alluviation 1.33-1459 5.23-5.07 152 Light grey (2.5Y N7) gleyed clay, rare small stones; gleying Upper alluvium creates a blueish hue. The relationship with layer 105 is not channel fill; clear. Gleyed channel infill 1.49-1.72 5.07-4.84 186 Pale brown (10YR 6/3) clay with some mottling and orange Upper alluvium and grey streaks channel fill; 1.72-1.85 4.84-4.71 183 Brown (10YR 5/3) gritty gleyed clay, mottled with grey and Grey alluvium orange streaks. It occurs as a thick band which is laminated in some places and blocky elsewhere; pre-dates the occupation, diffuse boundary 1.85-2.46 4.71-4.10m+ 184 Dark grey (10YR 4/1) gleyed clay, slightly mottled and Grey alluvium veined. Wentlooge series

Sampling the sequence

Two main locations within the sequence were sampled. One was through the northern occupation area and building (Fig. 14), where disturbed bulk samples were taken through the sequences of land snails. The second location was within the ‘palaeochannel’ between the occupation areas. Two sample sequences here comprised a monolith sequence through the alluvium and the deposits underlying all settlement activity, and a comparable sequence of bulk samples at 0.1 m intervals taken 5.5 m to the north (see Fig. 13).

56 WA Hallen Marsh

Section Fig.7

Plate 4. Round-house 1 at Hallen after excavation

House 1

House 2

Plate 5. Round-house 2 at Hallen after excavation

0 10 m

Figure 12. Hallen Marsh: Summary plan of the excavation

57 WA Hallen GGAT G018

100 100 6 m O.D. 101

102/151

183 0 10 20 30 x10-8 SI/kg 5 m O.D.

WA Hallen

WA Hallen 100

6 m O.D.

101 101 Pollen Zones

151 151 2 152 5 m O.D. 186 152 183 186

1 184 184

0 10 20 x10-8 SI/kg 4 m O.D.

Alluvial silty clays (brownish red hue) Alluvial silty clays (bluish grey hue) Alluvial silty clays (colour unknown) 01 m

Figure 13. Hallen Marsh; full site profiles

58 A. WA Hallen Summary S N B1 B2

1 m Section break see below 5 m Vertical scale Vertical Horizontal scale

B. Detail 7 m O.D. S N B1 100 6 m O.D.

101

181 5 m O.D. 182 183 184 4 m O.D. Continued below

7 m O.D. S N B2 100 6 m O.D.

101

181 5 m O.D. 182 183 184 Continued above 4 m O.D.

Main post-occupation alluvial sequence 1 m Primary post-occupation alluvial sequence Humic land surface above abandoned settlement Alluvial silty clays (brownish red hue) 5 m

Iron Age occupation deposits scale Vertical Horizontal scale Alluvial silty clays (bluish hue)

Figure 14. Hallen Marsh Wentlooge Formation profiles

59 Textural features by J.R.L. Allen

Sixteen sediment samples from the channel between the settlement foci were analysed and grain- size patterns are shown graphically (Fig. 15). The triangular graph shows the scatter of values of the clay-silt-sand ratio and shows the sediments to range from clayey silts to sandy-clayey silts and to be broadly similar to other, similarly situated salt-marsh deposits from the outer Severn Estuary.

The stratigraphic sequence of samples reveals three textural patterns. The earliest few samples (layers 184–183; 4.46–4.76 m OD) display no overall textural trend. From top of layers 184–151 (4.86–5.33 m OD) there is a marked fining-up trend. A strong coarsening-up trend is evident from the top of the sampled sequence (layer 101; 5.43–5.73 m OD), the two topmost samples being coarser-grained than any others in the profile.

Textural patterns from Holocene salt-marsh sequences cannot be given unequivocal interpretations as two reinforcing factors are at work (Allen 1995). Firstly, the mud deposited over a marsh is advected by the tide, which enters the marsh across its edge and along the major creeks. Because the heavier particles settle out earliest, the deposit consequently becomes finer-grained with increasing distance inland. If the marsh-edge is retreating, the successive deposits formed at a point on the marsh will increase in coarseness over time. The opposite trend is expected if the marsh edge is building out. The latter is favoured by a negative sea-level tendency – a decline in the rate of sea- level rise – whereas a positive tendency – an increase in the rate of sea-level rise – encourages marsh-edge retreat. Secondly the sea-level tendency influences the height of a marsh relative to the tidal frame and, therefore, the general coarseness of the sediment advected over the marsh. During times of positive tendency, the marsh surface is depressed, allowing coarser sediment from faster- moving water to enter, The opposite effects are expected during a period of negative tendency. in the case of the Hallen Marsh profile, the site seems to have experienced (1) a period of stability of the marsh edge and surface, followed by (2) a period of negative sea-level tendency and/or marsh outward growth, succeeded in turn by (3) a period of marked positive sea-level tendency and/or marsh-edge retreat.

Pollen analysis by Robert G Scaife

A number of questions were posed by the archaeology which it was anticipated that pollen analysis could answer. These questions relate to the environment of deposition of the lower grey sediments (middle Wentlooge Formation) and an overlying alluvial deposit. Furthermore, it appears that the Iron Age settlement was abandoned and lies under thick alluvial silts. It was anticipated that multi- disciplinary analyses might demonstrate the environment and causes of this settlement abandonment and the basic elements of environmental and habitat change.

Pollen was extracted from all 17 samples examined. Absolute pollen frequencies were generally low in the largely minerogenic sediments and this dictated the pollen sum counts made. However, with the exception of the uppermost level, counts of greater than 300 grains per level were made throughout thus providing statistically adequate data. These pollen data are presented in standard pollen diagram form (Fig. 16).

Absolute pollen frequencies were variable, ranging from lowest values at the top of the profile of 1700 grains/ml to 67,500 grains/ml at 5.24 m OD – values are higher in the

60 Hallen Marsh Clay (100%)

50% 50%

Clay

Silt (100%) Sand 50% Sand (100%)

Figure 15. Hallen Marsh; particle size triangle

61 uppermost of the two recognised pollen assemblage zones. In pollen zone HALL:1, APF values are in the order of 10–15,000 grains/ml. Thus, the values and available pollen for counting is, in general, small. Pollen preservation was also variable with some deteriorated grains of Lactucoideae especialy in zone HALL:2 indicating longer and differential preservation of this taxon. However, the majority of pollen grains were moderately well preserved in spite of the low absolute pollen frequencies and the minerogenic character of the sediments. This is possibly a product of the taphonomy and alluvial preservational environment.

Two distinct local pollen assemblage zones have been delimited in the Hallen Marsh pollen sequence. These are delimited and characterised from the base of the profile as follows.

HALL:1 4.20–4.80 m OD, Quercus-Corylus avellana type-Chenopodium type-Plantago maritima-Poaceae. This zone is characterised by higher values of tree pollen than in zone HALL:2 with substantially higher values of halophytes-Chenopodiaceae and Plantago maritima. Trees are dominated by Quercus (to 43%) especially in the lower levels of the zone. Betula (7%), Pinus, Ulmus, Tilia, and Fraxinus are also more abundant but values are relatively small. Corylus avellana type is the principal shrub (18%). Herbs are important (to 60% of pollen) dominated by Poaceae (20–30%) with Chenopodium type (to 16%), Plantago maritima (peak to 19%) and Plantago lanceolata (10%). Lactucoideae are incoming mid-way through the zone. Freshwater marsh taxa comprise Alnus (to 8%) with Cyperaceae (8%). There are sporadic records of other fen taxa including Typha angustifolia/Sparganium type, Typha latifolia and Littorella uniflora at the top of the zone. Halophytes are Chenopodiaceae, Plantago maritima type, Spergularia type, Aster type and Potamogeton type. The latter, Potamogeton type may be Potamogeton and/or Triglochin. Spores of ferns include dominant Pteridium aquilinum (peaking to 50% at 4.60 m OD). Dryopteris type (monolete) spores (average 10–12%) and Polypodium vulgare (6%) are most important. There is a small but consistent presence of Sphagnum moss spores. Pre-Quaternary palynomorphs (including Jurassic-Cretaceous Cycadopites) are consistent (10%). Of note is a peak of Dinoflagellates at the top of the zone.

HALL: 2: 4.80–5.48 m OD, Plantago-lanceolata-Lactucoideae-Poaceae. This upper zone is delimited by reductions in Quercus (to 10–15% av.), and Betula, Pinus, Tilia, and Alnus and the herbs Plantago maritima type (to <2%), Chenopodiaceae (5%), and possibly Potamogeton type (if Triglochin). Contrasting are significant expansions of Plantago lanceolata (to 35%), Plantago coronopus type (8%), Lactucoideae (8%), Poaceae (50%), and some increase in Cereal type. Plantago lanceolata and Poaceae are dominant. Within the trees and shrubs, Quercus and Corylus remain the most important and consistent taxa. Fagus is present in the lower half of the zone. Although there is a substantial reduction in the halophytes Chenopodiaceae and Plantago maritima remain. Marsh taxa comprise Alnus reduced to (4–5%) with Cyperaceae expanding to a peak of 15% at the top of the profile (5.32 m OD). There are occasional freshwater aquatic/megaphytes (Myriophyllum alterniflorum, Lemna, and Typha/Sparganium type and algal Pediastrum). Spores remain dominated by Pterdium aquilinum (expanding to 55%) with Dryopteris type (to 14%) and Polypodium vulgare (values increasing upwards). There is a peak of Dinoflagellates/Hystrichospheres at the lower zone boundary (4.84–4.68 m OD). There is also an expansion of Pre-Quaternary palynomorphs (Jurassic).

Vegetation and Environmental Change at Hallen Marsh The two pollen assemblage zones represent a distinct change in environment and pollen taphonomy. The Iron Age archaeology of the Hallen site corresponds with this horizon of change at c. 4.80 m OD.

Pollen zone HALL:1 contains substantial quantities of halophytes including Chenopodiaceae (includes Chenopodium, Atriplex, Salicornia), Plantago maritima (sea plantain) and possibly Aster type (?sea aster) and Potamogeton type (?Triglochin-sea arrow grass). These halophytes and the character of the sediments suggest that the on-site environment was one of salt-marsh (Upper Wentlooge Formation). This would have been subject to brackish water/marine inundation at mean high water spring tide.

62 A: Hallen

TREES & SHRUBS HERBS

PE F. PE Y F T TY PE DI E Y RE E TA T P E UNY RA . T OPUSO . NA PE PE F L F GRADED Y Y 1 CEA TIMA TYPEON F E E E IF E E PE OR CEOLA F LLA T YPE T Y A RI R PE E DI A /D d) E T M P A ND YP YP AN UNI NIGRA TYP E D PE V U T T T MAJ L CO PE E EA N CE de A US PE U A ARI Y P E YPE A D U P A IE ra TY M IC AE M LA T STYA T F S g S A TY AE UL I RE OI e US U CUL ER CE GO IU H GO GO GO MAGO TYPM SI AE S d IS HUS TY OPODIUME TYPEL EAE TYPEX P A REL ISA ( RN P C UM NS TY AU CEAE E PO RCU A A N AC ICA O ENTILL YGONUMO AVICULA NT I C ER HE EMI G US MU E I I AXINU ICA LLUNNUN LVA B D IF IA ME R ANTAANTA ANTATO T CTUCIA REAL TYIDENTI Strat. ICEA L L BU NA A IANT A E P C T UC DE RSIUM A L AR BETULAPIN P UL QU TI TI FR FAGUSPRUNUSCORYLVI ER CA RA SI HORNUNGIAcf. HYPM D SPERGULARICHE F M TR POTA POLRU S PL PL PLA PL LI GALS BI ASTANTAR CI CENTL LI POACE CE L UN Zone 5 10

15 20 25 30 35 Zone 2 40 45 50 55 60 65 70 Depthcm 75 80 85 90 95 100 Zone 1 105 110 115 120 125 20 40 20 20 20 20 40 20 20 40 Sum = % Total Dry Land Pollen For Key See Figure 7a

Figure 16a. Hallen pollen diagram.

63 B: Hallen cont.

MARSH SPORES MISC. TOTALS

HS M RU LO PE M YNOMORP A E TY E AT U E IA AL OR P L IS AR CT P l. O S N TYPE TY A TYPF LI PE LG U RY /m I I STR A TES s G VU P A in UM ALTERNIF UNIFL E U S L a ES UST C STY M S TAL A G EA A M IU ERNA S TH LL OTYLEATIFOLN L ERI UM RUMGEL S TO 0Gr L A ORT E N OGETONE T OD N LA ARI S S A M OC RAC IDIUM AQUILINUMP P G F S S EN .X1 A OR R HA HA R Y RW HOCERA SC E B L F D P P PE ETESUISETU E DIAST E ORE ITT Y Y O SMUNDA RE OL PHA V NO P. ALNUS SALIXMYRIOPHYLLLEMNAMENYPOTL H TY TY C IS EQ O PTE DRYO P S LI ANT PE DI PRE-QUAT cf. A TR SHRUBERICALHER MARSHSP MISC POL A. Zone

5 10 15 20 25 30 35 Zone 2 40 45 50 55 60 65 70 Depthcm 75 80 85 90 95 100 Zone 1 105 110 115 120 125

20 20 20 40 60 20 20 40 20 40 60 20 20 40 60 80 20 20 40 60 80 20 40 60 400 200 400 600 800

Note that alder (Alnus sp) is included with fen species in Fig 16b)

Figure 16b. Hallen pollen diagram.

This zone also shows evidence of woodland dominated by Quercus (oak) with Corylus avellana type (hazel or possibly sweet gale) with some lesser representation of Tilia and Fraxinus (ash). The latter are, however, usually very under represented in pollen spectra (Andersen 1970; 1973) and as such, may have been more important on drier soils. However, given that even these values are small and the possible taphonomy of the pollen it seems likely that we are seeing the last vestiges of areas of previous important lime dominated woodland such as described for Awkley and Vimpenny’s Lane. Betula (birch), Pinus (pine), and Ulmus (elm) are considered here to be of extra-local origin.

The change to zone HALL:2 at 4.80 m OD shows a change in environment from upper salt- marsh to apparently one of pasture ‘on-site’. Poaceae (grasses) are dominant with substantial quantities of Plantago lanceolata (ribwort plantain) and sporadic herbs typical of pasture (Fabaceae) and Lactucoideae (dandelion types). The pasture was probably not short cropped/grazed or tall herb pasture but was probably wet meadow of medium length pasture (based on the presence of Plantago lanceolata flowering). Plantago coronopus is also present and increasing in zone HALL:2. This is more typical of coastal grassland swards on sandy soils and dunes and unstable ground. It is possible that the pollen source was from path/trackways on the more sandy sediments of the floodplain or from near coastal grassland habitats (dunes?).

Reduction in tree pollen from zone HALL:1 represents an important taphonomic change. Zone HALL:1 saw the accumulation of salt-marsh sediments along with the typical salt- marsh/marine silts and included pollen. This would have been from periodic (possibly monthly) high water inundation. However, in zone HALL:2 there is evidence of a change in taphonomy from salt-marsh/brackish water and marine derived pollen to probable freshwater alluvium containing pollen from different areas of the catchment. In zone HALL:2, there is evidence of sediment sources derived from the local and/or regional (Jurassic) geology; derived palynomorphs are more abundant in zone HALL:2. The reduction in oak and hazel may not, therefore, be a real decrease in the extent of such woodland at this time period (ie, pollen zone boundary).

The cause of this change from salt-marsh to rough pasture is not yet clear. It seems most likely that it is the result of some negative eustatic tendency (decline in sea-level relative to land), though much more localised factors cannot be ruled out. This would have allowed colonisation of the grassland on the subsequently drier soils and as a consequence, Iron Age settlement/activity. However, there is clear evidence for continued sedimentation, possibly caused by seasonal flooding (and overbank sediment deposition).

Some remaining halophytes (eg, Chenopodiaceae) indicate that salt-marsh remained within the near region. Alternatively, there may have been occasional marine/brackish water inundation depositing these halophytic taxa. Diatom analysis (Cameron, below) may also demonstrate this and provide a clue as to whether the halophyte pollen is of airborne derivation or fluvially transported (ie, if marine/brackish water diatom taxa are found). Such periodic inundation and sedimentation may have been responsible for abandonment of the archaeological site.

Some cereal pollen and possible associated weeds are present throughout both zones, but especially in zone HALL:2. This is clear evidence that cereal cultivation was taking place. However, values are small and it is suggested that the pollen is derived from some distance (fluvially or airborne) from suitable interfluve soils in the region.

65 Summary of environment and habitat changes i) Bronze Age/Iron Age deposition of typical grey, marine salt-marsh sediments (Wentlooge Formation) as a consequence of progressive rise in relative sea-level (positive tendency). Habitat ecology was upper salt-marsh dominated by Chenopodiaceae (eg, oraches but also transported glassworts from the lower salt-marsh?) and Plantago maritima. Dry land vegetation was apparently oak and hazel although there were perhaps the last vestiges of lime woodland (ie, just post-Late Bronze age lime decline) and possibly some secondary ash. Pine and birch are present but are probably long distance transported components. ii) Sedimentological and pollen taphonomic changes at c. 4.80 m. The previous salt-marsh changes to a grassland pasture environment dominated by grasses and ribwort plantain. Also at this horizon there is archaeological activity and habitation. This may have been a response to the availability of this new pasture/land surface. The cause of this may have been drainage by the Iron Age communities (but there is no archaeological evidence to confirm this) but more likely due to a negative eustatic tendency (ie, small lowering in relative sea-level). Unexplained at present is a peak of Hystrichospheres across this zone of change. Was this flooding? The sediment source changed from marine/salt-marsh to probably freshwater alluvial material. This deposited sediments containing geological, derived pollen, eroded and transported from the river catchment. This was possibly seasonal and as such, the question of seasonal occupation of the floodplain and grazing activity should be considered (such as by Mark Robinson in his discussion of the Thames floodplain at Farmoor, Oxfordshire). Furthermore, introduction of winter sown wheat during the Iron Age would have caused greater soil instability, soil erosion and thus sediment supply within the fluvial catchment. There is some evidence for cereal cultivation at distance on the better drained interfluves and island soils. Oak and hazel woodland probably remained with reduced pollen values due to the taphonomic changes described. Alternatively, if there was increased Iron Age activity and intensification of land use, there may also have been woodland clearance. Charred plant remains and querns found on-site and the small values of cereal pollen in section would tend to indicate that pre-processed cereals were imported for domestic use on the site. This was an apparent phenomenon of emmer and spelt cultivation during the Iron Age (producer and consumer sites) with these cereals being transported as whole ears to consumer sites. Areas of marine influence (inlets or rivers) remained within the proximity of the site: periodic inundation of nearby salt-marsh? iii) Expansion of pine, birch and hazel and Cyperaceae may indicate increasing wetness of the site; the latter indicating fen, the former from alluvial transport? There is a reduction of ribwort plantain, hoary plantain and to some extent, grasses which may also be due to waterlogging of the soils. This appears to have become mudflat, with sediments derived from fluvial sources and relating to the local channel infill; there is no indication of marine inundation.

Conclusion There is clear evidence of environmental change from salt-marsh to grassland pasture. The transition is also at the point of Iron Age activity and confirms suggestions that the occupation was sited on an alluvial plain. This may, however, have been drying out salt-marsh in transition to pasture. Whether this activity was causative or a response to change is at present enigmatic. It is realised that some of the suggestions made may be a bit tenuous and it is expected that integration with other environmental aspects in the future may produce a more definite conclusion.

66 Diatom analysis by Nigel G. Cameron

One diatom assemblage, from the grey alluvium, 184 (4.20 m OD), was counted from Hallen Marsh and was composed of approximately 85% marine species, almost 10% marine-brackish species and less than 5% brackish water species. Halophilous and freshwater diatoms comprise less than 1% of the diatom assemblage. The marine planktonic diatom Paralia sulcata comprises about 50% of the assemblage, whilst the marine tychoplanktonic diatom Cymatosira belgica accounts for over 25% of the diatom flora of this sample. This assemblage clearly represents a diatom flora deposited under estuarine conditions, perhaps with a high proportion of allochthonous, marine planktonic diatoms transported from the outer estuary. However, the local habitat is exposed to high salinity and tidal conditions. The diatom assemblage is consistent with the context (184) description for this sample which indicates a dark grey clay interpreted as part of the Wentlooge Formation.

Molluscs by Michael J. Allen

Two columns of samples were taken through the relatively short/shallow sequence at Hallen Marsh. The first, a sequence of ten samples through House 1 and the occupation deposits, containsed very few shells. The robust Hydrobia species is only present in one sample and Ovatella mytosis, which does not occur in the second sequence, is also present in one sample. This is an estuarine species, but one common under driftwood and rubbish (Macan 1977); whether this has washed in, or was living in brackish pools and rubbish associated with settlement cannot be determined. The terrestrial assemblage is sparse, but significantly richer than that from the second sequence (Table 8). The presence of Pupilla muscorum and Vallonia costata tends to indicate some drier habitats, though both are recorded in floodplain meadows (Davies 1996; 1998). The molluscan evidence does not amplify that provided by the pollen record.

