Over the past 30 years, humans have largely been held responsible for the major changes in vegetation in the past. Thus the `elm decline' in north-west Europe around 5,000 years ago was thought partly the result of selective lopping of elm-branches for animal fodder. The northward and westward expansion of beech, and the westward expansion of spruce in the last 5,000 years, were attributed to soil impoverishment following early agriculture. The widespread development of blanket peat in the British uplands was put down to the same cause.

This emphasis on human impact arose for two reasons. First, it was assumed the postglacial climate has been essentially unchanging (apart from such `minor' fluctuations as the Little Ice Age of the 15th-19th centuries AD). Second, the pollen record documents numerous coincidences between evidence for human activity and past vegetation change. Both assumptions, however, are now under question.

It is now recognised that the postglacial climate has been not only changable but highly complex with different changes taking place in different regions. For instance, the warm-summer period of c 7000-5000BP, once thought a global event, can now not even be recognised throughout Europe. Moreover, changes in the Earth's orbit around the sun, which strongly influenced climate changes in the Quaternary (or glacial) period, are now known to be affecting the postglacial climate as well.

Moving on to the apparent coincidences between human activity and past vegetation change, much of the evidence is no more than circumstantial. In north-west Europe, the `elm decline' coincides often with the first evidence of cereal cultivation but not always. In some areas agricultural activity long precedes it. In eastern North America a `hemlock decline' is recorded during the mid-postglacial, linked not to human activity but to attack by a pathogen. Why not the `elm decline' too?

As for beech and spruce, in many places their first appearance does indeed follow a disturbance associated with human activity and the presence of cereal pollen; but ecologists know that forest-disturbance often acts as a `proximal trigger' - or immediate cause - for changes in plant-composition, whose ultimate origin lies in changes in the broader environment. Human activity is not the only type of proximal trigger: others include forest fire, wind-storm and even attack by defoliating insects. The ultimate causes of the spread of beech and spruce were changes in seasonal temperatures and the distribution of rainfall.

What of the development of upland peat? Peat develops as a result of soil waterlogging; particularly if the climate is cool and wet, and especially if `podsolisation' has occurred - that is, when the soil is impoverished of nutrients at the surface. Pollen evidence suggests that in Britain the early postglacial was characterised by warmer summers and a consequent reduction in soil waterlogging (especially in the west and north), but that in the last 5,000 years the climate has progressively cooled, leading to more waterlogging. Moreover, podsols developed in the British uplands as a result of the extensive cover of glacial deposits derived from acidic rocks. Although human forest disturbance may sometimes have acted as a proximal trigger for blanket peat development, once again the ultimate cause lies in changes in the climate.

The expansion of Neolithic agriculture across Europe took place at a time of warm summers and mild winters. Subsequently, in northern Europe, there have been 5,000 years of gradual deterioration in conditions for cereal cultivation. Superimposed on this trend have been warmer/colder fluctuations (such as the Little Ice Age), some of which may have favoured phases of colonisation of marginal areas followed by abandonment. As we refine our knowledge of postglacial climate, we may begin to understand that macro-historical changes have frequently been responses to the changing environment; and that it is only in the last two centuries that humans have exerted a large enough impact to overwhelm natural patterns of environmental change.

Dr Brian Huntley is a Senior Lecturer in Biological Sciences at the University of Durham

For centuries, geographers and historians working in the Mediterranean have pointed to the evidence for significant landscape change in the past. The ruins of once great cities stood in the midst of impoverished countryside. Rivers known to have been navigable in classical or medieval times were now silted up. What had caused such changes?

Once modern geological reasoning replaced biblical notions of great floods and deluges, the debate has centred on two major agencies of landscape change: climatic change, and human actions such as forest-clearance and the herding of animals.

In 1969, Claudio Vita-Finzi provided the first scientific investigation of the scale and nature of Mediterranean landscape change, with a clearly-argued model of causation. In fieldwork throughout the Mediterranean basin, he found evidence for a major phase of river sedimentation which he dated to Roman and perhaps medieval times (from its relationship with buried archaeological features such as Roman dams at Lepcis Magna, and from the inclusion of Roman and medieval potsherds). Given how widespread it was, and apparently of similar age throughout its extent, he concluded it was explained by climatic change - a period of cooler, wetter weather - rather than by human actions.

Since then, however, other geomorphologists have found evidence for earlier and later episodes of alluviation, and have tended to explain them in terms of human impact on the environment rather than climatic change. A similar debate has been developing amongst pollen analysts on how best to explain evidence for changes in vegetation.

The problem with the debate is that geographers have collected evidence for environmental change, and archaeologists and historians have collected evidence for settlement trends and systems of land use, but rarely in the same area and hardly ever working together as a team. As a result, one discipline's set of data can't be tested against those of the other two.

To advance the debate, we need integrated methodologies linking geomorphology, archaeology and history. An example of this kind of approach is a study of the Biferno valley, east of Rome in central-southern Italy, published this month as A Mediterranean Valley: Landscape Archaeology and Annales History in the Biferno Valley (Leicester). In this project, I co-ordinated a team of archaeologists, historians and geographers, working together to reconstruct the environmental and human prehistory and history of the valley.

What we found was that the main episodes of prehistoric settlement all coincided with small-scale clearance. Then, in the pre-Roman Samnite period (c 500-80BC), when the entire valley was densely populated and intensively farmed, there was a predominantly open landscape and massively accelerated erosion. Following the Roman period, population numbers fell dramatically, with a return to small-scale clearances.