A second sequence of 16 samples from the same section as the pollen monoliths (Fig. 13) and grain size analysis was through the deposits between the two structures, and encompassed pre- occupation, and immediately post-occupation deposits. Shells are absent in the lower, pre- occupation, and upper (later) profile and not abundant throughout. The terrestrial component is restricted to only two samples with Pupilla muscorum and one fragment of Cepaea sp. The aquatic assemblage is dominated by Hydrobia, especially H. ventrosa. Most occur in the immediate post- occupation deposits (layers 152 and 151) and all relate to pollen zone HALL2. This species is predominant in brackish salt-marsh creeks away from the marine mudflat edge. The lack of any other species may, however, indicate that these were washed in on tidal flooding. Numbers of shells fluctuate through the profile; whether this is a factor of preservation or of episodic inwash cannot be determined. This does, however, tend to indicate marine flooding or ingression in localised creeks and reens.

This may represent a flooding episode at soon after 390–110 cal BC resulting from rising sea-level and equating with the abandonment of the Iron Age settlements in peats marginal to marsh areas at the Mere and Glastonbury lake villages.

Table 8. Mollusc data from Hallen Marsh

Between houses In house 1 Sample pre-occupat post occupation later pre- floor post occupa later Sample 4 5 17 16 12 1 3 8 7 6 10 15 21 19 20 20 25 Sample 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1115 1116 1030 1031 1032 1033 Context/Feature 183 183 152 152 152 152 152 151 151 101 101 183 321 105 105 102 101 Depth (cm) 100- 90- 80- 80- 75- 70- 60- 50- 40- 30- 20- 80- 70- 60- 50- 40- 30- 110 100 90 85 80 75 70 60 50 40 30 90 80 70 60 50 40 Wt (g) 1750 1750 1750 1550 1750 1750 1750 1750 1750 1750 1750 1750 1750 1750 1750 1750 1750 Pupilla muscorum (Linnaeus) 1 - - - - 3 - - - - - 1 2 1 - - 3 Vallonia costata (Müller) ------1 - - - - Cepaea spp. - - - + ------Land mollusc Taxa 1 0 0 0 0 1 0 0 0 0 0 1 2 1 0 0 1 LAND MOLLUSC TOTAL 1 0 0 0 0 3 0 0 0 0 0 1 3 1 0 0 3 Hydrobia ventrosa (Montagu) 1 1 5 4 61 13 58 19 57 18 - - 2 - - - - Hydrobia ulvae (Pennant) ------1 4 ------Hydrobia spp. ------1 ------Ovatella myosotis (Draparnaud) ------2 - - Aquatic Taxa 1 1 1 1 1 1 2 2 1 1 1 0 1 0 1 0 0 FRESHWATER TOTAL 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 MARINE TOTAL 1 1 5 4 61 3 59 23 1 18 1 0 2 0 2 0 0 TOTAL 2 1 5 4 61 16 49 23 57 18 1 1 5 1 2 0 3 Magnetic Susceptibility

(note: only samples with shells tabulated: totals exclude C. acicula)

Charcoals by Rowena Gale

Charcoal was sparsely present in Iron Age contexts at Hallen Marsh. Two samples collected during the excavation in 1992 were examined and identified to genus (Table 9). Samples were from an oven and floor within the occupation area.

A few thin flakes of oak (Quercus sp.) charcoal were recovered from the floor occupation area of House 1. The charcoal probably contained sapwood although it was too fragmented to be certain. Charcoal fragments from an irregular patch north of the oven were exceedingly small and difficult to examine. Oak formed the major component, with variable growth rates from average to slow. Extraneous materials deposited throughout the wood cells prevented an assessment of maturity.

Table 9. Charcoal from the roundhouse at Hallen Marsh

Sample Context Description Quercus Other 1086 195 Floor occupation area 19 - 1089 311 Irregular patch north of oven 9 1 (see text) The number of fragments identified is indicated.

Both samples were recovered from the House 1 and, by inference, the charcoal represents fuel debris from a hearth within the round-house, and indicates the domestic use of oak. The settlement at Hallen suggests that, during the Midddle–Late Iron Age, local woodland resources were sufficiently plentiful to provision the building of round-houses and other structures. Although speculative, these would almost certainly have incorporated oak timbers.

Evidence from mid-Flandrian levels at sites somewhat closer to the estuary than Hallen Marsh demonstrates that dense woodland once flourished on the peat-levels bordering the coastline (Allen 1992a; see also Bell et al. 2000). At Grange Pill, Hills Flats, Oldbury Flats, Gold Cliff, and other sites their remains are now exposed daily, lying in situ where they fell on the present-day foreshore of the Severn Estuary. The well preserved slender trunks, crowns and root-systems estimated to be

68 roughly 50 years or so in age before succumbing to windblow, suggest successional woodlands of relatively young trees. At this time the marginal wetlands and moist soils were probably dominated by oak, alder (Alnus glutinosa), birch (Betula sp.), and willow/poplar (Salix sp./Populus sp.).

Evidence from Late Bronze Age/ Early Iron Age charcoal deposits from Avonmouth at Bristol Sewage Works (Gale unpubl.) and Kites Corner (Walker et al. 1999a), and pollen from Cabot Park (Walker et al. 1999b) and Rockingham Farm (Walker et al. 1998b), attest to local woodland composed of oak, ash (Fraxinus excelsior), alder, yew (Taxus), holly (Ilex aquifolium), birch, the hawthorn/Sorbus group (Pomoideae), maple (Acer campestre), elm (Ulmus sp.), hazel (Corylus avellana), willow/poplar, and blackthorn (Prunus spinosa).

The Roman settlement at Rumney Great Wharf on the Welsh coastline of the estuary produced evidence of wetland reclamation and a small-scale metalworking industry (Fulford et al. 1992). Charcoal deposits were indicative of woodlands somewhat similar in character to those from the Late Bronze Age/Early Iron Age sites discussed above, with oak, alder, hazel, elm, gorse/broom (Ulex/Cytisus), spindle (Euonymus europaea), Prunus, and probably maple, member/s of the hawthorn/Sorbus group, and willow/poplar (Gale 1992).

4. Northwick

The sequence

Observations of geotechnical pits during advance works in 1990 revealed the presence of ditches containing Iron Age and Romano-British pottery (Russett 1990, 12). Evaluation test pits at these locations recorded a Romano-British ditch (Lawler et al. 1992, 51–6) and postulated the presence of a more extensive and long-established archaeological site. As with the archaeological investigations at Hallen, large areas were stripped of topsoil and alluvium to expose the archaeological site and, consequently, an area of c. 19,325 m² spread over three fields was excavated in plan (Fig. 17). The site was preserved beneath extant ridge and furrow, so a representative portion of the section has been described, though the surface undulation thus created fluctuations in the depth of deposits.

The Romano-British site was only buried under less than 0.5 m of alluvium so the sequences exposed are relatively shallow, but they relate well to the upper sequences recorded elsewhere. Beneath the humic alluvial gley topsoil uniform brown and yellowish brown silty clay alluvium occurred to depths of c. 0.4 m. This ‘upper alluvium’ (251) was lightly gleyed and mottled and locally contained indistinct laminae (possibly flood couplets). A gleyed horizon (354) approximately 0.1 m thick and with slightly higher susceptibility levels occurred at the base of this sequence at c. 5.90 m OD and lay unconformably on the deposits into which the Romano-British features archaeological features were cut. Higher susceptibility levels probably reflect enhancement due to hydromorphism rather than pedogenesis (cf Allen & Macphail 1987). The archaeological features were cut into a dark brown–brown silty clay with clear prismatic structure typical of the ‘upper alluvium’.

Although deep Romano-British ditches (up to 1.2 m depth) were excavated, these contained sedimentary sequences largely local to the features and are not described here. A deeper sondage was, however, recorded through the sediments to 2.0 m OD during the evaluation (Lawler et al. 1992, 51–4, fig. 9). This revealed a very narrow band of humified peat at c. 4.6 m OD with very low pollen counts, beneath which were low energy estuarine alluvial deposits similar to those reported elsewhere. At depth, however, c. 2.20 m OD was lower organic horizon overlying beach

69 WA Northwick

530

Trench 5

254 327

Trench 8 Trench 27

328

0 10 m

Figure 17. Northwick; summary plan of the excavations

70 WA Northwick WA Northwick GGAT G024

6 m O.D.

0 10 20 x10-8 SI/kg

5 m O.D.

4 m O.D.

3 m O.D.

Alluvial silty clays (brownish red hue) Alluvial silty clays (colour unknown)

01 m

2 m O.D.

Figure 18. Northwick Wentlooge Formation profiles

Depth (m) OD Layer Description Comment 0-0.12 6.26-6.14 250 Dark brown (10YR 4/3) dark brown, silty clay loam Topsoil with high organic matter, medium granular to subangular blocky structure, many fine roots, clear boundary, Topsoil 0.12-0.38 6.14-5.88 251 Yellowish brown (10YR 4/5) to brown (10YR 5/3) Upper alluvium & silty clay loam, massive and structureless, but locally ridge and furrow both weak medium blocky structure and very faint laminae are present, and common diffuse ferruginous mottling, rare small local stones, rare inclusion of weathered and worn pottery, bone and charcoal with clear to gradual wavy boundary 0.38-0.46 5.88-5.80 354 Grey (5Y 5/1) to brown (10YR 5/3) compact gleyed Upper alluvium silty clay with coarse angular blocky to prismatic occupation interface structure, 0.5% fine macropores, common to many diffuse ferruginous yellowish brown (10YR 5/6) mottles and common fine to very fine manganese flecking, clear smooth boundary 0.46+ 5.80+ 316 Dark brown (7.5YR 4/4) to brown (10YR 5/3), coarse Upper alluvium prismatic silty clay with common greyish hues and grey (5Y 5/1) interped surfaces due to gleying, and rare fine manganese flecking sands and gravels. The relatively shallow sequence replicates the upper profiles at Awkley Lane, Vimpenny’s Lane, and Hallen Marsh so sampling was analysis was restricted to land snails, charcoals, and charred plants, largely from archaeological features.

Molluscs by Michael J Allen

The sedimentary sequence was not sampled for snails at Northwick. Small samples were processed but were devoid of shells. Instead two deep, alluvial filled Romano-British ditches associated with field boundaries and land tenure were sampled.

Ditch 237 A sequence of seven samples was taken through the shallow ditch (0.6 m deep) and into the natural alluvium through which it was cut. Although shells only survived in the upper profile, Romano- British pottery is still prevalent in these contexts, but this occur above a significant peak in magnetic susceptibility indicating deposits directly contemporary with human activity. Shell numbers are typically low (Table 10) in the three samples and only one aquatic species (Ovatella mytosis) is present which is more terrestrial than aquatic in its habitat (Kerney 1999, 46). The remaining impoverished terrestrial assemblage includes Pupilla muscorum, Cepaea spp. and rupestral Clausiliidae. It seems likely that the rupestral species represent damp habitats, possibly even wet wood and leaf mould collected in the ditch and this is the niche exploited by O. mytosis. Certainly C. bidentata is also tolerant to human disturbance.

72 Table 10. Mollusc data from Northwick

Feature ditch 254 dicth 327 TEMP Sample 12 13 10 11 1038 1039 1040 Sample 1008 1009 10101011 1038 1039 1040 Context 265 265 257 257 333 835 325 Depth (cm) 110- 110- 90- 80- 70- 60- 50- 120 110 100 90 80 70 60 Wt (g) 1400 1500 16001425 1750 1750 1750 Pupilla muscorum (Linnaeus) - - - - 2 13 4 Cochlodina laminata (Montagu) ------5 Clausilia bidentata (Ström) - - 1 - 1 - 7 Helicella itala (Linnaeus) - - 1 - - - - Trichia hispida (Linnaeus) - - 1 - - - - Cepaea spp. - - + + + 2 2 Land mollusc Taxa 0 0 4 0 2 2 4 LAND MOLLUSC TOTAL 0 0 4 0 3 15 18 Bithynia tentaculata (Linnaeus) - 1 - - - - - + Lymnaea truncatula (Müller) - - 1 - - - - Hydrobia ventrosa (Montagu) 9 - 1 - - - - Hydrobia ulvae (Pennant) 1 16 - - - - - Hydrobia spp. - 17 - - - - - Ovatella myosotis (Draparnaud) - - - - - 3 - Aquatic Taxa 2 2 2 0 0 1 0 FRESHWATER TOTAL 0 1 1 0 0 0 0 MARINE TOTAL 10 33 1 0 0 3 0 TOTAL 10 34 6 0 3 18 18

note: only samples with shells tabulated: totals exclude C. acicula

Ditch 254 In contrast, ditch was 1.5 m deep and 17 samples were taken in columns through the ditch deposits and into the natural. Shells occur in low numbers in four samples (Table 10). These are at the top of the primary fill (layer 265), and within the main ‘occupation’ related deposit; ie, that containing most artefacts. The assemblages are also distinctly different from the shallow ditch 327. The primary fill contains only aquatic species; the majority of which are brackish species (Ventrosa sp.) with only one specimen of a freshwater taxa (Bithynia tentaculata). The presence of H. ulvae as the dominat Hydrobia species in this feature is particularly interesting as this has a much more estuarine and coastal preference than H. ventrosa. We may infer that this either represents near coastal conditions or, more likely, that these are derived from habitats and washed in by local flood events relating to rising relative sea-level (RSL) at this time. In the secondary fill, relating to ‘occupation’, however, aquatic species are minimal and a low number of terrestrial species not previously present indicate localised open drier habitats locally.

Comment There is a distinct variation between the assemblages from the two ditches; whether this is a factor of depth (4.90 m OD vs 5.40 m OD), or time cannot be determined. As occupation and the ditches seem, however, to cover a relatively short time span in the first century AD and probably not extending much beyond a century, it is most likely that the local variation in depth, and use or function of the ditches, rather than differences in time, account for the variation between the two groups of assemblages.

73 5. Awkley Interface

The sequence

Four large machine cut soil pits (and one evaluation pit) were excavated along a transect of c. 125 m on the footslope of Awkley Hill (ST 895 863) to illucidate the relationship between estuarine innundation and colluvial deposits on the north-west edge of the Avon Levels (Fig. 3). Trench 1, almost at the foot of the slope, was only 130 m east of the deeply stratified mineralogenic silts and peat sequence at Awkley Lane. The composite section (Fig. 19) can be extended into the back fen to the Awkley Lane sequence by virtue of an evaluation trench and augerholes conducted by GGAT (Lawler et al. 1992, 58–62).

The excavations revealed alluvial and colluvial deposits overlying Mercian mudstones (Kueper Marl and Tea Green Marl). Evidence of limited Romano-British activity was present near the base of the footslope. The basal deposits are highly variable forms of Kueper Marl and Tea Green Marl bedrock with many mudstone fragments (Fig. 19). Weathered Mudstone occurred beneath the Keuper Marl on the slopes of Awkley Hill itself. Deep freshwater fen peats occur below Awkley Hill (Awkley Lane) but, as the sequence and bedrock rise above c. 4.2 m OD, they are absent. Overlying the peat are sandy alluvial depoists (cf Crowther in Lawler et al. 1992, 98), with moderate magnetic susceptibility readings (c. 20 SI10-8Kg-1) with an increasingly colluvial component both upslope and up profile. The higher magnetic susceptibility readings here than in other alluvial sequences probably reflect the colluvial and terrestrial soil input. The Keuper Marl bedrock forms a bench at the base of the slope and marks the location of former Romano-British activity in the form of field boundary ditches, pottery and charcoal. This is also the location of the former Old Awkley Lane and Awkley Road.

To the east, and upslope of this point, alluvial and estuarine deposits give way to coarse colluvial soils derived from Awkley Hill. Magnetic susceptibility values generally remain similar to that of the alluvium, but peaks (up to almost 60 SI10-8Kg-1) clearly define horizons of archaeological activity

Samples were removed from trenches 1 and 4 in columns of contiguous samples at 0.1m intervals for soil magnteic susceptibility and land snail analysis. These sequences were nearer the base of Awkley Hill, with the aim of examining any alluvial-colluvial interface. Bulk samples from the archaeological features produced no charred remains.

74 Trench 1; base of slope Depth (m) approx OD Layer Description Comment 0-0.14 7.25-7.11 100 Dark greyish brown clay loam Topsoil. 0.14-0.70 7.11-6.55 101 Weak red silty clay. One small sherd of Romano-British Alluvial /colluviual pottery present 0.70-1.60 6.55-5.65 102 Grey silty clay, a single limestone slab recorded in layer. [GGAT E] Layer sits in what appears to be a slight hollow 1.20-1.35 6.05-5.90 118 Light grey silty clay, occasional flecks of brown sand Alluvial [GGAT D1]. 1.35-1.60 103 Brown sandy clay, common mineral staining. Alluvial [GGAT C1] 1.60-1.70 104 Grey/pinkish grey mottled sandy clay, very occasional ?occupation charcoal flecks and small pebbles [GGAT C1] 1.70-1.75 105 Light grey sandy clay Alluvial [GGAT C1] 1.75-1.85 106 Pinkish grey/brown mottled sandy loam Alluvial [GGAT C1] 1.85-2.00 107 Red silty clay, thin striations of pinkish grey clay throughout Alluvial layer [GGAT C1] 2.00-2.05 108 Greyish brown silty clay Alluvial [GGAT C1] 2.05-2.65 109 Dark reddish brown sandy clay, occasional lenses of orange Kueper Marl brown sand [GGAT A] 2.65-2.70 110 Weak red silty clay, layer is involuted and is not present Keuper Marl across whole trench [GGAT A] 2.70-2.90 111 Light greenish grey silty clay with lenses of weak red clay, Tea Green/Keuper layer is involuted and is not present across whole trench Marl [GGAT A] 2.75-2.80 112 Weak red silty clay, thin lense of material only present in Keuper Marl part of trench [GGAT A] 2.80-2.82 113 Light greenish grey gleyed silty clay, thin lense of material Tea Green Marl only present in part of trench [GGAT A] 2.90-3.20 114 Reddish brown (5YR 3/4)/weak red sandy clay mottled Keuper Marl material, layer is involuted [GGAT A] 3.20-3.25 115 Light greenish grey (5G 6/1) gleyed silty sand, layer is Tea Green Marl slightly involuted [GGAT A] 3.25+ 116 Weak red silty clay Keuper Marl [GGAT A]

Trench 3 Depth (m) OD Layer Description Comment 0-0.10 300 Very dark greyish brown loam, occasional limestone, chalk Topsoil. and charcoal flecks. 0.10-0.25 301 Reddish brown sandy loam, occasional charcoal flecks. 0.25-0.60 302 Reddish brown sandy loam, veins of crushed limestone and colluvial some mineral staining. 0.60-0.80 303 Brown silty loam, occasional charcoal flecks, only appears buried soil. in N-W end of trench. 0.60-1.10 304 Weak red silty clay loam, lenses of pink coarse silty loam. colluvial 0.75-1.10 305 Dark red silty clay loam, occasional lenses of pink sand. colluvial Layer only occurs occasionally through section. F312 Small cut 0.35m deep only seen in section, filled with ?post-hole. reddish brown silty clay loam with occasional limestone and mineral staining 1.10-1.20 306 Reddish brown silty clay loam, lenses of sandy material and colluvial/weathered occasional mineral staining. Keuper Marl. 1.20-1.40 308 Contorted reddish brown silty clay loam, thick veins of Keuper Marl. greenish grey material and veins of crushed limestone run diagonally through layer into 309 below. 1.40-2.10 309 Contorted reddish brown silty clay loam, common veins of Keuper Marl greenish grey material and crushed limestone running diagonally through layer. 2.10-2.35 310 Reddish brown/dark red coarse dandy loam with up to 50% Keuper Marl & mudstone. Mudstone. 2.35+ 311 Dark red sandy clay loam with up to 70% mudstone. Mudstome

75 Trench 4 Depth (m) OD Layer Description Comment 0-0.20 8.18 409 Very dark greyish brown (10YR 3/2) loam, occasional Topsoil manganese frags and crushed limestone pieces. Diffuse wavy boundary 0.20-0.50 410 Brown (5YR 5/2) fine sandy loam, occasional manganese frags and crushed limestone pieces. Gradual boundary 0.50-0.80 411 Reddish brown (5YR 4/3) fine sandy loam, very occasional Largely colluvial manganese frags and crushed limestone and small stones. [GGAT E] 0.80-0.95 400/ Dark reddish grey (10YR 4/2)/brown (7.5YR 4/4) silty loam Occupation layer? 412 to to silty clay loam, common medium limestone pieces, 13 = GGAT 396 [D1] sherds of Romano-British pottery recovered. Gradual to diffuse boundary 0.95-1.20 413/ Reddish brown (10YR 4/3 - 10YR 5/3) coarse sandy clay Occupation deposit ? 423 loam. Occasional manganese staining rare limestone pieces [GGAT C1] and occasional charcoal and fired clay, with occasional lenses of olive green – pale yellow (2.5YR 7/4) towards at base. Sandier than 412 above 1.20-1.30 422 Reddish brown (5YR 5/3) fine sandy loam, occasional [GGAT C1] mineral staining, mottled with pink and yellow sand. F405 Irregular cut feature up to 0.75m deep only seen in section. ?Ditch Probbaly a ditch, filled with reddish brown silty and sandy loams over a dark reddish brown silty clay loam with olive brown lenses. F428 Irregular cut feature up to 0.70m deep only seen in section. ?Ditch Probably a ditch, filled with yellowish red coarse silty loam with occasional charcoal and cut by F405. 1.30-1.35 407/ Brown silty loam with frequent limestone and charcoal Occupation layer 425 F401 Broad shallow 0.50m deep ditch only seen in section, filled ?Ditch with dark reddish brown silty clay loam with occasional charcoal and limestone fragments, cut by F428 F402 Irregular cut up to 0.55m deep only seen in section, filled ?Pit with dark brown silty clay loam with occasional charcoal 1.30-1.50 415 Light brown (5YR 6/4) loamy sand, mottled with reddish alluvial yellow (7.5YR 6/4) sand, occasional mineral staining and =GGAT 401. charcoal. Gradual boundary [GGAT C1] 1.50-1.70 416 Reddish brown (2.5YR 4/4) sandy clay loam, very alluvial occasional small stones and mineral staining, distinguished = GGAT 403 by presence of manganese and crushed limestone fragments. [GGAT C1] Gradual boundary 1.70-2.00 426 Reddish brown silty clay with veins of grey green clay alluvial = GGAT 404 [GGAT C1] 2.00-2.10 417 Red (2.5YR 4/6) fine silty clay loam with veins of grey alluvial green clay and lenses of pinker material. = GGAT 404 2.10-3.00 418 Reddish brown (2.5YR 4/4) /dark red (2.5YR 3/6) coarse Mudstone, sandy loam: mixed interleaved layers of mudstone and Keuper Marl [GGAT coarse loamy sand. A] 3.00+ 419 Greenish grey (5GY 6/1) sandy clay. Tea Green Marl [GGAT A].

76 WA Awkley Lane WA Awkley Interface NW SE Awkley Hill 12 m O.D. Occupation Zone 11 m O.D.

10 m O.D. Tr.2 Old 9 m O.D. Awkley Lane Tr.3 Tr.4 A 8 m O.D. Awkley Lane Tr.1 C A 7 m O.D. B D 6 m O.D. C A 5 m O.D. D

A Keuper Marl Mudstone 4 m O.D. B Easturine Sediment 3 m O.D.

2 m O.D. Peat

1 m O.D.

0 100 m

Figure 19. Awkley Interface composite transect

77 Molluscs by Michael J. Allen

Shells were absent from all samples of alluvium from trench 1 and were were sparse from those in trench 4, only being present in six samples (Table 11) and were all restricted to the upper 0.8 m of the colluvial profile. The depauperate assemblages were wholly terrestrial, with only one specimen of fresh or brackish water species present, confirming the absence of developed salt-marsh and inundation in the sequence with surviving shells. The low numbers of shells make any detailed palaeo-environment comment impossible. Shells were recovered from the latest and uppermost mixed layer of the putative Romano-British occupation deposits (layer 412) and from sandy loam colluvial sealing it. The main occupation layer (413) and all deposits beneath it were devoid of shells. Most of the assemblage is typical of open long, vegetation with a dense sward proving local mesic habitats at ground level. Some species tolerate marshes or damper habitats such as Vallonia pulchella and Vitrea crystallina. One specimen in particular, however, is worthy of note; Vertigo lilljeborgi. This rare species, which has never before been recorded in southern England (Kerney 1999, 97), is tolerant of acid conditions and is restricted to ‘saturated decaying vegetation in Carex and Juncus swamps, sometimes shaded by alders’. Although it has mainly been found on the margins of highland lakes, it has occasionally been found in small isolated mires.

The limited assemblages here may indicate essentially terrestrial land on the margins of damper land with mires and possibly salt-marsh in Roman and post-Roman times.

Table 11. Mollusc data from Awkley Interface, trench 4

Sample 1035 1036 1037 1038 1039 1040 1041 1042 1043 Context 412 411 411 411 411 410 410 409 409 Depth (cm) 75- 70- 60- 50- 40- 30- 20- 10- 0- 80 75 70 60 50 40 30 20 10 Wt (g) 1750 1750 1750 1750 1750 1750 1750 1750 1750 Vertigo lilljeborgi (Westerlund) - 1 ------Vallonia costata (Müller) - - - - 3 2 2 - - Vallonia pulchella (Müller) no shells cf 1 - - 1 - - - - - Vallonia spp. below 80cm 1 ------Discus rotundatus (Müller) ------+ - - Vitrea crystallina (Müller) ------1 - - Aegopinella nitidula (Draparnaud) ------+ - - Limacidae - - - 4 - - - - - Cecilioides acicula (Müller) ------2 - - Trichia hispida (Linnaeus) ------+ - - Land mollusc Taxa 2 1 0 2 1 1 2 0 0 LAND MOLLUSC TOTAL 2 1 0 5 3 2 3 0 0 Bithynia tentaculata (Linnaeus) - - - - - 1 - - - Aquatic Taxa 0 0 0 0 0 1 0 0 0 FRESHWATER TOTAL 0 0 0 0 0 0 0 0 0 MARINE TOTAL 0 0 0 0 0 1 0 0 0 TOTAL 2 1 0 5 3 3 3 0 0

(note: only samples with shells tabulated: totals exclude C. acicula)

Comment

Most of the Holocene sedimentary sequence exposed on the slopes of Awkley Hill is Romano- British or later in date. Only at the flootslope do these layers seal colluvial or alluvial deposits which remain undated. Despite sampling these deposits, few microfossils were found and no real evaluation of the environment of the colluvial–alluvial interface could be determined, except that marsh (snails) existed locally.

78 DISCUSSION and REVIEW The physical and vegetational evolution of the Avon Levels

Stratigraphic overview

The Wentlooge Formation

The Wentlooge sequence, as described by Allen and Rae (1987), provides a basis for the interpretations presented here. Detailed analyses at a number of sites has generated local palaeo- environmental interpretations which, in conjunction with the large-scale auger survey of the superficial deposits, provide the opportunity to extend these interpretations laterally. In so doing an attempt is made to describe the development of the Avon Levels as a whole.

The excavations described have provided detailed local palaeo-environmental reconstructions. In general terms some consistency in interpretations can be seen within the major facies of the Wentlooge Formation between sites, although some local variation is clearly evident. Because of the nature of both of the two deeper excavations (limited to 5 m depth and exposing the middle and upper Wentlooge Formations), the archaeological excavations and the data from the auger survey (the latter limited to 2 m depth), only the upper facies of the Wentlooge Formation were examined (ie, middle and upper Wentlooge). Nevertheless, this information includes geomorphological, sedimentological, biological, and archaeological elements with which to examine the Late Holocene evolution and use of the Avon Levels and to review and re-examine the nature of the Wentlooge Formation and the development of the Severn Estuary Levels. When the palaeo- environmental data is combined with the auger survey a much more comprehensive picture of the entire Avon Levels can be suggested.

The lower Wentlooge Formation (Mesolithic) Deposits relating to this facies were at great depth and were only encountered as basal deposits (Awkley Lane), or in exploratory auger holes through the base of deep excavations (Vimpenny’s Lane, trench 1). These facies do, however, occur more accessibly as peat ledges (Allen 1990c) along the coastal margins of the Avon Levels (eg, Gravel Banks) and have been studied and reviewed by Druce (2001).

The middle Wentlooge Formation (later Mesolithic to Bronze Age) Interleaved deposits of peats and minerogenic clays were only truly encountered at Awkley Lane. The lower minerogenic sequence and narrow detrital peat bands at Vimpenny’s Lane also belong to the very top of the middle Wentlooge Formation. These represent freshwater fen peats and reed swamp which are inundated by minerogenic sediments in flunctating salt-marsh and mudflat conditions.

Boundary of the middle and upper Wentlooge Formations A clear boundary exists between the humic peats and lower minerogenic deposits, and the upper minerogenic deposits. This can be detected on physical criteria (characteristics such as colour and structure), and in terms of pollen zone boundaries (Fig. 22, below). The humic peats developed under carr (birch woodland) with local acid peat communities. The lower minerogenic deposits represent brackish/estuarine inundation at c. 2250 cal BC (4000 BP) in a variety of local salt-marsh and mudflat regimes. Although the upper minerogenic deposits are also silts and silty clays deposited under salt-marsh and mudflat conditions they indicate deposits which contain more fluvially derived sediments, with ferrous and manganese minerals giving rise to strong reddish and brown hues on gleying. Unlike the lower minerogenic deposits, they rarely contain any fluvial structures such as flood couplets or laminae, indicating that such features have been destroyed by 79 pedogenesis and biotic reworking. Thus, although this boundary is physically clearly defined, the change of environmental deposition between the upper and lower facies is not as clear. Its stratigraphic and chronological location is, however, extremely significant in relation to occupation of Iron Age and Romano-British date. It is discussed below as it relates to one of the main phases of late Holocene positive relative sea-level rise occurring in the later Bronze to Romano-British periods.

Upper Wentlooge Formation Heavy gleying with strong iron and manganese staining gives the upper Wentlooge Formation its distinctive reddish-brown hue, making it easily distinguishable in the field and in auger survey. It is uniformly later Iron Age and later, but its inception, although undated, is probably asynchronous across the region. Nevertheless, by the later Romano-British period, the upper Wentlooge Formation is recognised throughout the wetlands. This formation overlies later Iron Age features at Hallen Marsh, and encompasses later Romano-British ones at Northwick. Later medieval and post- medieval ridge and furrow are the latest ubiquitous topographic forms within this deposit.

Sea-level change and palaeo-geographic reconstruction

The peat and mineral sediment sequences of the lowlands surrounding the Bristol Channel contain a valuable archive of the local and regional changes in Holocene sea-level. The sediment sequences were influenced and/or accreted directly in response to changing RSL and the inter-relationships of peat, sediments, and their heights relative to present Ordnance Datum can provide data on the sea- level changes which have taken place during the Holocene period. Radiocarbon dating and microfossil analysis of the peat and microfossil analysis (especially pollen and diatoms) of the estuarine silts/clays has allowed some deductions regarding the position of the palaeo-environment and palaeo-geography.

Whilst the late-Devensian glacial and early Holocene period to c. 7000 BP globally saw sharply rising sea-levels from -120 m OD (Fairbanks 1989), the rate subsequently slowed, resulting in many and more local factors such as local neotectonics and variations in coastal dynamics (Long et al. 2000) having a greater impact on the spatial aspects and on the middle and late Holocene sea-level and shoreline position. It is, however, clear that general expansion of RSL to the present day has altered the configuration of the coastline and as such, the character and distribution of human resources along the coastal and estuarine fringe.

Reviews of sea-level change in this region have been given by Hawkins (1971; 1973) Work has concentrated on the Bristol Channel region on both the English and Welsh sides and has a long history dating back to Reid (1913, 129), Godwin (1945), and Churchill (1965). More recent studies include Balaam et al. (1987), Kidson and Heyworth (1973; 1978), and Heyworth and Kidson (1982). Published sea-level curves for the region are available and include those of Churchill (1965), Balaam et al. (1987), Heyworth and Kidson (1982), Scaife and Long (1994), Canti et al. (1993), Haslett et al. (1997), and Haslett et al. (1988). Figure 21a is a synthesis of data obtained by Kidson and Heyworth which incorporates sea-level change with datum points obtained from the Welsh side of the Second Severn Crossing at Caldicot Pill. Shennan (1989), examining neotectonics around Britain’s coastline, suggested that the Bristol Channel is/was experiencing crustal subsidence at a rate of between c. -0.2 mm to -0.5 mm OD per year; that is, substantially less than other regions of the country with postive changes (the uplifting North) or areas of greater tectonic subsidence such as the southern North Sea Basin and Thames Estuary (Devoy 1977; 1979; 1980; 1982) where net sea-level rise has been greater than the study region. For the purposes of this study, the work of Kidson and Heyworth (1978) and especially Heyworth and Kidson (1982) are most

80 relevent to the early and middle Holocene changes which resulted in the formation of the basic coastal morphology, whilst late Holocene changes have been addressed by Allen (1987a) and Allen and Rae (1988). After these sharply rising sea-levels, and sedimentation seen in the Somerset Levels, there was a retreat evidenced at many sites along the Bristol Channel (eg, at Caldicot Pill; Scaife & Long 1994; and Westward Ho! Balaam et al. 1987). This retreat resulted in the establishment of a peat forming reed-swamp and fen carr. It is from this point in time that studies here commence with the basal peat sequence dated to c. 3800 cal BC (5000 BP) at Awkley Lane.

There has, however, been some debate, with Heyworth and Kidson (1982) arguing in favour of a smooth exponential change in sea-level through the Holocene in this and other regions (Jelgersma 1961; 1966; Jardine 1976). Heyworth and Kidson (1982) suggest that after rapidly rising sea-level during the early and middle Holocene, there was a substantial slowing down between 4000 and 3000 BP, after which any increase has been minor. They suggest that any such oscillations of sea- level are of a lesser magnitude than the inherent uncertainties associated with typical studies of relative sea-level change. This view contrasts with other workers favouring the evidence for an oscillating sea-level which gave rise to alternating sequences of peat and mineral sediment as for example seen in the Lower Thames (Devoy 1979; 1980; 1982). Whilst this is, perhaps, the case when studying the typical inter-relationship between the principal peat facies intercalated within estuarine/marine sediment units here, examination of the latter in more detail suggests that some further resolution of sea-level and environmental fluctuations may indeed be represented.

Allen (1990a; 1990b), Allen and Rae (1988) and Haslett et al. (1998) argue that RSL continued to rise throughout the historic period at 0.41–0.82 mm per year after 3000 BP. Woodworth (1987) shows a continuing annual rate approaching 2 mm based on tide gauge evidence. This continuous change resulted in continued sedimentation in areas not subject to Romano-British or later reclamation or land division schemes of 1.58–1.92 mm per year maximum and 0.8–0.96 mm per year minimum. Allen and Rae (1988) also suggest rates of accumulation of 1.22 m since the Romano-British period, 1.05 m since the medieval period, and 0.21 m since 1945. Where reclamation and land division has occurred (?Nyland Hills in the Somerset Levels) an anthropogenically induced negative RSL has resulted.

It has been argued that the typical grey silts accumulating in salt-marsh and estuarine habitats may be unsuitable for establishing past sea-levels because salt-marsh halophytes with the ability to grow above the ground water table can be influenced by changes in RSL and local ground water conditions (Heyworth & Kidson 1982), which may themselves reflect very local changes in coastal morphology due to, for instance, the formation and breakdown of sand barriers, and the substantial effects of sediment compaction (Haslett et al. 1998). Thus, there are clearly problems in using sediment data as a proxy for establishing past relative sea-level. Allen (1992b), however, points to the undue emphasis placed on the intercalated peats whereas little is known about the tidally influenced marsh deposits of the Wentlooge Formation. The standard sea-level curves referenced above are based on radiocarbon dated peat/sediment contact horizons which mayor may not have had biostratigraphical/multi-proxy verification of their eustatic relevance. Within this continuously rising sea-level change curve is hidden evidence of minor RSL oscillations (Haslett et al. 1997). This is similarly in evidence in this study. Problems of sediment compaction, lack of understanding of sedimentation deposition processes in salt-marshes, and problems of obtaining absolute (radiocarbon measurements) from the predominantly minerogenic sediments negates the reconstruction of such minor fluctuations from the pollen data. This is further compounded by the fact that the Severn has the largest tidal range in Britain (Shennan 1982).

In this study, major changes brought about by positive RSL are clearly in evidence in the change from the peats of the middle Wentlooge at Awkley to the overlying upper Wentlooge sediments deposited under marine/estuarine conditions. Two sea-level data points have been obtained: at 81 Awkley Lane the determination of 4044±50 BP (NZA-12533) at a height of +3.14 m OD and at Vimpenny’s Lane a determination of 4182±55 BP (NZA-12527) at a height of +2.78 m OD. When plotted within the existing framework of Haslett et al. (1997) it can be seen that the datum points clearly fall the general curve of RSL change (Fig. 21a). When corrected to mean high water spring tide (based on data from Minehead) the former gives an OD of -2.42 m OD and the latter at -2.06 m OD. When plotted on the corrected RSL graph for Caldicot Pill (Scaife & Long 1994) on the north side of the Severn Crossing (Fig. 21b) it can be seen that the points lie above the general curve of Heyworth and Kidson. These points however, compare with data added for Caldicot Pill which are also on the high side. However, both datasets, when added to the Kidson and Heyworth curve, mirror the envelope of change. This phenomenon has also been noted by other recent workers in the region (A. Long pers. comm.).

Detailed pollen analysis of the upper Wentlooge sediments at the three sites (Vimpenny’s, Awkley, and Hallen; below) provide evidence for fluctuations between salt-marsh and mudflat conditions after the principal late Holocene marine incursion. In terms of vertical displacement this equates to some 1–2 m difference between mudflat and lower salt-marsh typically separated by a cliff (Allen 1989 suggests an alarming 5–10 m in places!). Given that human utilisation of a stabilised salt- marsh is now in evidence at Hallen, any such minor oscillations may have had a significant effect on these human activities.

Patterns and consistency; the mapped sediments

The auger survey successfully mapped a reddish brown alluvial unit over a greyish alluvial unit. These were clearly distinguishable by colour in the field. Although the survey recognised and recorded distinct peat and buried soil or stasis horizons, the survey was less able to map these as they were not extensive surfaces which could be linked and modelled (in 2001) using QuickSurf. Stabilisation surfaces and deep peats were highly localised in their occurrence as pools or shallow mires within the salt-marsh, or fringing the higher dryland at the east of the Avon Levels. The extent of deeper peats was not mapped by the auger survey which was restricted to the 2 m of the sedimentary profile. Nevertheless, peats of varying thickness (middle Wentlooge Formation) exist over much of the North Avon Levels as determined from numerous commercial bore holes (Russett 1990; Druce 2001; Allen 1990c)

Although the main area of the wetland is both low lying and relatively flat, subtle variations in the present day and prehistoric surface altitude can be seen from the records of excavated sites. Buried soils and sub-surface stabilisation horizons at depth are obviously related to this and as can be seen from Figure 20. Furthermore, some individual surfaces have a more complex buried topography than previously envisaged. The most obvious of these is at about 2 m below the present surface and fluctuating about the 4.5 m OD height. Indeed, an organic stabilisation horizon can be detected at this depth at a number of locations and is sometimes referred to as the ‘BARAS layer’ after its first record by the Bristol & Region Archaeological Services (BARAS 1998). Examination of Figure 20 shows that sections at Awkley Lane, Vimpennys Lane, Avonmouth Sewage Treatment works, Hallen Marsh, and Rockingham Farm all indcate stasis horizons at about this height. These vary in their nature and superficially may be considered to be a single unit. Radiocarbon determination on horizons at all of the noted sites, however, contradict this. Dates vary from 5880–5660 cal BC at Avonmouth to 2900–2620 cal BC at Vimpenny’s Lane, to 1190-810 cal BC and 1100-810 cal BC at Rockingham Farm and Avonmouth respectively (see Table 12 and Fig. 22). Indeed, the

82 Awkley Vimpenny Vimpenny Avonmouth Hallen Rockingham Cabot Tr.1 Tr.2 Farm Park

6 m O.D.

2192+_ 55BP _ 2778+ 55BP 2810+_ 70BP 5 m O.D. 2957+_ 55BP 3040+_ 60BP 5879+_ 70BP 6380+_ 50BP 4170+_ 70BP 4 m O.D.

4182+_ 55BP 3 m O.D. 3816+_ 50BP 3991+_ 55BP 4044+_ 60BP

4745+_ 50BP 2 m O.D.

_ 5603+ 50BP

1 m O.D.

?Stabilisation horizon Peat (wood peat vs humified peat) 0 m O.D.

Figure 20. Relative heights of soils and stasis horizons from various locations

Table 12. Stasis horizons in the Middle Wentlooge sequence in relation to their OD height and radiocarbon determination

Site layer mOD result date Cal description BC Rockingham 5.7 2850±40 BP, Beta-134902 1190–910 upper gleyed layer Farm

Later Bronze Age soils with archaeological activity Avonmouth 523 5.3-5.25 2778±55 BP, NZA-12725 1110–810 Pale blue clay, incipient 2957±55 BP, NZA-12726 1370–1010 stabilisation horizon, arch Rockingham 729 5.20-5.10 2810±70 BP, Beta-118379 1190–820 thin bands of organic material Farm 3040±60 BP, Beta-118378 1430–1120 Kites Corner 462 5.1 2610±70 BP, Beta-129554 920–520 charcoal patch archaeological activity Little Googs/ 5.1-5.2 2970±60 BP, Beta-134900 1390–1020 non humic soil Kites Corner 3350±60 BP, Beta-134901 1870–1500

BARAS layer Cabot Park 162-4 4.5 3970±60 BP, Beta-125795 2830–2290 organic clay 'BARAS' charcoal 4170±70 BP, Beta-125794 2900–2570 - activity in vicinity Seabank site ‘BARA 4.7 3930±50 BP, Wk-5804 2580–2280 S’ Avonmouth 525 4.85-4.75 5879±70 BP, NZA-12478 4930–4550 Dark bluish grey humic clay 6866±50 BP, NZA- 12495 5880–5660

Other pre-Late Bronze Age stasis horizons Awkley Lane 107 4.61-4.46 R26327/2 no result Dark grey (10YR 4/1) stonefree clay with up to 10% flecks of black material. Vimpennys Lane 207 4.20-4.05 4182±55 BP, NZA-12527 2900–2620 Very dark grey (10YR 3/1) stonefree slightly silty clay, humic/organic layer

organic horizon at Awkley Lane at c. 4.5 m OD which may appear equivalent to that at Vimpenny’s Lane, lies 1.5 m above a surface dated to 2840–2310 cal BC and is probably Roman or later in date.

Two major alluvial units were recognisable in most of the excavations described (named ‘Upper Alluvium’ and ‘Grey (lower) Alluvium’). One of the major concerns was whether this colour differentiation was a post-depositional effect of gleying and reduction creating ‘false’ stratigraphy. If this was the case then the significance of its mapping and of any lateral interpolation of palaeo- environmental information that may derive from it is greatly reduced. In order to verify the nature and significance of the two units, the general physical and biological characteristics of both were examined.

Characteristics Visually in the field the two units are differentiated by colour, rarely by structure, and by the occasional presence of weak laminae and possible flood couplets in the lower, grey alluvial unit. Both horizons were largely massive (unstructured), excepting the localised occurrence, or preservation, of laminae and flood couplets. In the field no major textural difference was observed between these two units and this was verified by grain size analysis (most variation occurs within them). The colour change is not related to depth of burial (compare Hallen Marsh and Awkley Lane or Vimpenney’s Lane), nor are the observed colours likely to be entirely a result of post- depositional redox processes that have overprinted the depositional strata. During pollen analysis, major zone boundaries were defined independent of the stratigraphy. In every case the main pollen zone boundaries lie on, or close to, this colour change. We can conclude that the observed colour

84 change may reflect some consistent change in the depositional environment. Although this change occurred at between c. 4.1 m OD (Awkley Lane and Vimpenny’s Lane), and 4.5–4.7 m OD (Hallen Marsh), and 5.5 m OD (Avonmouth), in every case they occur stratigraphically at about dated, or assumed, Iron Age–Romano-British horizons. The informal nature of the ascription of the divisions of upper, middle and lower Wentlooge Formation do not necessarily make it possible to define precise boundaries between these divisions. However, in this case the colour change might relate to the boundary between the middle/upper Wentlooge Formation which would occur at about this point in the sedimentary sequence (though it should not be taken as a definite chronostratigraphic unit). Thus, this boundary may map both a broad chronological marker and ecological change through large area of the Avon Levels. This boundary, therefore, seems to relate to the main phase of post-glacial/late Holocene positive RSL; ie, Late Bronze Age–Romano-British innundation.

Environmental change of the Avon Levels

Pollen taphonomy and interpretation

In the past, attention has tended to be been focused on the pollen analysis of peat intercalated in estuarine sediments (Allen 1992b) and the radiocarbon dating of the upper and lower transitions to silt for the purpose of establishing sea-level age/datum points. Because of the generally low APF values of the mineral sediments and the complex pollen taphonomy, the majority of studies have ignored these sediments and their palynological potential for environmental reconstruction. Allen (1992b) similarly points to the undue emphasis placed on peat whereas there is little data for the subfossil, tidally influenced marshes. However, study of these mineral sediments in the Avon Levels, and also recently from the Solent (Scaife in Tomalin et al. forthcoming) and North Kent (Scaife 2001) suggests that useful environmental data can indeed be gained. This is especially pertinent where there is a need to understand the on-site/local environment in relation to human activity. This is becoming increasingly relevant with recent studies of prehistoric activity and impact on the coastal zone.

The predominantly minerogenic character of the Upper Wentlooge sediment sequences (largely grey- estuarine silts) clearly implies that any pollen contained will have a complex taphonomy having come from a variety of different sources along with the sediments. (Allen 1990b; 1991). These sources may include the following.

Pollen fluvially transported from some distances up the river catchments entering the Avon Levels with the sediment (Grichuck 1967; Hall 1981; Scaife & Burrin 1992). Reworking of older sediments (from soils of the interfluves and/or on-site sediment) containing pollen. The more robust forms may remain in sediments for a long time. Pre-Quaternary palynomorphs are a good indication of this, although the presence of degraded/reworked Holocene Tilia is perhaps a better indication of the problem ‘Natural’ airborne pollen sources (cf. Tauber 1965; 1967; Moore et al. 1991; Scaife & Burrin 1992) Pollen from ‘on-site’ ie, autochthonous salt-marsh and fen mire vegetation Wetland taxa also fluvially transported from the river catchments including pollen evidence of carr woodland and aquatic megaphytes such as Myriophyllum and marginal aquatic fen taxa, also algal Pediastrum.

As a consequence of the taphonomic complexity, the generally small APF values and the rigorous pollen extraction procedures required, there have been few attempts to carry out full/detailed pollen analysis on the typical grey inorganic sediments. Discussion regarding the taphonomy of pollen, spores and other microfossils (diatoms, dinoflagellates, pre-Quaternary palynomorphs) in fluvially-

85 derived sediments have attempted to produced models as an aid to understanding this complexity (see for example Burrin & Scaife 1984; Brown 1985; Cundill & Whittington 1987; Scaife & Burrin 1992). In spite of these taphonomic problems and lack of research data associated with the pollen in salt-marsh sediments and alluvial deposits, the presence of pollen can provide data on the local and regional vegetation changes (Clarke & Patterson 1984; Jennings et al. 1993; Long et al. 1999). The pollen data obtained from the Vimpenny’s Lane site (and at Awkley Lane and Hallen) further demonstrate that useful information pertaining to local and regional environmental changes can be gained from areas where perhaps, more typical or satisfactory peat sequences are not available to the analyst.

Clearly, these factors have to be considered when interpreting the pollen spectra obtained from these sediments and similarly at Awkley Lane and Hallen Marsh. Because fluvial transport (freshwater, estuarine, and ?marine) may have transported pollen for substantial distances, the term ‘local pollen assemblage zone’ is not used and primary pollen zones recognised have been termed ‘pollen assemblage zones’. However, on-site (autochthonous) changes in vegetation ecology and environment of deposition are also present and discussed where relevent (local pollen assemblage sub-zones). These aspects relate principally to the changing status of the local marsh habitat. The stratigraphical change between peat formed under wet fen and fen carr conditions such as at Awkley Lane is reflected strongly in the pollen spectra first, by a marked reduction in absolute pollen frequencies in the sediments contrasting with the very high APF values of the fen peat and, secondly, there is an expansion of taxa which are typically over-represented or more abundant in fluvially derived sediments. These include: i) Pinus (pine) which is frequently over-represented in marine sediments. ii) Degraded grains of Tilia (lime) most probably derived from soils on the interfluves and eroded sediments. iii) Fern spores, especially Pteridium aquilinum (bracken), but also monolete Dryopteris type (typical ferns) and Polypodium vulgare (common polypody). The former is diagnostic (Peck 1973). iv) There are also substantial increases in pre-Quaternary palynomorphs and Dinoflagellates, the latter which may be marine/brackish water Holocene and as well as from older, derived geological microfossils.

Furthermore, variations in the pollen spectra within the minerogenic marsh sediments have become apparent from the sequential analysis of long profiles. In addition to the essential changes noted above, are variations in the autochthonous component from which data on the changing environment (and thus RSL) can be suggested. These changes relate to the accretion of sediments under salt-marsh (upper, middle, and lower) and mostly unvegetated mudflat. Such division between salt-marsh and mudflat is of importance since there may have been several metres difference between them due to typical stepped/salt-marsh cliff profiles along channels and the estuarine fringe. Allen (1989) notes that such salt-marsh cliffs are ‘strong and tall’ and in some cases up to 5–10 m in places. This will also have been a significant factor in human activity as suggested from the archeology and abandonment of the Iron Age site at Hallen.

Pollen is present in these sediments coming from the on-site (autochthonous) plant communities which may, as noted, enable differentiation between mudflat, salt-marsh and shingle communities.

Mudflat: This appears to be characterised by substantial numbers of reworked pre-Quaternary palynomorphs derived from eroded and/or reworked geological strata and/or Holocene alluvium. There are also frequently substantial numbers of Holocene spores, eg, bracken and monolete spores. All of these taxa are very resistant to decay. Also within the mudflat sediments are freshwater aquatic taxa and algal Pediastrum derived from rivers discharging into the estuaries (frequently also 86 evidenced in diatom assemblages). Numbers of Pinus grains usually increase because of its over- representation in fluvial environments (it has air sacs which give buoyancy). The dominant autochthonous pollen comes from Chenopodiaceae which produce substantial numbers of pollen grains and may come from on-site glassworts (Salicornia).

Salt-marsh: Pollen spectra show smaller percentages/APF values of all elements noted above for mudflats. In contrast, there is a better representation of salt-marsh taxa with poorer pollen production. Thus, typically Spergula type, Plumbaginaceae (Armeria and Limonium), Triglochin maritimum, Aster type (ie, Aster tripolium), Plantago maritima and pollen of halophytic grasses (eg, Spartina) with large (>45 m with thin exine and small pori, ie, not cereal).

Upper Salt-marsh, dune, and shingle habitats: These are less well represented but with occasional Hippophae rhamnoides, Calystegia sepium, and possibly Liliaceae (possibly coming from Scilla verna, S. autumnalis Hyacinthoides non-scripta). Some Chenopodiaceae may come from Chenopodium (goosefoots) and Atriplex sp. (oraches).

It should, however, be noted that many of the pollen taxa may contain more than one taxon and interpretations may be based on suites/assemblages of such types to differentiate plant communities/habitat. Many pollen taxa recorded are also of catholic distribution and thus not definable to specific habitats.

The late prehistoric vegetation of the Avon Levels

There is now a substantial corpus of palynological data from the Bristol Channel and Severn regions of North Somerset, Avon, and south Wales from which much information on the development of late prehistoric vegetation and environment has been gained. This is unusual for lowland England as a whole but, as with the Fenlands of East Anglia, is attributed to the extensive lowland fen marshes of these areas and the extensive spreads of peat and mineral sediments which have accumulated. These deposits have offered tremendous potential for pollen studies dating back to the inception of the discipline (eg, Godwin 1941; 1943). Interest in this area and again the Fenlands is also enhanced by the potential for preservation of wetland archaeology (Coles et al. 1975–8) including notably, the well documented Neolithic and Bronze Age trackways of the Somerset Levels and the Iron Age lake dwellings of Glastonbury (Bulleid & Gray 1911; 1917; Coles & Minnett 1995) and Meare (Bulleid & Gray 1948; Coles & Coles 1986). In contrast, the Avon Levels lying to the east, whilst offering potential for palaeo-environmental studies (vegetation, environment, human impact, and Holocene sea-level change) had not been examined in any detail until the Second Severn Crossing English Approaches Project necessitated archaeological and palaeoenvironmental investigations.

The regional pollen database

Whilst the Somerset Levels are some distance away from the Avon Levels study area, the wealth of pollen data and diagrams from the region as a whole offers a valuable framework within which the information gained pollen studies at Awkley Lane, Vimpenny’s Lane, and Hallen may be compared.

The substantial depths of sediment that has accreted in the low-lying areas adjacent to the Severn Estuary and Bristol Channel are clearly a response to post-(Devensian) glacial eustatic change. Increase in RSL and thus transgression from the Late Devensian was responsible for the deposition of basal marine clays (lower Wentlooge) seen in the Severn Estuary Levels. This transgression was recognised at Shapwick Heath by Godwin (1943) and in the later extensive excavations of Coles in the Somerset Levels (Coles et al. 1973). Subsequent, negative changes in RSL resulted in coastal

87 emergence, development of extensive fen carr and peat accumulation over the underlying clays, and some incipient pedogenesis, as seen at Hallen. During the Late Bronze Age–Iron Age marine inundation saw the re-incursion of marine influence, further deposition of marine clays over the fen peat, and, locally, ponding back of the perimarine zone and fluvial systems causing flooding as evidence at Glastonbury and Meare Lake dwellings. Changes in RSL are viewed as the principal regional controlling factor in the character of sediment accretion in these low lying areas adjacent to the Bristol Channel/Severn Estuary (Kidson & Heyworth 1976; Allen 1987a; Allen & Rae 1987; Hibbert 1980; Scaife & Long 1994).

The north Somerset region has been extensively studied for many years. Of particular significance was Godwin’s work (1940; 1941; 1943; 1946; 1948; 1956; 1960). This work established the relationships between the rising post-glacial sea-level, consequent environmental changes, and the importance and effect on human populations in the region. For example, Dewar and Godwin (1963) recognised two primary phases of arable and pastoral agricultural activity. Without the aid of radiocarbon dating at this time, he attributed these phases to the Middle–Late Bronze Age and secondly to the Iron Age and Romano-British periods. This primary data has been much extended by archeological excavation of the Somerset Levels trackways and associated environmental studies (plant macrofossil, pollen and diatom analyses, and radiocarbon dating (Coles et al. 1973; 1975–8; Beckett & Hibbert 1979; Hibbert 1978). With detailed and multiple (three dimensional) pollen diagrams constructed in association with the prehistoric trackways, palaeo-geographical reconstruction, and landscape history has been made possible demonstrating for example, localised Neolithic clearance on the sandy islands of the Burtle Beds (Hibbert in Coles et al. 1970; Coles et al. 1973). Typical features such as the prehistoric ‘elm decline’ (eg, Chilton track (Coles et al. 1970); Sweet Track, Factory Site (Beckett 1979); and Abbot’s Way (Beckett & Hibbert 1976)) have also been recorded and radiocarbon dated. From the now large number of studies, regional pollen assemblage zones and overall vegetational history has been established (Beckett & Hibbert 1978, 86). This has proven to be especially important providing useful comparative data for the pollen study presented here from Awkley Lane.

The area of the Avon Levels has been less well studied than the Somerset Levels in spite of the obvious potential of this low lying region adjacent to the Bristol Channel. Thus, there is little data with which to compare directly, the local pollen sequences. Where data do exist, these are preliminary investigations or assessments undertaken in response to commercial developments and have not yet been fully analysed. These Avonmouth sites include Rockingham Farm (Walker et al. 1998b), Cabot Farm (Walker et al. 1999a), and Kites Farm (Walker et al. 1999b).

Relevent data are also available from south Wales which also date back to earlier periods of pollen analysis (Godwin & Newton 1938; Godwin 1940; 1943) and associated with the foreshore archaeology and peat accumulations in south Wales in the Gwent Levels. The Welsh side of the Second Severn Crossing has also been subject to palaeo-environmental studies including diatoms (Cameron 1993). Pollen analysis (Scaife 1993; 1994), and sea-level change (Scaife & Long 1994). Clearly, comparisons and relationships can be drawn between the south Wales and Avon Levels stratigraphical and biological sequences. This has previously been attempted by Smith and Morgan (1989; and see Bell et al. 2000).

Avon Levels sites

Although only at three sites has pollen been analysed, the data obtained provide clear indications of the environment of the Avon Levels and its fringing areas and how this changed throughout the later prehistoric period from the c. 4500 cal BC (5600 BP), the late middle Holocene Atlantic period (Flandrian II), until the Romano-British period in the 1st century AD. Godwin’s discussions

88 concerning the relationships between rising post-glacial sea-levels, environmental changes, and the importance of and effect on human populations in the Somerset Levels, discussed above, similarly apply to the Avon levels and are reflected in the environmental analyses which have been undertaken at the sites of Awkley Lane, Hallen, and Vimpenny’s Lane.

Sea-level change As discussed above, the substantial depths of sediment which accreted at these sites was a response to rising post-glacial RSL, the principal controlling factor in the character of sediment accretion in these low lying areas adjacent to the whole of the Bristol Channel/Severn Estuary. It is acknowledged that, because of the sensitivity of organic sedimentation facies to forcing factors, combined with lag effects, substantial intrinsic uncertainties may be expected in the sea-level curves discussed above constructed from radiocarbon dated peats (Allen 1995, 42–3).

Because of the closeness of the Somerset and Avon Levels, it is highly likely that the same controlling factors will have been responsible for the middle and late Holocene development of the latter. The Late Bronze Age–Iron Age marine inundation of the Somerset Levels which caused deposition of marine clays over the fen peat, ponding back of the perimarine zone and fluvial systems and ultimately the flooding of the Glastonbury and Meare Lake dwellings, may also have been a major cause of human abandonment of the Iron Age site at Hallen Marsh although the on-site stratigraphy reflects the position of the site in the centre of the Avon Levels floodplain. Certainly, the transition from fen peat/poor fen to estuarine sedimentation at Awkley Lane bears close comparison with the Somerset (Shapwick Heath) model and fall within the framework of RSL datum points presented by Haslett et al. (1997) and Scaife and Long (1994; see Figs 21a and b).

The changing vegetation environment The palynological study of the three sites here embraces the late middle Holocene period (Flandrian chronozone II) the sub-boreal and sub-Atlantic period (Flandrian chronozone III); that is, the late prehistoric and historic period until the medieval period.

Pollen data presented provide evidence for the changing vegetation and environment of the better drained soils of the terrestrial zone and the on-site (autochthonous) development of fen wood, poor fen, and salt-marsh and mudflat habitats, the latter in response to fluctuations in RSL. The former demonstrates the demise of the dominant middle Holocene woodland through progressive late prehistoric woodland clearance. An attempt has been made to correlate these changing vegetation and environment of these sites/periods using this data and the available radiocarbon dates and stratigraphical architecture recorded from the three profiles examined. Local pollen assemblage zones described from the three sites have been collated (Fig. 22).

The terrestrial vegetation: Awkley Lane has the longest temporal/stratigraphical sequence being the only one which provides evidence for the middle Holocene environment (pollen zone AWK:1). and first effects of human impact on the climax woodland communities. The determination of 5603±50 BP (NZA-12774) (4530–4350 cal BC) places the basal peat at Awkley Lane (AWK:1) at the end of the Atlantic, Flandrian Chronozone II. From the pollen data this is clearly just prior to the ‘primary (Neolithic) elm decline’ (see below) which is evidenced in AWK:2. This phase represents the culmination of early Holocene vegetation migration and establishment and stability in the middle Holocene. Typically, the woodland of this period comprised dominant oak, elm, lime, ash, and hazel (Birks et al. 1975) with alder growing on-site in carr woodland (see below).

The importance of lime/linden woodland: Tilia (Tilia cordata), with the high pollen percentages and plant macrofossils (Clapham above) seen in AWK:1, demonstrates that this was locally dominant on the nearby well drained soils. This adds further to the now numerous data from southern Britain which has demonstrated that this was the dominant or at least co-dominant taxon (with oak and/or 89 elm) at this time (eg, Moore 1977; Scaife 1980; 1988; 2000; Waller 1994). After its arrival in the late Boreal (Godwin 1940; 1975a; 1975b) or early Atlantic period, Tilia cordata (lime/linden) became an important and, in some areas, the dominant woodland component of the middle Holocene (Atlantic period; Flandrian Chronozone II). This is in fact the most characteristic phenomenon of the late prehistoric woodland in many areas of southern England (Birks et al. 1975; Birks 1989; Moore 1977; Scaife 1980; Greig 1982) along with its decline at various times during the late prehistoric period, particularly during the Middle–Late Bronze Age. Whilst Tilia has been recorded many times in studies of the Somerset Levels, at Awkley, the proximity of the sample site to well drained land has recorded percentages to 33%. Given the very great under-representation of Tilia in pollen spectra (Andersen 1970; 1973), these values must be regarded as significant. This, plus the presence of the fruits, provides clear evidence for such domination on suitable soils in the west country. Lime growth was considered to be a characteristic only of the well drained soils of the interfluves and here, on the localised islands which occur on the floodplain. However, it should also be considered that some recent studies are now, indicating that far from being restricted to such ‘good’ soils, lime may have also been a constituent of poorer sandy soils (Waller 1994, 96) and even very ‘damp’ woodland on peaty substrates (Scaife in Tomalin et al. forthcoming) as shown by studies at Wootton-Quarr, Isle of Wight (Scaife 2000).

Whilst Tilia was clearly the dominant woodland element, other tree and shrub taxa were also of importance in the local environment. It is possible that oak, ash, and hazel formed parts of the dry carr woodland or woodland fringing the edge of the mire – perhaps on the thicker, down-slope soils. Ilex aquifolium (holly) is recorded at Awkley (AWK: 1) and is usually extremely poorly represented in pollen spectra and thus, as with lime, implies some importance. Alternatively and very likely, these woodland trees formed lesser elements in the dominant lime woodland during the middle Holocene (AWK:1).

The first evidence of human disturbance: It is suggested that the zone AWK:l/2 transition is the early part of the prehistoric elm decline at c. 5000 BP or, slightly later if c. 3550 cal BC (c. 4800 BP) if dates for the Somerset Levels pertain. This is confirmed by the radiocarbon date of 3630–3360 cal BC (4683±55 BP, NZA-12530) near the base of zone AWK:2. Further reduction in elm pollen at 1.86 m OD (AWK:2 sub-zone a and sub-zone b) and sharply expanding values of Poaceae (grasses) with sporadic occurrences of cereal type pollen, Plantago lanceolata (ribwort plantain) and peaks of other herbs including Asteraceae (Bidens type and Aster type), Urtica (nettle and pellitory), and Chenopodiaceae (goosefoots and oraches) are all indicative of changes in the local dry-land vegetation. This reduction in lime and elm at the zone AWK:l/2 transition is the prehistoric elm decline at c. 4400–380 cal BC (5500–5000 BP) (Smith 1970; Smith & Pilcher 1973). The increase in herbs which occurs is also strongly indicative of the first impact of human disturbance and agriculture on the environment and is attributed to the establishment of a Neolithic economy with localised cereal cultivation. Whilst the decline in elm was likely to be the result of spread of disease instigated by Neolithic opening of the forest (Girling 1988), the reduction of other areas of woodland was perhaps due to clearance for occupation and agriculture. Subsequently, oak, lime, ash, and hazel woodland remained the most woodland trees on the better drained soils. Typically, there is evidence of a secondary regeneration of elm which is only seen at Awkley (AWK:2c) which has similarly been evidenced in other areas of southern England (Scaife 1988). Here, this is also associated with a regeneration of lime and ash (both as noted, are poorly represented in pollen spectra). This occurred at c. 2900–2800 cal BC (4300–4200 BP), the late Neolithic. Whilst there have been arguments that Late Neolithic woodland expansion resulted from depopulation (Whittle 1978), there is also

90 5 A Vimpenny's Lane Awkley 0 Lane

-5

-10

-15

-20

-25 024610 8 Calender years BP (thousands)

Figure 21a. Sea-level curve using the existing framework of Haslett et al. (1997) from a synthesis of data obtained by Kidson and Heyworth

0 B Site 3 5 Site 11 Vimpenny's Lane Oscar 3 Awkley Lane 10 Context 333-340

Sea-level (metres below present sea level)35 Altitude (mOD) 109876543210 Radiocarbon years BP (thousands)

Figure 21b. Sea-level curve using the the corrected RSL graph for Caldicot Pill on the north side of the Severn Crossing

91 E 100 W 100 6 m O.D. 101

101 102/151 WA Northwick 201 183 202 Upper Wentlooge 104 151 WA Hallen 203 fl 152 l ud ate 5 m O.D. M 204 Mud flat nn 186 cha 105 rsh 183 a m 205 lt 106 Mud flat Sa 184 107 Salt marsh 206 108 lt mars h 109 207 Sa 4 m O.D. Mud flat arsh 208 m lt 209 Sa WA Hallen

110 marsh 210 t al S as Mud flat 4182+_ 55BP 211 3 m O.D. Log Sedge gr s 212 112 flat d 113 Mu 113 rr WA Vimpenny Middle Wentlooge 121 rch ca Bi

114 edge 120 s 2 m O.D. 115 fen 116 Alderd Ree 118

119 lercrd ar A WA Awkley Lane 1 m O.D.

Lower Wentlooge

0 m O.D. +ve RSL

-ve RSL 01 m =equal

Figure 22. Correlation of the sedimentary facies from the reported sites

92 substantial evidence that the woodland regeneration occurred through changing agricultural practice to woodland based pastoralism (Scaife 1980; 1988).

After the Late Neolithic regeneration phase there is progressive reduction in woodland which ultimately left oak and hazel as the principal arboreal constituents but with some remaining lime, ash, and beech. Evidence for these changes is complicated by the changing environment of deposition and thus sedimentology and pollen taphonomy. Clearly, however, the next significant event was the widespread removal of lime woodland from its areas of dominance. This was a result of increasing human pressure on available land and reflects the expansion of Bronze Age activity in the region. This decline in lime is evidenced at all three sites examined at zone boundaries AWK:4/5; VIMP:1/2 and to a lesser extent at Hallen Marsh HAL:1.

Later Holocene deforestation: the lime decline: The decline of lime from southern England was initially attributed to climatic change by Godwin (1940; 1956) and, without radiocarbon dating, it was previously assumed that the lime decline was synchronous and represented climatic worsening from the sub-Boreal to Sub-Atlantic period at c. 1000–500 BC. Turner (1962), however, demonstrated the possibility that prehistoric clearance for agriculture may have been a major factor in its decline by identifying the expansion of agricultural weeds which occurred along with, or shortly after, declining lime pollen percentages. This has subsequently been widely illustrated from sites in southern England (eg, Scaife 1980; Waton & Barber 1987; Waller 1993). Furthermore, the development of radiocarbon dating has demonstrated that the decline was not synchronous but took place from the Neolithic period onwards. Whilst the majority of dates come from the Late Bronze Age, at Awkley Lane, the late Atlantic dominance is reduced at the time of the elm decline possibly representing clearance of lime for agriculture on the locally well drained soils or even for use of its bast fibres. Neolithic decline is relatively rare but see on the Isle of Wight (Scaife 1980), Somerset Levels sites (eg, Beckett 1979; Beckett & Hibbert 1976; 1978) and London (Scaife 2000). Other important explanations for the lime decline have also been forthcoming in recent years. Waller (1993; 1994) has suggested that expanding wetland (in Sussex) may have forced lime dominated woodland farther away from basin sample sites, thus causing reduction in pollen percentages at that site.

Taphonomic causation may be foremost and relevent to the decline in Tilia at the Avon Level sites where pollen assemblage zone boundaries are associated with major changes in sediment regime and thus pollen taphonomy. However, superimposed on these taphonomic changes, lime must have been removed during the late prehistoric period; since it no longer exists and there is no regional evidence for its extensive growth during the historic period. From the discussion presented, it seems plausible that this was during the Bronze Age and associated with the overall reduction in trees seen in the three pollen profiles. During the span of pollen zone AWK:4; VIMP:1a–c and HALL:1 the low percentages of Tilia in the mineral (estuarine) sediments possibly represents growth but at distance from the sample site. However, above the critical pollen/stratigraphical zone boundaries (see Section B, above) of AWK:4/5, VIMP:1/2 and HALL:1/2, numbers are substantially reduced and it seems likely that the final removal occurred during the temporal hiatus suggested for this stratigraphical boundary.

From the evidence at Hallen we can see that this boundary occurs in the later Iron Age, with incursion and sedimentation sealing an Iron Age seasonal (summer) occupation dated to 390–110 cal BC (2195±55 BP, NZA-12727). If this boundary is consistent and contemporaneous across the region and is marked by the colour changes discussed above which has been postulated to relate to the middle/upper Wentlooge boundary, then we can examine the dated evidence at Northwick. Although no palaeo-environmental studies were undertaken on this relatively shallowly buried Romano-British site (of the 1st century AD), it is clear (see Fig. 18) that the features of this site lie 93 above the blue (lower) alluvium of the middle Wentlooge and are wholly within the upper alluvium (upper Wentlooge). This would enable us to postulate this boundary occurring at the beginning of the 1st millennium AD.

Awkley provides the best evidence of terrestrial vegetation changes being in closer proximity than Hallen and Vimpenny’s Lane. In zone AWK:3 there is evidence of soil deterioration perhaps caused by such woodland clearance and subsequent acidification/podzolisation of the sandy soils of the islands. This is shown by the expansion of acidophilous communities (Erica, Calluna, and Sphagnum: see below) and the possible presence of charcoal in the macrofossil samples during the period c. 2800–2400 cal BC (c. 4300–3900 BP) (based on the dating of the organic sediments at Awkley Lane zone AWK:3; see Table 1 and Fig. 5). Similarly at Vimpenny’s Lane there is an expansion of such acidophiles although here, their expansion is associated with the change from organic to mineral sediments suggesting that pollen was derived fluvially from some distance.

Overall, it appears that significant environmental changes were taking place during the late prehistoric period brought about by a positive RSL tendency and the increasing intensity of human activity throughout the Bronze Age and into the Iron Age. Thus, whilst the sedimentological and taphonomic changes caused by marine inundation of the fen community at Vimpenny’s Lane and Awkley Lane at c. 2550 cal BC (4000 BP) resulted in the reduction of some elements such as Tilia (see above), it may be argued that the extension of the pollen catchment through fluvial transport (rather than largely airborne) might provide a more realistic picture of vegetation in the wider geographical area; that is, assuming that pollen had become incorporated into sediments in the estuary directly or from rivers exiting into the estuary.

From this period (ie, post-lime decline), there is consistent evidence of woodland dominated by Quercus (oak) with Corylus avellana type with some lesser representation of other and less well represented pollen taxa including Tilia, Fraxinus, Fagus, Populus, Ilex aquifolium, Taxus baccata. Such importance of oak and hazel woodland during the later prehistoric period until present is characteristic of southern Britain and is seen in the majority of later pollen sequences cited from this region. The importance of oak and hazel pollen during the later/historic period results from the high pollen productivity of these taxa and the fact that they were maintained in managed (coppice with standards) woodland. Populus and the more common taxa (Betula and Pinus) which are likely to be from airborne transport represent growth within the region as a whole, and from longer distances. Remaining Tilia as well as Ilex aquifolium and Fraxinus and Fagus sylvatica are taxa which are generally substantially under-represented in pollen spectra and, if not fluvially transported from elsewhere in the catchment, probably indicate local growth. Associated with the demise of woodland is the continued expansion of agriculture giving increased herbaceous diversity and pollen percentages from pasture and arable habitats (ruderal/segetals). These are superimposed on the pollen from the autochthonous communities.

The changing wetland/floodplain/estuarine habitat

Increasing RSL during the early Holocene and the early middle Holocene periods was responsible for the deposition of the lower Wentlooge Series, the basal estuarine and marine clays seen of the Somerset Levels and Avon Levels. This initial transgression was recognised in the Somerset Levels at Shapwick Heath by Godwin (1943) and later by Coles et al. (1973). Subsequent negative tendancy in RSL resulted in coastal emergence and development of extensive fen carr and peat accumulation (middle Wentlooge) over these underlying clays. The middle Wentlooge peats have been analysed at Awkley where they formed under a fen carr woodland community along the fringes of a more extensive reed swamp fen evidenced in the lower (analysed) levels at Vimpenny’s Lane. The former

94 dating to 4530–4350 cal BC (5603±50 BP, NZA-12774) to 2840–2310 cal BC (3991±50 BP, NZA- 12754) and the latter from 3350–2900 (4420±90 BP, GU-3121) to 2900–2620 (4182±55BP, NZA- 12527).

This lowest peat (AWK:1) is only seen in this study at Awkley Lane where it accumulated under a typical alder fen carr woodland (Alnetum glutinosae) from c.5600–c. 5000 BP. The best environmental data for this community comes from plant macrofossil analysis. Numerous macro- remains and large pollen frequencies of Alnus glutinosa were recovered along with evidence of a typical fen carr ground-flora (Chrysosplenium alternifolium, Lycopus europaeus, Eupatorium cannabinum). One of the characteristic species of this type of woodland is the alternate-leaved golden saxifrage which was found in some abundance in samples (Awkley) at 1.44 m OD and 1.48 m OD. Typical of this habitat were localised areas of wetter habitats/pools including for example, Hippuris vulgaris, Alisma plantago-aquatica, and Cladium mariscus and the eggs of water-fleas. There were also drier areas of fen carr with a ground flora of ferns (eg, Dryopteris felix femmina and D. felix mas-female and male fern), sedges such as Carex paniculata (tussock sedge), Urtica dioica (stinging nettle), Solanum dulcamara (woody nightshade), and Rubus sp. Taxus baccata, although only a single (pollen) record, was typically a tree of such dry fen carr woodland (Godwin 1975a; Scaife in Sidell et al. 2000).

In stratigraphical contexts 114, 120, 115 (zone AWK:2.b at c. 3500 cal BC (4700BP)), there is evidence of an ephemeral phase of increasing wetness as indicated by increasing pollen frequencies of reed-swamp taxa including Cyperaceae (sedges), Typha latifolia and Typha angustifolia/ Sparganium type (bulrushes and burreed), and aquatic megaphytes such as Potamogeton (pond weed), Callitriche (water starwort), and Lemna (duckweed). Plant macrofossils similarly demonstrate an expansion of a wet fen with Phragmites australis reedswamp with for example Mentha aquatica, Thalictrum flavum, Apium nodiflorum, Alisma lanceolatum, and Ceratophyllum demersum (see Table 4).

In contrast, Alnus (alder) displays a sharp reduction in sub zone AWK:2b in response to this increasing wetness. It is possible that local woodland clearance (charcoal was present) on the nearby interfluve may have caused a reduction in evapotranspiration, an increase in surface run-off and a resultant higher ground water table. Alternatively, a fluctuation in ground water table through more regional change in RSL may have been responsible. The fragile balance of alder carr and water table along the fen margins was clearly disturbed resulting in the demise of the carr woodland. Typically, alder will tolerate flooding of the peat surface around the root boles for some 3-4 months of the year during winter (Tansley 1949; McVean 1953; 1956). With the exception of a single peak of Alnus at 2.38 m OD, values of alder were and remained much reduced from its earlier dominance (seen in zone AWK:1; context 118).

Fluctuation of the reed-swamp taxa as indicated in the pollen record (see for example the Cyperaceae and Typha latifolia in Fig. 7) indicates a rather unstable, variable wetland habitat. The small increase in Salix (willow) pollen may be also significant since willows are largely under-represented in pollen spectra and relate to fringing willow carr. Because of the very substantial numbers monolete Dryopteris spores (typical ferns) it is also likely that these were ferns growing within the wet, fen carr marsh habitat-although subsequently the diagnostic marsh fern (Thelypteris palustris) is well represented.

From c. 3550 cal BC (4800 BP) the peat at Awkley (context 121; zone AWK:3) contains substantial fragments of wood dated to 3090–2700 cal BC (NZA-12528, 4286±55 BP). This sample was the middle sample from the uppermost peat horizon. Pollen in zone AWK:3 exhibits a very marked change in the character of local woodland with Betula (birch) becoming dominant while Alnus remains consistent at levels seen in earlier pollen sub zones. Macrofossil evidence (wood, bark, 95 fruits, bud-scale) also attests to the development of birch carr woodland. This is comparable with similar birch dominance in some phases of the Somerset Levels (Beckett 1979). This clearly represents a major environmental change. Apart from the birch plant remains few other macrofossils were present except for substantial quantities of Sphagnum. This is similarly attested in the pollen analysis with peaks of Sphagnum spores and Ericales.

At Awkley Lane, there was a developing hydrosere where reed-swamp/tall-herb fen raised the surface above the watertable, via the accumulation of peat leading to the development of alder carr (AWK:1 and AWK:2) and birch carr (AWK:3). At Awkley, however, removal of trees from the adjacent sand island may also have caused local acidification of the soils (with heath communities) and increased the acidity of the mire (the birch carr of AWK:3) giving local Sphagnum bog. A thin peat facies at Vimpenny’s Lane allowed radiocarbon dating allowing correlation with the primary sequence examined at Awkley. Here, at 2900–2620 cal BC (4182±55 BP, NZA-12527 at 3.24 m OD), no evidence of similar birch (or alder) carr was found. Although there is evidence of acidophilous communities, these occur in the overlying mineral sediments and as such may reflect fluvial transport from other sites such as Awkley at some distance. This suggests that vegetation development at Awkley represents the areas fringing the floodplain and adjacent to higher ground/terrestrial environment. Vimpenny’s is more clearly situated in the outer floodplain where an overall wetter habitat was present supporting a wet, poor fen.

Summary

Overall, therefore, there is evidence of change in habitat types through the Awkley (middle Wentlooge) peat sequence and evidence of regional variation as seen at Vimpenny’s lane. At Awkley, the lower peat horizon was dominated by an alder carr woodland with an undergrowth of wet/damp loving species. The middle peat horizon, (contexts 115, 120, and 114) shows a tall herb fen/reedswamp was the dominant vegetation with very little or no indication of woodland vegetation being present within the local mire habitat. This was then followed by a change in vegetation to one dominated by Sphagnum moss with strong evidence of birch growing on and near the site. The provision of radiocarbon dates makes it is possible to give a rough idea of the time span of these developments at Awkley Lane. The date for the bottom of the sequence is 4530–4350 cal BC (NZA- 12774, 5603±50 BP), the date for detrital peat (context 120) from the middle peat horizon is 3640– 3380 cal BC (NZA-12529, 4745±50 BP), which suggests that the change from alder woodland to reed-swamp/tall herb fen is approximately 900 years. The date for the topmost sample (context 121) of peat is 3090–2700 cal BC (NZA-12528, 4286±55 BP) at 2.42 m OD which gives the time span for the change from reedswamp/tall herb fen as approximately 500 years.

Caution must, however, be used when expressing the time span between successions in this way. It should be considered that the peat bands are not contiguous and are separated by thin bands of alluvium and, therefore, it can be very difficult to give a more accurate time span for these vegetation changes to occur. However, in general terms it can be said that a period of approximately 1500 years saw a vegetation change from an alder woodland to one dominated by Sphagnum moss at Awkley whereas, further out in the floodplain at Vimpenny's Lane (context 211), such birch carr woodland was not present and instead, a poor fen (reed swamp) dominated by Phragmites australis was present (also confirmed by evidence of the insect remains). This fen was perhaps periodically inundated by brackish water (see also diatom evidence). These data may be comparable with the Somerset levels (Godwin 1975a, 31) where the development of ombrotrophic mire was often preceded by a stages of Phragmites and Cladium fen and fen carr woodland. At Meare Heath (Beckett and Hibbert 1979; Hibbert 1980) for example, a raised Sphagnum ombrogenous mire had developed and was inundated by estuarine conditions at c. 910–760 cal BC (SRR-914, 2624±45 BP).

96 The mineral, estuarine and freshwater sediments

Radiocarbon dates of 2900–2620 cal BC (4182±5 5BP, NZA-12527 @ 3.14 m OD) at Vimpenny's Lane and 2840–2310 cal BC (3991±50 BP, NZA-12754 @2.92 m OD) at Awkley Lane represent the uppermost (and thus datable) peat at these sites prior to estuarine sedimentation. These dates represent an earliest commencement (terminus post-quem) for this major environmental change/phase since it is highly likely that some of the uppermost organics will have been eroded. Clearly, however, this marked stratigraphical change, after which there are only ephemeral phases of organic accumulation during the major phase of late Holocene sedimentation – the Upper Wentlooge Formation. Allen (1990b; 1992b) has noted the little understood nature of the tidally influenced marshes preserved within the Wentlooge formation, and, whilst it appears unlikely that the sediments will provide a useful record of dating and relative sea-level change (see above) there is clear evidence of the changing environment of the salt-marsh as evidenced by the changing palynological assemblages.

The lowest levels analysed at Vimpenny’s correlates temporally with the uppermost peat/sediment contact at Awkley Lane at c. 2550 cal BC (4000 BP). At Vimpenny’s Lane, the position of this site further out on the floodplain/marsh suggests that there will be less of the fen and carr peat which in general thickens towards the terrestrial zone (Allen 1992b). This appears to be the case although the full stratigraphical extent/depth was not examined. At Vimpenny’s the lowest analysed sequence at 2.90 m OD (context 212) shows strong evidence of halophytic vegetation and reworked palynomorphs which suggest a salt-marsh or possibly mudflat environment. The organic facies of context 211 (zone VIMP 1a) has been dated at 2900-2620 cal BC (4182±55 BP, NZA-12527) appears represent a phase of negative sea-level tendency which allowed the development of reedswamp dominated by Phragmites australis and other rooting, poor fen taxa. However, it appears that this habitat was periodically inundated since salt-marsh taxa remain important in the pollen spectra. This was, however, an ephemeral phase of reed-swamp which culminated with further deposition of grey estuarine/marine clays and silts (upper Wentlooge). This was caused by a positive sea-level tendency from c. 2550 cal BC (4000 BP) and correlates with a similar event at Awkley Lane (AWK:4 context 110). At Vimpenny’s Lane, this initially formed salt-marsh (context 210) and then mudflat (209) (VIMP:1b). These arguments are based on specific characteristics and fluctuations of the palynological assemblages as described in above; that is, the prevalence of pre-Quaternary palynomorphs, Holocene spores derived from fluvial sources and deposited with sedimentary fines on mudflats and presence and/or absence of autochthonous halophytic taxa from the more stable salt- marsh habitat.

At Awkley Lane, the upper Wentlooge Formation is similarly characterised by estuarine silts resting on the middle Wentlooge peat (zones AWK:1–3). The upper Wentlooge transgressive event as with Vimpennys occurred at c. 2550 cal BC (4000 BP). Palynologically these events seem to suggest that at Awkley, the minerogenic sediments accreted in a mudflat environment rather than salt-marsh. The sediment source was, derived largely from rivers (Allen 1990b; 1991) exiting and depositing their load into the now incumbent estuarine conditions. Such transported material is apparently better represented in the sediments of mudflats. Thus, and unsurprisingly, there is evidence from the sites analysed of significant variations in the local character of the floodplain and estuarine environment with mudflat, salt-marsh and incised channels. As Allen (1990b) points out, theoretical, mudflat- marsh growth is determined by rates of minerogenic and organogenic sedimentation, the rate of change and tendency of relative sea-level and the rate of ‘long-range’ sediment compaction. The fringing fen and accumulating peat would have been moved transgressive inland with rising RSL giving higher ground water table and thus potential for anaerobic conditions along river valleys and along the terrestrial fringes. Negative tendency may also give rise to peat accumulation in areas previously subject to salt-mash/tidal influence; effectively an off-lapp situation such as indicated in

97 context 211 at Vimpenny’s Lane. This reflects the complex causes and responses to a range of controlling factors in estuarine environments/habitats (Long et al. 2000).

At Hallen although the lower levels (zone HALL:1) are not clearly dated, it is thought that they correlate with the lower part of the upper Wentlooge Formation. As at Awkley, it appears that mudflat developed into salt-marsh. Whether this was positive or negative sea-level tendency is not clear but its seems more plausible that the on-site changes reflect a critical balance between sediment supply and vegetation dynamics.

A significant stratigraphical break has been detected which occurs within the upper Wentlooge Series at three of the sites: at 3.98 m OD at Vimpenny’s Lane, 3.95 m OD at Awkley Lane, and 4.80 m OD at Hallen. This horizon appears to represent a stasis horizon resulting in a hiatus in sediment deposition caused by a significant negative tendency of sea-level change during the Iron Age period. This horizon is visible in many sections of the Avon Levels and appears to be delimited by change from lower, typical grey, estuarine/marine silts to brown oxidised silts/clays (see above). Once thought to be a post-depositional colour change, this is now shown to be due to differing sediment sources. This stasis horizon has important archaeological implications since at sites such as Hallen this standstill or negative tendency allowed increased human activity on the salt-marsh. At Hallen this horizon is intricately linked with Iron Age occupation dated to 390-110 cal BC (2192±55 BP, NZA-12727) where there is some evidence that the marsh developed into a drier and possibly rough pasture community on which Iron Age occupation existed at this time.

Our model indicates that this stasis event was terminated by a final (Iron Age–Romano-British) positive tendency which saw progressive waterlogging, sediment accretion, and a return to estuarine salt-marsh and mudflat conditions. At Hallen and Vimpenny’s Lane, this became initially mudflat. However, there was an ephemeral phase of salt-marsh development at Awkley at c. 4.88 m OD (context 204; VIMP:2b). This may represent an oscillation of RSL (negative) or some other local factor (drainage?) which allowed development of middle/lower salt-marsh dominated by halophytic plants. Unfortunately there was too little organic material to allow radiocarbon dating of this horizon and the possibility of an anthropogenically (post Iron Age) induced negative RSL tendency through drainage cannot be precluded. This event was, however, short lived and there was a return to mudflat (context 203).

Archaeology and Environment; marsh landscape and archaeology

Some archaeological implications

The detailed and dated environment studies have provided information of the evolution of the Avon Levels (presented above). They also provide a framework for understanding the patterns and nature of human action and intervention in this landscape which seems to be localised and both specialised and intermittent. Obviously human action itself is also an important factor in the development and modification of this landscape. The changing wetland environment facilitates and curtails exploitation of some resources and provides strict temporal, seasonal, and spatial parameters for settlement or occupation. The archaeological evidence from the sites discussed here (especially Hallen Marsh and Northwick) is reviewed in relation to other recent archaeological finds to provide an explanation of human behaviour which is integrated with the changing nature of the landscape and its potential resources. It is recognised that throughout prehistory and the early historic period the Levels provided but one small area (physically and economically) of larger social and economic systems.

98 Limited evidence of changing RSL has been obtained, but detailed fluctuations of positive and negative tendency have not been isolated, and it is these that might have a significant effect on the parameters of human behaviour and patterns of activity. The importance of understanding the evolution and changing nature of these small Levels cannot be over-emphasised in terms of providing a framework and setting for human action. In particular, changes in the environmental conditions are closely examined for the role they may have played in enabling and curtailing utilisation, exploitation and settlement within the Avon wetlands. Obviously the earlier Holocene (Mesolithic) activity, although sparse, is not considered here, as no deposits relating to this epoch were examined. Instead the reader is directed to research and review by Druce (1997; 1998; 2001) which covers this aspect more adequately from peat data at minus OD heights exposed on the foreshore around Gravel Banks.

History of human exploitation patterns and behaviour

Conventionally the Avon Levels is essentially devoid of any evidence of Neolithic activity (Darvill, in Aston & Iles 1986). Unlike the Somerset Levels, there is little evidence for the Neolithic occupation of the Avon Levels. The Middle Holocene environment comprised patchy discontinuous fen carr woodland fringing more extensive reed-swamp fen with localised dryer marshes and woodlands of oak, elm, ash and hazel. Woodland cover that existed seems to have persisted later than in many other areas. Although substantial clearances were made in parts of the Somerset Levels at about 3500 cal BC (Beckett & Hibbert 1978), much of the lower Severn Valley was not seriously deforested until the later part of the Neolithic (Brown 1982). Evidence of regional and climatic vegetation succession is recorded during the Neolithic and the elm decline is present at Awkley Lane (c. 3550 cal BC, 4800 BP). In addition limited human interference of the climax woodland can be detected, and there clear evidence of disturbed ground (nettles, ribwort plantain, etc).

Although Neolithic artefacts are present at Hills Flats as stray finds on the foreshore to the north in the inner estuary (Allen 1997; 1998); they are absent from the main Avon Levels. It is possible that, if present, they are buried by several metres of later Holocene minerogenic sediments, however, the Holocene environment does not seem to be conducive to settlement, and as yet no ‘islands’ of archaeology, or trackways such as are present in the Somerset Levels, are known. Despite this, there is some cultivation on the edge of, or adjacent to, the Avon Levels as indicated in the presence of cereal pollen at Awkley Lane from about 4530–4350 cal BC (c. 1.5 m OD), with sporadic occurrences throughout the Neolithic peat. From this sparse evidence, enhanced by the very limited archaeological intervention in the Levels with information at about this date, we can envisage the area as one cautiously and periodically exploited for a range of fish, fowl and plants (reeds, Phragmites) to compliment the resources from the few Neolithic sites on the surrounding higher land (flint scatters at Almondsbury (unpublished SMR data)). Whether cereal cultivation look place on the Levels or slightly higher and drier land of the narrow scarp foot bench on its margins is undetermined. Nevertheless, such activity took place within the wider context of the Levels landscape.

Within the Avon Levels themselves the only artefacts are occasional Neolithic axes (three) from intertidal and near wetland contexts along the coastal margins (Allen 1990d), and Neolithic flint knapping sites on the coastal margins in the inner estuary at Oldbury Flats (Allen 1990c, 34) and Hills Flats (Allen 1997). These may represent localised flintworking, and perhaps the exploitation of flints from mixed, now almost offshore, gravels in the fashion of exploitation regimes postulated for the river gravels in Langstone Harbour, Hampshire (Allen and Gardiner 2000; 2007).

99 Very sparse evidence of later Neolithic and earlier Bronze Age activity, save a few localised flint scatters which have become exposed in coastal locations such as Hills Flats (Allen 1997; 1998), tend to suggest that the area was a backwater, perhaps providing less important resources on rare forays into the area.

Local stasis horizons are noted within the middle Wentlooge Formation comprising varyingly organic surfaces. One such weakly organic horizon at about 4.5–4.7 m OD at the Cabot Park and seabank sites (see Table 12 above and Locock 1999) indicates localised soil ripening and gave consistent dates around 2700 cal BC. This might have provided an opportunity for increased human exploitation. But to date, this horizon has been devoid of evidence of human exploitation wherever it has been encountered, though fragmentary charcoal has been recovered from samples (Locock 1997a; 1999) possibly indicating anthropogenic activity in the vicinity.

It is not until the later Bronze Age that a series of ‘settlements’ can be detected in the Levels. A cluster of similar sites occurs around Cabot Park (Locock et al. 1998) and Avonmouth but this concentration is largely an artefact of recent industrial development at this location which has led to archaeological exploration, and we can expect similar sites to be widely dispersed over much of the Levels.

Fen carr and swamps had been inundated by fluvial and intertidal sediment creating a more uniform environment of salt-marsh; freshwater marsh occurred locally, fringing the higher ground and associated with streams draining from the hills to the coast. This open salt-marsh with some higher and drier patches enabled limited soil formation; typically of a saline alluvial gley soil (Avery 1990) with some soil ripening. This is reflected in a stasis horizon observed at 5.1–5.2 m especially around Cabot Park and Avonmouth (Locock et al. 1998, table 1; Locock 1999), consisting of a stabilisation horizon with incipient gleying, evidence of soil ripening and visually appearing slightly more humic. Evidence of human occupation has not been found at every location that this ‘Kites Corner’ horizon occurs, but its present indicates some localised high spots or drier patches within a wider salt-marsh which graded gently to mudflat at its estuarine and river margins within the salt-marsh.

The archaeological sites are ‘set back’ from the present coastal margins (Fig. 23), in contrast to sites of the same date in the Gwent Levels, where there has been a 1–2 km loss of the Levels over the last 2000 years (Bell et al. 2000), which lie on and below peat shelves on the present day tidal edge, and are appreciably higher than those in Wales (5.1 vs 2.5–3.9 m OD). The Avon sites are ephemeral and comprise charcoal patches, and concentrations of animal bone, burnt stone, imported stone, and pottery, with no, or scant evidence of structures (Locock et al. 1998). These sites are not distinguishable on the basis of the physical evidence and tend to indicate localised temporary activities on the salt-marsh edge, perhaps near to former small water courses. Locock (2000) has suggested that these might be temporary seasonal ‘camps’ associated with summer grazing, saltmaking and possibly even potting. The sites were not long lived, nor re-visited, even though the general vicinity was. What is also evident from the radiocarbon determinations is that this activity persisted almost unchanged for about one millennium from 1780–1510 cal BC to 930–520 cal BC.

These locations seem to have operated in spatial and temporal isolation from each other; there is no evidence of any infrastructure or route-, or track-, ways. This contrasts with Caldicot Levels where brushwood trackways have been found at Cold Harbour Pill (Locock 1997b) that probably date to around 1300–450 cal BC. Evidence of trackways and more permanent infrastructures in the Gwent Levels fits with evidence of later Bronze Age round-houses and more substantial settlements at fig 23

100

Chapel Tump (Whittle 1989), Rumney Great Wharf (Allen 1996a), and Collister Pill (Bell et al. 2000, 309–11). Similarly short-lived trackways existed in the Somerset Levels (Coles & Coles 1986; 1998) linking foci of semi-permanent Bronze Age activity on islands of slightly higher ground. These more complex social systems do not seem to have developed, or have not yet been discovered, in the Avon Levels.

Kites Corner dense charcoal patch 3m across, and wider spread of charcoal, some stake-holes 5.1 m OD 2610±70 BP (920–520 cal BC) Avonmouth 1 scatter, stone and charcoal, some pottery 5.3 m OD 2778±55 BP (1110–810 cal BC) 5.3 m OD 2957±55 BP (1370–1010 cal BC) Little Googs dense scatter of burnt stone, bone, pottery, and charcoal (6 m area) 5.1-5.2 m OD 2970±60 BP (1390–1020 cal BC) 5.1-5.2 m OD 3350±60 BP (1870–1500 cal BC) Stinkums 3 sites, burnt stone, animal bone, and pottery

By about 450 cal BC all of the later Bronze Age sites known in the Avon Levels were inundated by minerogenic sediments relating to salt-marsh or mudflat environments. Abandonment is asynchronous, in keeping with the activity, and suggest a general diminishing of exploitation rather than large- and wholesale changes in the environment rapidly curtailing these summer forays. This fits with the environmental evidence of gradual changes and fluctuations in positive and negative RSL at this time. So, as environmental change gradually lead to the lessening of human activity, a period of seldom/sparse human invention followed.

By the later Iron Age (350 cal BC), whereas mudflats were more widespread previously, now they only fringed open salt-marsh along alluvial or estuarine inlets and much of the sedimentation was essentially marine. The subtle transition to increased fluvial sedimentation, slightly longer and drier summer spells and increased freshwater fluvial input into the wetland system led to subtle changes in the observed sedimentary sequence and prescribed the change from the middle to upper Wentlooge Formation, as described earlier. Unfortunately we have no firm dates for this change, but is likely to asynchronous across the region, varying in accordance to proximity to the shore, tidal inundation sources and freshwater streams.

Very little Iron Age evidence is known from the Levels as a whole (Fig. 23) but at Hallen Marsh, by the Late Iron Age (200 cal BC), relatively substantial settlement is present, and to date there is no evidence of any earlier Iron Age activity in the Avon Levels. By the later Iron Age complex societies had developed with both permanent and specialised sites. Hillforts are present on the upland overlooking the Avon Levels; the Hallen site is situated within the Levels, in contrast to those in north Somerset which seem to be specifically located on the fen edge or in the Brue Valley, where the freshwater peats have produced the famous sites of Glastonbury and Meare (Coles & Coles 1996; Coles & Orme 1983; 1985).

Although the Iron Age activity at Hallen Marsh is considerable more substantive than that of the later Bronze Age sites, it nevertheless, does not indicate permanent settlement such as those known on the higher lands. The settlement at Hallen Marsh comprises at least two complex round-house structures (Plate 4) sited on either side of a small stream. All of the artefacts were readily portable and it seems that all the necessities of basic life, including staple foods, were brought to the site, either in containers or on the hoof. The site seems to have been occupied on a seasonal basis, specifically for the grazing of sheep, with some cattle, horses, pigs, and dogs present. The animal bone assemblage indicates a ‘normal’ herd structure, with animals being killed as required for immediate consumption. It seems likley that Hallen represents evidence for transhumance between

101 353000 354000 355000 356000357000 358000 359000 360000

188000

187000

186000 Northwick sCae Green Awkley Interface Lane Bri tol h nn l

185000

Rookery Farm Ellinghurst Farm 184000 Caltybrook Farm Brynleaze Farm Almondsbury Fort

183000

Crookmarsh Farm 182000

Elmington Manor

181000

Rockingham Farm Cabot Hallen 180000 Park

Katherine’s Farm

179000

Lawrence Blaise Castle Land over 10m Weston Long Cross Bronze Age Barrow/findspot 178000 Kings Late Bronze Age Site Weston Kings Weston Hill Middle-Late Iron Age Site Romano British Site

177000

Figure 23. Location of archaeological sites in the Avon Levels

102 the limestone uplands and the rich summer pasture of accessible parts of the Levels (see Gardiner et al. 2002 for a full discussion and Fig. 24).

Why does settlement occur?

There is no major change in the local environment, but then there need not be large changes in the ‘marginal’ landscape to enable it be utilised for exploitation and grazing. As we have discussed there is a minor change in local conditions (the middle/upper Wentlooge boundary), but this may not in itself have been great enough to engender the onset of human exploitation and settlement. As we cannot invoke environment deterministic ideals, we may examine this in terms of environmental possibilism. It is the combination of a subtly and slowly changing landscape enabling some pasture and seasonal occupation, in combination with increasingly complex and larger societies (the Avon Levels in the later Iron Age formed part of the territory of the Dobunni tribe; Cunliffe 1991) which enabled and facilitated the necessity to utilise the Avon wetlands (Fig. 24).

The site at Hallen Marsh is specialised and seasonal, undoubtedly concentrating on summer grazing and production of local and portable crafts. The lack of charred grain from the settlement (see Clapham in Gardiner et al. 2002) is likely to be a real, rather than taphonomic, phenomenon. Grain was present for preparation and consumption as a stored food stuff, but agriculture was not part of the Hallen economy.

The two round-houses lie either side of stream, which itself seems to be upper profile of a much early palaeochannel that cuts through the earlier, middle Wentlooge peats, at this location (Druce 2001; Juggins 1982). This situation was probably important for the welfare of both humans and livestock. Although not clearly ascertained from analysis, this is likely to have been largely freshwater, but inundated with brackish water on higher and storm tides.

What caused Iron Age abandonment?

Cessation for settlement is not due to environmental catastrophe or rapid change. There is evidence at Hallen of increased brackish and marine sediment input, probably from overbank flooding of the stream during the period of later Holocene positive RSL. This led to increasingly wetter conditions prevailing during longer periods of the year. Summer grazing of stock, especially sheep, became unfeasible. Ultimately the settlement itself was inundated with mineral sediment and the artefacts washed over the site and incorporated in the overlying, newly forming alluvial gley soil in the salt- marsh. This was inevitable and was understood by the inhabitants of Hallen, for not only were the buildings located on slightly higher land but attempts were made to consolidate the entrances-ways of each by adding imported stone, and discarded bone and broken pottery to the paths. The flooding was exacerbated locally by the proximity of the stream, but ultimately represents a wider phenomenon. Thus, although ultimately we can see that the local environmental conditions were responsible for abandonment of the Late Iron Age summer grazing regimes at Hallen Marsh, this was a gradual retreat of Iron Age use of the area. A continued cycle of mudflat/salt-marsh – salt- marsh/mudflat during positive RSL made the region unpredictable and thus, in economic terms, unviable. This rise in RSL is seen as the cause of the abandonment of a number of Iron Age– Romano-British sites throughout the Somerset (Hibbert 1980) and Gwent Levels (see Boon 1980), and was first defined by Godwin (1940; 1943).

103 Topographic Neolthic Later Early Mid-Late Early Late location Bronze Age Iron Age Iron Age Roman Roman

Upland Monuments Barrows No Hillforts Rural Villas and evidence settlements temples

Some Development settlement Development of of settlement tribal territory hierarchy and communications Flint & Flint Farmsteads stone axes, scatters flint scatters Farmsteads

Lower Possible No Small towns Slopes clearance Some evidence some arable clearance ?Fields arable Rural Fen Edge Fish and No Mixed farmsteads, fowl evidence farming mixed farming

Reed Open Mudflat/ Increasing seasonal flooding Fen swamp salt-marsh salt-marsh leading to permanent waterlogging

Pot Increasing Seasonal making Seasonal seasonal Probably No grazing with Fen grazing flooding not evidence field systems and temporary leading to Islands present for presence and temporary Summer settlement permanent grazing settlement waterlogging

Intertidal Mudflat/ Saltworking Mudflat/ Increasing seasonal flooding Zone salt-marsh salt-marsh

Figure 24. Suggested model for the exploitation of the Avon Levels.

105

A change of use

Some human exploitation also occurs in the Levels in the earlier Romano-British period. Field systems seem to have been established in the salt-marsh in a few locations such as Northwick, bounded by relatively deep, but well-defined ditches. We have no evidence of major environmental change at this time though local variations in ground-water regimes will have been important factors; minor and subtle negative changes in RSL would make these developments feasible. This activity will have been accompanied by some occupation in the Levels – quantities of pottery, animal bone, and charcoal occur in the ditches at Northwick, but the nature and extent of such activity was not elucidated in excavation here. The limited evidence that so far exists for Romano- British occupation in the Levels themselves seems to be specifically early Roman in date (1st–2nd century AD). Crook’s Marsh is, to date, is the only later Roman site to have been identified within the Levels (Everton & Everton 1980; Allen & Fulford 1986; Juggins 1982, 21–31). Here a series of ditches was identified, not dissimilar to Northwick. Both probably represent paddocks, rather than arable fields, and more formal and controlled stock husbandry; presumably based on summer grazing. These fields, with often water-filled ditches, were probably used for seasonal graze and, therefore, as an economic activity were less jeopardised by gradual rising RSL and increased periodic inundation.

Other Romano-British settlements are sited on the fen-edge and on the slopes of the higher ground overlooking the Levels. A number of sites north of the Avon and centred around Lawrence Weston and Kings Weston are particularly notable, with habitation spanning the 1st–4th centuries AD but the majority of dated artefacts are of later Romano-British date (3rd–4th century AD).

In general, although we can see formalised control of stock in the Levels, the utilisation and occupation of the area is at a lower and more ‘rural’, nature than that seen in both the Gwent and the Somerset Levels. Whether there is continuity of occupation at sites like Northwick, where Iron Age pottery also occurs, cannot yet be determined but certainly by the later Romano-British period settlement was sited on the distinctly higher land peripheral to the Levels themselves. Ultimately, even the fen-edge fields too, were inundated by further minerogenic sedimentation. Agricultural development of the area resumed in the medieval and post-medieval periods with large-scale ridge and furrow agriculture (Fig. 24).

Land claim

Much has been made previously of Roman land claim of coastal margins of the Severn Levels (eg, Rippon 1997), and this is well demonstrated at Rumney Great Wharf and along the Wentlooge Levels (Allen & Fulford 1986). Similar piecemeal land claim is also evident along the narrow inner estuary at Longney (Allen & Fulford 1990a), Elmore (Allen & Fulford 1990b), and Oldbury-on- Severn in Gloucestershire (Allen & Fulford 1990b; Riley 1998). In keeping with the lower level of activity, along the Avon Levels land claim is, however, less well testified, and we can argue that such activity did not take place along this stretch of coastline. This has implications for the nature and intensity of Romano-British activity. Unlike the Wentlooge Levels where major settlement occurs within the former fens and industrial activity (smelting) occurs, a much smaller scale of operation seems evident in the Avon Levels.

105 Archaeological significance

Here we have summarised some of the main events defined by the archaeological evidence and relating to human behaviour. In this landscape, which is subject to distinctive changes resulting from minor fluctuations in RSL and runoff, changes in the vegetation and local environment are inextricable tied to the nature and phases of human activity. The analyses reviewed in this paper provide a more detailed geoarchaeological picture of the Avon Levels than was previously possible. A full discussion of the rchaeological evidence and a model for the prehistoric exploitation and georachaeology is presented elsewhere (Gardiner et al. 2002).

Conclusions

The analytical programme discussed here has enabled us to define the middle and upper Wentlooge Formations in the field and analytically in terms of subtle changing vegetation patterns. It has also provided the basis for re-examination of patterns of human behaviour form the Neolithic to Romano-British periods which are summarised here and explored in more detail elsewhere (Allen & Gardiner, in Gardiner et al. 2002). We can, therefore, summarise the main events in terms of the Wentlooge Formations as follows.

Formation deposit Period Environmental conditions Archaeology Upper Wentlooge reddish Romano-British and return to saltmarsh, some ditched field systems & estuarine later (500 BC +) gradual flooding & drained landscape silts/clays inundation especially associated with channels.

Middle blue & grey Bronze Age–Iron salt-marsh & intertidal Human exploitation of Wentlooge estuarine silts/ Age (4500–800 BC) mudflats the saltmarsh: Cabot clay over peat park & Avonmouth (LBA), Hallen Marsh (IA), Northwick (RB)

Lower Wentlooge peats or Mesolithic–Neolithic salt-marsh, wet alder carr Potential for fishing and blue/grey (5500–4500 BC) with local Phragmites reed fowling. Limited estuarine silts swamps, raised bogs, minor artefacts & occupation over sands rivers & numerous small streams

Acknowledgements The authors would like to thank all of those who contributed in the field and to the assessment and post-excavations programmes of the English Approaches Project, and especially to those who contributed to this paper. Their written words are produced here but the text also includes their thoughts and ideas from long and detailed discussions. The principal authors would like to thank all the specialists who contributed to the programme and to this report. In addition we thank Mark Noel and associates (archaeomagnetic dating) and particularly Karen Nichols for interpreting our rough drawings and workings-out to produce the illustrations.

This paper has benefitted from discussions with and information provided by numerous colleagues and specialists, of whom Julie Gardiner, Anthony Long, Martin Locock, and Steve Rippon deserve special mention. Steve Juggins allowed us to use and quote from his long-lost undergraduate dissertation and Denise Druce was especially generous in discussing the results of the PhD research and supplying extracts of her thesis. Julie Jones and Vanessa Straker provided information and

106 copies of various reports unavailable to us. Thanks are also due to Richard Newman whose foresight prior to and during fieldwork in 1992–4 has finally been realised by this and its complimentary papers. The paper was edited for publication by Julie Gardiner, who would like to thank all the contributors and referees for their patience and assistance. The post-excavation project was funded by Highways Agency, managed for them by Ian Smith, and monitored on their behalf by Janet Miller of W.S. Atkins Heritage.

107

Appendix: Analytical Techniques

Pollen analysis by Robert G. Scaife

Standard techniques were used for the extraction of the sub-fossil pollen and spores (Moore & Webb 1978; Moore et al. 1991). Micromesh sieving (10 m) was used to aid the removal of the substantial clay fraction in the grey sediments. Samples of 1.5–2 ml were used for the peat and 5 ml for the overlying grey minerogenic sediments. Absolute pollen frequencies were calculated using added exotics to known volumes of sample (Stockmarr 1971). Pollen was identified and counted using an Olympus biological research microscope fitted with Leitz optics. A pollen sum of 400 grain of dry land taxa plus all extant spores and pollen of marsh taxa (largely Cyperaceae and Alnus), fern spores, and miscellaneous pre-Quaternary palynomorphs was counted for each level where preservation made this possible. This was generally achieved except in some levels of the upper minerogenic sediments. Very substantial total pollen numbers total were counted for the lower peat deposits where pollen of Alnus, Betula, and fern spores were extremely abundant. An extensive reference/comparative collection of modern taxa was available to aid identification. Pollen diagram(s) have been constructed and plotted using Tilia and Tilia Graph. Percentages have been calculated as follows:

Sum = % total dry land pollen (tdlp) Marsh/aquatic = % tdlp+sum of marsh/aquatics Spores= % tdlp+sum of spores Misc.= % tdlp+sum of misc. taxa.

Alnus, with its extremely high values and on-site dominance in the lower pollen assemblage zone, has been excluded from the pollen sum and is incorporated into the marsh/aquatic category as originally detailed by Janssen (1969). Taxonomy in general follows that of Moore and Webb (1978) modified according to Bennett et al. (1994) for pollen types and Stace (1997) for plant descriptions. These procedures were carried out in the Palaeoecology Laboratory of the Department of Geography, University of Southampton.

Diatoms by Nigel G. Cameron

Diatom preparation followed standard techniques. Two coverslips, each of a differing concentration of the cleaned solution, were prepared from each sample. Where the dilutions were not satisfactory for counting a second set of slides was prepared. Slides were scanned thoroughly at magnifications of x400 and at x1000 under bright-field, phase contrast, and differential interference contrast illumination.

Waterlogged plant macrofossils by Alan J. Clapham

3 Sub-samples of 300 cm from each sample were processed via wet sieving through a series of granulometric sieves with mesh diameters of 1 mm, 500 μm and 300 μm. The resulting fractions were sorted under water using a low-powered stereomicroscope (x6–x40) and the critical taxa were compared with the modern seed reference collections housed in the George Pitt-Rivers Laboratory, department of Archaeology, University of Cambridge. The taxa identified were then counted. All nomenclature follows that of Stace (1997).

108 Molluscs by Michael J. Allen

Bulk samples of up to 2 kg were processed by standard wash-over flotation methods (Evans 1972). Because of the sparse and fragile nature of shells from these deposits, extreme care was taken in their dissaggregation; in cases where the silts did not disaggregate upon immersion in warm water, they were subject to up to to three cycles of freeze-thaw (see Bell in Balaam et al 1987) and re-immersed in warm water. After disaggregation the flots were retained on a 0.5mm mesh and the residues passed through a 0.5 mm mesh. The lack of any residue negated the necessity of passing the minerogenic fraction through a stack of sieves.

Insects by Mark Robinson

Samples of 2 kg were subjected to paraffin flotation to recover insect remains. The flot was caught on a 0.2 mm sieve, washed in detergent and sorted under a binocular microscope. Nomenclature for Coleoptera follows Kloet and Hincks (1977).

Sediments by J.R.L. Allen

Samples were analysed for grain-size distributions and plotted on triangular diagrams to show values of the clay-silt-sand ratio. Each sample was coarsely but non-destructively crushed and quartered to yield a sub-sample. This sub-sample was crushed more finely and reduced to a further sub-sample weighing a few grams. This further sub-sample was then mixed with a weak aqueous solution of Calgon to form a paste and spread out over a glass plate. From the ‘biscuit’ so formed, a final sub-sample of a few tenths of a gram was removed by jabbing a tiny spatula into the biscuit at numerous, well-scattered points. The final sub-sample was then introduced to a Coulter LS laser granulometer and a count allowed to proceed for 1–2 minutes. Using the Fraunhofer with PIDS model as the basis for calculating the results, the resulting machine output chiefly lists measures of central tendency (especially mean and modal grain sizes), selected percentile values, and the clay- silt-sand ratio.

Charcoal by Rowena Gale

Bulk samples were processed by floation with flots and residues retaiend on a 0.5 mm mesh. Samples of charcoal were prepared for examination using standard methods. Fragments from each sample were fractured to expose fresh transverse surfaces (TS) and sorted into groups based on the anatomical features observed using a x20 hand lens. Representative fragments from each sample were selected for detailed study at high magnification. Additional surfaces to show the wood structure in the tangential (TLS) and radial planes (RLS) were also prepared. The fragments were supported in washed sand and examined using a Nikon Labophot microscope at magnifications of up to x400. The anatomical structures were matched to prepared reference slides. When possible the maturity (ie, heartwood/ sapwood) of the wood was assessed. Classification follows that of Flora Europaea (Tutin, Heywood et al 1964–80).

109 Magnetic susceptibility by Michael J. Allen

Samples for magnetic susceptibility analysis were extracted from the sample columns at 10 cm intervals through the upper alluvial sequences at each site, and from the Romano-British ditches at Northwick. These were air-dried, crushed and passed through a 2 mm sieve with 10 g of the <2 mm fraction measured using a Bartington MS2 meter coupled to a MS2b dual frequency sensor. Measurements were made at 0.46kHz (low frequency) and at higher frequency (4.6kHz).

110

Bibliography

Allen, J R L, 1986 A short history of salt-marsh reclamation at Slimbridge Warth and neighbouring areas, Gloucestershire. Trans Bristol Gloucestershire Archaeol Soc 104, 139–55 Allen, J R L, 1987a Late Flandrian shoreline oscillations in the Severn Estuary: the Rumney Formation at its type site (Cardiff area). Phil Trans Roy Soc London B 315, 157–84 Allen, J R L, 1987b Dimilington Stadial (Late Devensian) ice-wedge casts and involutions in the Severn Estuary, southwest Britain. Geol J 22, 109–18 Allen, J R L, 1988 Modern-period muddy sediments in the Severn Estuary (south western UK): a pollutant-based model for dating and correlation. Sediment Geol 58, 1–21 Allen, J R L, 1989 Evolution of salt-marsh cliffs in muddy and sandy systems: a qualitative comparison of British west- coast estuaries. Earth Surface Processes & Landforms 14, 85–92 Allen, J R L, 1990a Late Flandrian shoreline oscillations in the Severn Estuary: change and reclamation at Arlingham, Gloucestershire. Phil Trans Royal Soc London A330, 315–34 Allen, J R L, 1990b The Severn Estuary in southwest Britain; its retreat under marine transgression, and fine-sediment regime. Sediment Geology 66, 13–28 Allen, J R L, 1990c The post-glacial geology and geoarchaeology of the Avon wetlands. Proc Bristol Natur Soc 50 (special issue), 28–46 Allen, J R L, 1990d Three Neolithic axes from the Severn Esturay. Trans Bristol Gloucestershire Archaeol. Soc. 108, 171–74 Allen, J R L, 1991 Fine sediment and its sources, Severn Estuary and inner Bristol Channel. Sediment Geol 75, 57–65 Allen, J R L, 1992a Trees and their response to wind: mid Flandrian strong winds, Severn Estuary and inner Bristol Channel, southwest Britain. Phil Trans Roy Soc London B, 338, 335–64 Allen, J R L, 1992b Tidally influenced marshes in the Severn Estuary, southwest Britain, in Allen, J R L and Pye, K (eds), Saltmarshes; morphodynamics, conservation and engineering significance. Cambridge: Univ Press, 123–47 Allen J R L, 1995 Salt-marsh growth and fluctuating sea level: implications of a simulation model for Flandrian coastal stratigraphy and peat-based sea-level curves. Sediment Geol 100, 21–45 Allen, J R L, 1996 Three final Bronze Age occupations at Rumney Wharf on the Wentlooge Level, Gwent. Studia Celtica 30, 1–16 Allen, J R L, 1997 A scatter of Neolithic–Bronze Age flintwork from the intertidal zone at Hills Flats, South Gloucestershire. Trans Bristol Gloucestershire Archaeol Soc 115, 265–276 Allen, J R L, 1998 A prehistoric (Neolithic–Bronze Age) complex on the Severn Estuary Levels, Oldbury-on-Severn, South Gloucestershire. Trans Bristol Gloucestershire Archaeol Soc 116, 93–115 Allen, J R L and Fulford, M G, 1986 The Wentlooge Level; a Romano-British saltmarsh reclamation in southeast Wales. Britannia 17, 91–117 Allen, J R L and Fulford, M G, 1988, Romano-British settlement and industry on the wetlands of the Severn Estuary. Antiq J 67, 237–89 Allen, J R L and Fulford, M G, 1990a Romano-British wetland reclamations at Longney, Gloucestershire, and evidence for the earlier settlement of the inner Severn Estuary. Antiq J 70, 288–326 Allen, J R L and Fulford, M G, 1990b Romano-British and later reclamations on the Severn salt-marshes in the Elmore area, Gloucestershire. Trans Bristol Gloucestershire Archaeol Soc 108, 17–32 Allen, J R L and Rae, J E, 1987 Late Flandrian shoreline oscillations in the Severn Estuary: a geomorphological and stratigraphical reconnaissance. Phil Trans Roy Soc London B 315, 185–230 Allen, J R L and Rae, J E, 1988 Vertical salt-marsh accretion since the Roman period in the Severn Estuary, southwest Britain. Marine Geol 83, 225–235 Allen, M J, and Gardiner, J P, 2000 Our Changing Coast; a survey of the intertidal archaeology of Langstone Harbour, Hampshire. York: Counc Brit Archaeol Res Rep 124 Allen, M J and Gardiner, J P, 2007 The Neolithic of Langstone Harbour: dryland resources preserved by a wetland environment, in Haughey, F and Siddell, J (eds), Current Research in Neolithic Intertidal Archaeology. Oxford: Neolithic Studies Group, 59–70 Allen, M J, and Macphail, R I, 1987 Micromorphology and magnetic susceptibility studies; their combined role in interpreting archaeological soils and sediments, in Fedoroff, N, Bresson, L M, and Courty, M A, (eds), Soil Micromorphology. Paris: Association Francaise pour l'etude du sol, 669–76 Allen, M J and Richie, K, 2000 The stratigraphy and archaeology of Bronze Age and Romano-British deposits below beach level at Brean Down, Somerset. Proc Univ Bristol Spelaeol Soc 22, 5–52 Andersen, S Th, 1970 The relative pollen productivity and pollen representation of North European trees, and correction factors for tree pollen spectra. Danm. Geol. Unders. Ser I 196, 99

111 Andersen, S Th, 1973 The differential pollen productivity of trees and its significance for the interpretation of a pollen diagram from a forested region, in Birks, H J B and West, R G, Quaternary Plant Ecology. Oxford: Blackwell, 109–115 Andrews, J T, Gilbertson, D D and Hawkins, A B, 1984 The Pleistocene succession of the Severn Estuary: a revised model based upon amino acid racemiztaion studies. J Geol Soc London 141, 967–74 Aston, M and Iles R, 1986 The Archaeology of Avon. Bristol: Avon County Council Avery, B W, 1990 Soils of the British Isles. Wallingford: CAB (Commonwealth Agricultural Bureau) International Balaam, N D, Bell, M G, David, A E U, Levitan, B, Macphail, R I, Robinson, M, and Scaife, R G, 1987 Prehistoric and Romano-British sites at Westward Ho! Devon: archaeological and palaeoevironmental surveys 1983 and 1984, in Balaam, N D, Levitan, B, and Straker, V, Studies in palaeoeconomy and environment in southwest England. Oxford: Brit Archaeol Rep 181, 163–259 BARAS, 1998 Archaeological watching brief on the laying of an effluent pipeline between Avonmouth sewage works and Saebank power station, Severnside, Bristol for Wessex Water plc. BARAS report 265/1998 Barnes, I, 1993. Second Severn Crossing: English Approaches an interim statement on the 1992/3 fieldwork. Ann Rep Severn Estuary Levels Comm, 3–30 Beckett, S C, 1979, The environmental setting of the Sweet Track Drove site, in Coles, J M (ed), Somerset Levels Pap 5, 75–83 Beckett, S C and Hibbert, F A 1976, An absolute pollen diagram from the Abbot’s Way, in Coles, J M et al Somerset Levels Pap 2, 24–27 Beckett, S C and Hibbert, F A, 1978 The influence of man on the vegetation of the Somerset Levels – a summary, in Coles, J M (ed) Somerset Levels Pap 4, 86–90 Beckett, S C and Hibbert, F A, 1979 Vegetational change and the influence of prehistoric man in the Somerset Levels. New Phytol 83, 577–600 Bell, M G, Caseldine, A, and Neumann, H, 2000 Prehistoric Intertidal Archaeology in the Welsh Severn Estuary. London: Counc Brit Archaeol Res Rep 120 Bennett, K D, Whittington, G and Edwards, K J, 1994 Recent plant nomenclatural changes and pollen morphology in the British Isles. Quat Newslet 73, 1–6 Birks, H J B, 1989 Holocene isochrone maps and patterns of tree-spreading in the British Isles. J Biogeog 16, 503–40 Birks, H J B, Deacon, J and Peglar, S, 1975 Pollen maps for the British Isles 5000 years ago. Proc Royal Soc London B189, 87–105 Boon, G C, 1980 Caerleon and the Gwent Levels in early historic times, in Thompson, F H, (ed), Archaeology and Coastal Change. London: Soc Antiq London Occas Pap NS 1, 24–36 Boyden, C R, Crothers, J H, Little, C, and Mettam, C, 1997 The intertidal invertebrate fauna of the Severn Estuary. Field Stud 4, 477–554 Brown, A G, 1982 Human impact on the former floodplain woodlands of the Severn, in Bell, M, and Limbrey, S, (eds), Archaeological Aspects of Woodland Ecology. Oxford: Brit Archaeol Rep S146, 93–104 Brown, A G, 1985 The potential use of pollen in the identification of suspended sediment sources. Earth Surface Processes and Landforms 10, 27–32 Bulleid, A and Gray, H S G, 1911 The Glastonbury Lake Village. Vol 1. Taunton: Glastonbury Antiq Soc Bulleid, A and Gray, H S G, 1917 The Glastonbury Lake Village. Vol 2. Taunton: Glastonbury Antiq Soc Bulleid, A and Gray, H S G, 1948 The Meare Lake Village. Vol 1. Taunton, Taunton Castle Burrin, P J and Scaife, R G, 1984 Aspects of Holocene valley sedimentation and floodplain development in southern England. Proc Geol Assoc 95, 81–96 Cameron, N G, 1993 Diatom analysis of the Oscar3 Profile, in Godbold, S and Turner, R C, Second Severn Crossing Archaeological Response, Phase 1 – The Intertidal Zone in Wales. Cardiff: Cadw, 61–4 Cameron, N G, 1997 The diatom evidence, in Nayling, N & Caseldine, A, Excavations at Caldicot, Gwent: Bronze Age palaeochannels in the Lower Nedern Valley. York: Counc Brit Archaeol Res Rep 108, 117–28 Cameron, N G & Dobinson, S J, 1998 Diatom analysis of sediments and caulking material, in Nayling, N, The Magor Pill Mediaeval Wreck. York: Counc Brit Archaeol Res Rep 115, 38–40 Cameron, N G & Dobinson, S J, 2000 Diatom analysis of sediments, in Bell, M., Caseldine, A., & Neumann, H, Prehistoric Intertidal Archaeology in the Welsh Severn Estuary. York: Counc Brit Archaeol Res Rep 120, 262–6 Canti, M, Heal, V, Jennings, S, McDonnell, R, and Straker V, 1993 Archaeological and paleoenvironmental evaluation of Porlock Bay and Marsh. Archaeol Severn Estuary, 49–69 Chambers, F, 1989 The evidence for early Rye cultivation in North West Europe, in Milles, A, Williams, D and Gardner N (eds), Beginnings of Agriculture. Oxford: Brit Archaeol Rep S496, 165–75 Churchill, D M, 1965 The displacement of deposits formed at sea level 6500 years ago in Southern Britain. Quaternaria 7, 239–49 Clarke, J C and Patterson, W A, 1984 Pollen, Pb-210 and opaque spherules: an integrated approach to dating and sedimentation in the intertidal environment. J Sediment Petrol 54, 1249–63 Coles, B and Coles, J M, 1986 Sweet Track to Glastonbury; the Somerset Levels in Prehistory. London, Thames & Hudson 112 Coles, B, and Coles, J, 1998 Passages of Time. Archaeol Severn Estuary 9, 3–16 Coles, J and Minnitt, S, 1995 Industrious and Fairly Civilized; the Glastonbury lake Village. Taunton: Somerset Levels Proj/Somerset Co Counc Mus Serv Coles, J and Orme, B, 1983 The Sedgemoor survey, 1982. Somerset Levels Pap 9, 6–8 Coles, J and Orme, B, 1985 Archaeology in the Somerset Levels 1984, Somerset Levels Pap 11, 5–7 Coles, J M, Hibbert, F A and Clements, C F, 1970 Prehistoric roads and tracks in Somerset – Neolithic. Proc Prehist Soc 36, 125–50 Coles, J M, Hibbert, F A, and Orme, B J, 1973 Prehistoric roads and tracks in Somerset, 3 – the Sweet Track. Proc Prehist Soc 39, 256–93 Coles, J M, et al. 1975–1988 Somerset Levels Papers 1–14. Cundill, P and Whittington, G, 1987 Flood plain sedimentation in the Upper Axe Valley, Mendip, England: a comment. Trans Inst Brit Geog 12, 360–2 Cunliffe, B W, 1991 Iron Age Communities (3rd edn). London: Routledge Davies, P, 1996 The ecological status of Pupilla muscorum (Linné) in Holocene overbank alluvium at Kingsmead Bridge, Wiltshire. J Conch Soc London 35, 467–71 Davies, P, 1998 Numerical analysis of subfossil wet-ground molluscan taxocenes from overbank alluvium at Kingsmead Bridge, Wiltshire. J Archaeol Sci 25, 39–52 Denys, L, 1988 Fragilaria blooms in the Holocene of the Western Coastal Plain of Belgium, in Simola, H (ed), Tenth International Diatom Symposium, Koenigstein: Koeltz, 397–405 Denys, L, 1992 A check list of the diatoms in the Holocene deposits of the Western Belgian Coastal Plain with a survey of their apparent ecological requirements: I. Introduction, ecological code and complete list. Service Geologique de Belgique. Professional Pap 246, 41 Devoy, R J, 1977 Flandrian sea level changes in the Thames estuary and the implications for land subsidence in England and Wales. Nature 270, 712–15 Devoy, R J N 1979 Flandrian sea level changes and vegetational history of the Lower Thames estuary. Phil Trans Roy Soc London B285, 355–407 Devoy, R J N 1980 Post-glacial environmental change and man in the Thames estuary: a synopsis, in Thompson, F H (ed), Archaeology and Coastal Change. London: Soc Antiq London Occas Pap NS 1, 134–48 Devoy, R J N 1982 Analysis of the geological evidence for Holocene sea-level movements in south-east England. Proc Geol Assoc 93, 65–90 Dewar, H S L and Godwin, H, 1963 Archaeological discoveries in the raised bogs of the Somerset Levels, England. Proc Pehist Soc 29, 17–49 Dimbleby, G W, 1988 The palynology of Archaeological sites. London: Academic Druce, D, 1997 Palaeoenvironmental constraints on Mesolithic to Romano-British communities of the Severn Estuary. Archaeol Severn Estuary 8, 99 Druce, D, 1998 Late Mesolithic to early Neolithic environmental change in the central Somerset Levels: recent work at Burnham-on-sea. Archaeol Severn Estuary 9, 17–30 Druce, D, 2001 Mesolithic to Romano-British Archaeology and Environmental Change of the Severn Estuary England. Unpubl PhD thesis, Univ Bristol Evans, J G, 1972 Land Snails in Archaeology. London, Seminar Everton, A, and Everton, R, 1980 Romano-British occupation at Crooks Marsh Farm, Avonmouth. Bristol Archaeol Res Group Rev 2, 57–8 Fairbanks, R W, 1989 A 17,000 year glacio-eustatic sea-level record: influence of glacial melting rates on Younger Dryas event and deep ocean circulation. Nature 342, 637–42 Fulford, M, Allen, J R L, and Rippon, S J, 1992 The settlement and drainage of the Wentlooge Level, Gwent: excavation and survey at Rumney Great Wharf 1992. Britannia 25, 175–211 Gale, R. 1992 Wood and charcoal, in Fulford et al 1992, 205–6 Gardiner, J, Allen, M J, Hamilton-Dyer, S, Laidlaw, M, and Scaife, R G, 2002, Making the most of it: Late prehistoric pastoralism in the Avon Levels, Severn Estuary, Proc Prehist Soc 68, 1–40 GeoQuest Associates, 1993. Palaeomagnetic analysis of sediments from sites R10 [Vimpennys Lane], R11 [Hallen] and R15 [Awkley Lane], English Approaches. Unpubl rep for Wessex Archaeol GGAT = Glamorgan Gwent Archaeological Trust, 1992, Archaeology of the Second Severn Crossing: assessment and recommendations for the English Approaches. Unpubl Client Rep Gimmingham, C H, 1984 Ecological aspects of birch. Proc Roy Soc Edinburgh 85B, 65–72 Girling, M A, 1977 Fossil insect assemblages from Rowland’s Track. Somerset Levels Pap 3, 51–60 Girling, M A, 1988 The bark beetle Scolytus scolytus (Fabricius) and the possible role of elm disease in the early Neolithic, in Jones, M. (ed), Archaeology and the Flora of the British Isles. Oxford: Oxford Univ Comm Archaeol, 34–8 Godwin, H, 1940 Pollen analysis and forest history of England and Wales. New Phytol 39, 370–400 Godwin, H, 1941 Studies of the post-glacial history of the British vegetation. VI Correlations in the Somerset Levels. New Phtyol 40, 108–32 Godwin, H, 1943 Coastal peat beds of the British Isles and North Sea. J Ecol 31, 199–247 113 Godwin, H, 1945 Coastal peat-beds of the North Sea region, as indices of land and sea-level changes. New Phytol 44(1) 29–69 Godwin, H, 1946 The relationship of bog stratigraphy to climatic change and archaeology. Proc Prehist Soc 12, 1–11 Godwin, H, 1948 Studies of the post-glacial history of the British vegetation. X Correlation between climate, forest composition, prehistoric agriculture and peat stratigraphy in Sub-boreal and Sub-atlantic peats of the Somerset Levels. Phil Trans Roy Soc London B233, 275–86 Godwin, H, 1956 History of the British Flora. Cambridge: Univ Press Godwin, H, 1960 Prehistoric wooden trackways of the Somerset Levels: their construction. age and relation to climatic change. Proc Prehist Soc 16, 1–36 Godwin, H, 1975a, History of the British Flora. A Factual Basis for Phytogeography (2nd edn). Cambridge: Univ Press Godwin, H, 1975b History of the natural forests of Britain: establishment, dominance and destruction, Phil Trans Roy Soc London B 271, 47–67 Godwin, H and Newton, L, 1938 The submerged forest at Borth and Ynyslas, Cardiganshire. New Phytol 37, 333–44 Greig, J R A, 1982 Past and present lime woods of Europe, in Bell, M and Limbrey, S (eds), Archaeological Aspects of Woodland ecology. Oxford: Brit Archaeol Rep S146, 23–55 Greig J R A, 1991 The prehistoric riverside settlement at Runnymede: the botanical story, in Needham S and Longley D, Excavation and Salvage at Runnymede Bridge, Berkshire, 1978. London: British Museum, 134–212 Grichuk, M P, 1967 The study of pollen spectra from recent and ancient alluvium. Rev Palaeobotany Palynology 4, 107–12 Groot, J J. 1966 Some observations on pollen grains in suspension in the estuary of the Delaware River. Marine Geol 4, 409–16 Hall, S A, 1981 Deteriorated pollen grains and the interpretation of Quaternary pollen diagrams. Rev Palaeobotany Palynology 32, 193–206 Haslett, S K, Davies, P, and Strawbridge F, 1997 Reconstructing Holocene sea-level change in the Severn Estuary and Somerset Levels: the foraminifer connection. Archaeol Severn Estuary 8, 29–40 Haslett, S, Davies, P, Curr, R H F, Davies, C F C, Kennington, K, King, C P, and Margetts, A J, 1998 Evaluating Holocene Late Holocene relative sea-level change in the Somerset Levels, southwest Britain. Holocene 8, 197–207 Hawkins, A B, 1971 The Late-Weichselian and Flandrian transgression of South-West Britain. Quaternaria 14, 115–30 Hawkins, A B, 1973 Sea level changes around South-West England, in Blackman, D J, Marine Archaeology. London: Butterworths, 67–87 Hewlett R and Birnie J, 1996 Holocene environmental change in the inner Severn Estuary UK; an example of the response of estuarine sedimentation to relative sea level change. Holocene 6, 49–61 Heyworth, A, and Kidson, C, 1982 Sea-level changes in southwest England and Wales. Proc Geol Assoc 93, 91–111 Hibbert, F A, 1978 The vegetational history of the Somerset Levels, in Limbrey, S and Evans J G, (eds), The Effect of Man on the Landscape: the Lowland Zone. London: Counc Brit Archaeol Res Rep 21, 90–5 Hibbert, F A, 1980 Possible evidence for sea-level change in the Somerset Levels, in Thompson, F H, (ed), Archaeology and Coastal Change. London: Soc Antiq London Occas Pap NS 1, 103–5 Hustedt, F, 1957 Die Diatomeenflora des Fluss-systems der Weser im Gebiet der Hansestadt Bremen. Ab. naturw. Ver. Bremen 34, 181–440 Janssen, C R, 1969 Alnus as a disturbing factor in pollen diagrams. Acta Bot. Neere. 8, 55–8 Janus, H, 1979 The Young Specialist Looks at Land and Freshwater Molluscs. London: Bourke Jardine, W G, 1976 Some problems in plotting the mean surface level of the North Sea and the Irish Sea during the last 15000 years. Geol. Foreningens 98, 78–87 Jelgersma, S, 1961 Holocene sea-level changes in the Netherlands. Meded geol. Sticht ser. (6)7, 1–100 Jelgersma, S, 1966 Sea level changes during the last 10,000 years, in Sawyer, J S, (ed), World climate 8000–0 BC. Roy Met Soc, 54–71 Jennings, S, Carter, R W Gm and Orford, J D, 1993 Late-Holocene salt-marsh development under a regime of rapid relative sea level rise: Chezzetcook Inlet, Nova Scotia. Implications for the interpretation of palaeo-marsh sequences. Canadian J Earth Sci 30, 1374–84 Juggins, S, 1982 The Geoarchaeology of Hallen Marsh. Unpubl undergrad dissert, Univ Bristol Kerney, M P, 1999 Atlas of the Land and Freshwater Molluscs of Britain and Ireland. Colchester: Conch Soc Grt Britain Ireland Kidson, C and Heyworth, A , 1973 The Flandrian sea-level rise in the Bristol Channel. Proc Ussher Soc 2, 265–84 Kidson, C and Heyworth, A, 1976 The Quaternary deposits of the Somerset Levels. Quart J Engin Geol 9, 217–35 Kidson, C and Heyworth, A, 1978 Holocene eustatic sea level change. Nature 272, 748–50 Kloet, G S and Hincks, W D, 1977 A Check List of British insects, (2nd edn, rev): Coleoptera and Strepsipter, London: Roy Entomol Soc Handbk Ident Brit Insects 11(3) Lawler, M, Parkhouse, J and Straker, V, (eds), 1992 Archaeology of the Second Severn Crossing: assessment for recommendations for the English Approaches. Swansea: Glamorgan-Gwent Archaeol Trust Lindroth, C H, 1974 Coleoptera: Carabidae. London: Roy Entomol Soc Handbk Ident Brit Insects 4(2)

114 Locock, M, 1997a Rockingham Farm, Avonmouth: mouted enclosures on the North Avon Levels. Archaeology in the Severn Estuary 8, 83–8 Locock, M, 1997b A prehistoric trackway at Cold Harbour Pill, Redwick, Gwent. Archaeol Severn Estuary 8, 9–12 Locock, M, 1999 Buried soils of the Wentlooge Formation. Archaeol Severn Estuary 10, 1–10 Locock, M, 2000 A Later Bronze Age landscape on the Avon Levels: settlement, shelters and saltmarsh at Cabot Park, in Brück, J, (ed), Landscape and Settlement in Bronze Age Britain. Oxford: Oxbow, 121–8 Locock, M, and Lawler, M, 2000 Moated enclosures on the North Avon Level: survey and excavation at Rockinhgham Farm, Avonmouth, 1993–7. Trans Bristol Gloucestershire Archaeol Soc 118, 93–122 Locock, M, Robinson, S, and Yates, A, 1998 Late Bronze Age sites at Cabot Park, Avonmouth. Archaeol Severn Estuary 9, 31–6 Long, A J, Scaife, R G, and Edwards, R J, 1999 Pine pollen in intertidal sediments from Poole Harbour, UK; implications for late-Holocene sediment accretion rates and sea-level rise. Quat Inter 55, 3–16 Long, A J, Scaife, R G, and Edwards, R J, 2000 Stratigraphic architecture, relative sea-level, and models of estuary development in southern England: new data from Southampton Water, in Pye, K, and Allen J R L (eds), Coastal and Estuarine Environments: sedimentology geomorphology and geoarchaeology. London: Geol Soc London Spec Publ 175, 253–79 McVean, D N, 1953 Biological flora of the British Isles. Alnus glutinosa (L) Gaertn. J Ecol 41, 447–66 McVean, D N, 1956 Ecology of Alnus glutinosa ( L) Gaertn. VI Post-glacial history. J Ecol 44, 331–3 Macan, T, 1977 A Key to the British Fresh- and Brackish-water Gastropods. Kendall: Freshwater Biol Assoc Sci Publ 13 (4th edn) Moore, P D, 1977 Ancient distribution of lime trees in Britain. Nature 268, 13–14 Moore, P D and Webb, J A, 1978 An Illustrated Guide to Pollen Analysis. London: Hodder & Stoughton Moore, P D, Webb, J A and Collinson, M E, 1991 Pollen Analysis (2nd edn). Oxford: Blackwell Peck, R M, 1973 Pollen budget studies in a small Yorkshire catchment, in Birks, H J B and West, R G (eds), Quaternary Plant Ecology. Oxford: Blackwell, 43–60 Porter, D, 1990 The Archaeology of the Second Severn Crossing (preliminary survey). Unpubl, Avon County Planning Dept Pye, K, and Allen, J R L, 2000 Past, present and future interactions, management challenges and research needs in coastal and estuarine environments, in Pye, K, and Allen J R L (eds), Coastal and Estuarine Environments: sedimentology geomorphology and geoarchaeology. London: Geol Soc London Spec Publ 175, 1–4 Riley, H, 1998 Intertidal survey at Avon mouth and Oldbury-on-Severn 1998. Archaeol Severn Estuary 9, 79–80 Reid, C, 1913 Submerged Forests. Cambridge: Univ Press Rippon, S, 1991 The Somerset Levels in the Roman period. Ann RepSevern Estuary Levels Res Comm, 43–6 Rippon, S, 1994 The evolution of a historic wetland landscape: the Gwent Levels historic landscape study. Archaeol Severn Estuary, 21–34 Rippon, S, 1996 Gwent Levels: the evolution of a wetland landscape. York: Counc Brit Archaeol Res Rep 105 Rippon, S, 1997 The Severn Estuary; landscape evolution and wetland reclamation. Leicester: Univ Press Russett, V, 1990 The Archaeology of the Second Severn Crossing (observations and geotechnical trial pits). Unpubl. Avon County Planning Dept Scaife, R G, 1980 Late-Devensian and Flandrian Palaeoecological Studies in the Isle of Wight. Unpubl PhD thesis, Univ London Scaife, R G, 1988 The elm decline in the pollen record of south east England and its relationship to early agriculture, in Jones, M, (ed), Archaeology and the Flora of the British Isles. Oxford: Oxford Univ Comm Archaeol, 21–33 Scaife, R G, 1993 The palynological investigation of the peats and sediments, in Godbold, S and Turner, R C, Second Severn Crossing. Archaeological response Phase 1–The intertidal zone in Wales. Cardiff: Cadw, 51–9 Scaife, R G, 1994 Pollen analysis and radiocarbon dating of the intertidal peats at Caldicot Pill. Archaeol Severn Estuary, 67–80 Scaife, R G, 2000 Holocene vegetation development in London, in Sidell et al 2000, 111–17 Scaife, R G 2001 Pollen Analysis of Peat and MineralSsediments at Motney (W4358), Gravesend (W1847) and Queenborough (W1872), North Kent. Unpubl rep for Wessex Archaeol Scaife, R G and Burrin, P J, 1992 Archaeological inferences from alluvial sediments: some findings from southern England, in Needham, S, and Macklin, M G, (eds), Alluvial Archaeology in Britain. Oxford: Oxbow Monogr 27, 75–91 Scaife, R G, and Long, A, 1994 Evidence for Holocene sea-level changes at Caldicote Pill. Archaeol Severn Estuary, 81–5 Shennan, I, 1982 Interpretation of Holocene sea-level data from the Fenland, England. Proc Geol Assoc 93, 53–63 Shennan, I, 1989 Late Quaternary sea-level changes and crustal movements in the British Isles. J Quat Sci 4, 1–94 Sidell, J, Wilkinson, K, Scaife, R G, and Cameron, N, 2000 The Holocene Evolution of the London Thames, Archaeological Excavations (1991–1998). London: Mus London Archaeol Serv Monogr 5 Smith, A G, 1970 The influence of Mesolithic and Neolithic man on British vegetation: a discussion, in Walker, D and West, R G (ed), Studies in the Vegetational History of the British Isles. Cambridge: Univ Press, 81–96

115 Smith, A G and Morgan, L A , 1989 A succession to ombrotrophic bog in the Gwent Levels, and its demise: a Welsh parallel to the peats of the Somerset Levels. New Phytol 112, 145–67 Smith, A G and Pilcher, J R, 1973 Radiocarbon dates and the vegetational history of the British Isles. New Phytol. 72, 903–14 Stace, C, 1997 New Flora of the British Isles (2nd edn). Cambridge: Univ Press, Stockmarr, J, 1971 Tablets with spores used in absolute pollen analysis. Pollen et Spores 13, 614–21 Tansley, A G, 1949 The British Isles and their Vegetation. London: Cambridge Univ Press Tauber, H, 1965 Differential pollen dispersion and the interpretation of pollen diagrams. Danm. Geol. Unders. II 89, 1– 69 Tauber, H, 1967 Investigation of the mode of pollen transfer in forested areas. Rev Palaeobot Palynol 3, 277–87 Tomalin, D J, Loader, R, and Scaife R G, forthcoming Wootton Haven: Coastal and port archaeology in a dynamic envirnment. Oxford: Brit Archaeol Rep Turner, J, 1962 The Tilia decline: an anthropogenic interpretation. New Phytol 61, 328–41 Tutin, T.G., Heywood, V.H. et al 1964–80 Flora Europaea 1–5, Cambridge: Univ Press Vos, P C and Wolf, H, de 1993 Diatoms as a tool for reconstructing sedimentary environments in coastal wetlands: methodological aspects. Hydrobiologia 269/70, 285–96 Waller, M, 1993 Flandrian vegetational history of south-eastern England. Pollen data from Pannel Bridge, East Sussex. New Phytol 124, 345–69 Waller, M, 1994 The Fenland Project, No 9: Flandrian Environmental Change in Fenland. Cambridge: E Anglian Archaeol 70 Walker, M J C, Bell, M, Caseldine, A E, Cameron, N G, Hunter, K L, James, J H and Smith, D N, 1998a Palaeoecological investigations of middle and late Flandrian buried peats on the Caldicot Levels, Severn Estuary, Wales. Proc Geol Assoc 109, 51–78 Walker, M J C, Caseldine, A E, Barrow, K, James, J H, Cameron, N G, and Kreiser, A, 1998b Palaeoenvironmental Assessment of Late Flandrian Sediments from Rockingham Farm, Avonmouth, near Bristol. Unpubl Rep, Palaeoenviron Res Centre, Univ Wales, Lampeter Walker, M J C, Caseldine, A.E., Crowther, J, Johnson, S, James, and Macphail, R I, 1999a, PalaeoenvironmentalAassessment of Late-Flandrian Sediments from Cabot Park, Avonmouth, near Bristol. Unpubl Rep Palaeoenviron Res Centre, Univ Wales, Lampeter Walker, M J C, Caseldine, A E, Barrow, K, Cameron, N G, James, J H, Kreiser, A, and Macphail, R I, 1999b Palaeoenvironmental Assessment of Late-Flandrian Sediments from Kites Corner, Avonmouth, near Bristol. Unpubl Rep Palaeoenviron Res Centre, Univ Wales, Lampeter Waton, P V, and Barber, K E, 1987 Rimsmoor, Dorset; biostratigraphy and chronology of an infilled doline, in Barber, K E, (ed), Wessex and the Isle of Wight, Field Guide. Cambridge: Quat Res Group, 75–80 Wessex Archaeology, 1994 The Second Severn Crossing: English Approaches Archaeological Survey and Landscape Analysis: Post-excavation assessment. Unpubl Client rep 35745 Wessex Archaeology, 2000 The Second Severn Crossing English Approaches Archaeological Post-Excavation Programme. Unpubl client rep T5512 Whittle, A W R, 1978 Resources and population in the British Neolithic. Antiquity 52, 34–42 Whittle, A, 1989 Two later Bronze Age occupations and an Iron Age channel on the Gwent foreshore. Bull Celtic Stud 36, 200–23 Wilson, D, Davies, J R, Smith, M and Waters, R A, 1988 Structural controls on Upper Palaeozoic sedimentation in south-east Wales. J Geol Soc London, 145, 901–14 Woodworth, P L, 1987 Trends in U K mean sea level, Marine Geol 11, 57–87

116 Wessex Archaeology Internet Report

Registered Head Office: Portway House Old Sarum Park Salisbury, Wiltshire SP4 6EB

Tel: 01722 326867 [email protected] www.wessexarch.co.uk

Registered Charity No. 287786 A company with limited liability registered in England No. 1712772