Archaeological and historical data show that population levels have fluctuated considerably over the past 1,000 years, and that every episode of population expansion coincides with increasing clearance and erosion. The evidence for human impact is increasingly dramatic: substantial in the 19th century, when huge numbers of the valley's population emigrated to America because conditions were so bad; larger still in the 1930s, when Mussolini made the peasants cultivate huge areas of unsuitable land for cereals; and on a colossal scale in recent decades, as the farmers have abandoned their traditional light ploughs for heavy ploughs, turning the valley from a mosaic landscape to an open prairie.

The evidence therefore seems clear that most of the changes to the valley's landscape over the past 5,000 years have been caused by people rather than by climate. The climate has not been stable through the period, and it still remains likely that a wetter phase accelerated erosion 1,000 or so years ago, as Vita-Finzi argued; but the reason for its dramatic effect was because the landscape then was intensively settled and cultivated. What is particularly frightening is that, whilst the Samnite and modern erosional sediments are equal in scale, the former is the record of three or four centuries of intensive farming, whereas the latter is the record of that of just two or three decades.

Graeme Barker is Professor of Archaeology at the University of Leicester

Everyone knows that the settlements of modern England vary from one district to another. In some, all the houses are gathered together in nucleated villages, some of which have regular plans, while in others they are scattered along lanes and roads, either in small groups or as isolated farms.

These differences reflect one of the great revolutions of English (and indeed European) social history - the formation of villages in the early medieval period, out of the largely scattered settlement pattern that had characterised the landscape since farming and settlement began. Yet exactly when and why the revolution took place, and why it did not happen in all areas - these questions have troubled archaeologists, geographers and historians for decades.

Nucleated villages tend to predominate in England's midland belt - from Durham in the north-east to Dorset in the south. By contrast, in the far south-west, the west and north-west, and in the south-east, a dispersed pattern is more common. Yet this broad distinction, between the Midlands and elsewhere, conceals a myriad of differences from one district to another, producing an overall pattern of such variety that it defeats all simplistic, single-cause explanations of nucleation such as population pressure, political will, the influence of landscape, or the growth of trade.

Research at Birmingham University, funded since 1991 by the Leverhulme Trust, suggests that a complex set of causes is needed to explain the origin of villages in England. The requirements of farming, the influence of lords, and the rise of marketing all played some role, as did the tendency of people to imitate models of the shape of settlements.

The research, carried out by Patrick Mitchell-Fox and Carenza Lewis (on secondment from the English Royal Commission), and led by me, involved an overview of the settlements and landscapes of the four east-midland counties of Bedfordshire, Buckinghamshire, Leicestershire and Northamptonshire.

These counties were chosen because they contain a variety of settlement and land-use; they also contain numerous surviving sites of abandoned villages - providing examples of medieval settlement unsullied by post-medieval and modern alterations; they have good archaeological data such as a datable sequence of pottery from the pre-Conquest period; and they are well provided with historical documents.

We began by looking at the maps compiled in the early 19th century, which provide the earliest systematic record of settlements over the whole area. We classified the different settlement forms as nucleated clusters and nucleated rows (villages laid out on a grid, or in a single street), interrupted or irregular rows, settlements on the edge of common land (colonising settlements), hamlets (groups of fewer than six houses) and farms. The abandoned villages of the area suggested the settlements of each locality in the 19th century still reflected the forms that had been laid out in the Middle Ages.

The area's settlement forms fall into clear zones. Much of Leicestershire, especially in the east, and most of Northamptonshire, are dominated by nucleated villages, either of the cluster or the row type. By contrast, in the west of Leicestershire, in Charnwood, and in the south of the region in the Chiltern Hills, most of the settlements are dispersed. In north Bedfordshire a mixture of forms is found.

The date of the first formation of villages has always posed problems. There seems to be a gap between the abandonment of the pre-nucleated settlements, which often seem to have gone out of use in the 9th century, and the earliest phases of the nucleated villages, which were rarely earlier than the 12th century.

Our research may help to close the gap. By scouring the region's Sites and Monuments Records for stray finds of 10th and 11th century pottery, we found that 80-90 per cent of the finds came from, or near, the sites of nucleated settlements, suggesting that many villages had begun to form by then.

Another obscure, and controversial, question is that of who made the decisions. Some argue that it was the lord who decreed that people should live in a single village, while others suggest that peasant communities themselves organised the new settlements. We have only one detailed contemporary document from our region that might shed light on the matter - a legal record describing the late 12th century reorganisation of the fields at Segenhoe in Bedfordshire.

There, two local lords took the initiative in making a redistribution of land at a time of uncertainty about ownership following the anarchy of King Stephen's reign. The lords, however, gained the consent of their tenants, who surrendered their lands to the judgement of six old men - old peasants who went around the village territory measuring new boundaries and assigning them to each tenant.

This process of land-redistribution would have been necessary whenever a village was created out of dispersed settlement; and Segenhoe provides one example of its being achieved with the participation of lords and peasants. No doubt there were other similar cases.

But why were villages formed? None of the single-cause answers offered in the past hold water. Take population pressure, for instance. It has been argued that as an area's population grew, room could only be found for grazing animals by a general agreement to share resources and leave a proportion of the land fallow each year - in other words, to create the classic midland two or three-field system, which is almost invariably associated with nucleated settlements. But the population pressure was not always very great, as in Rockingham Forest in Northamptonshire, where a thin distribution of population did not prevent the formation of several nucleated villages.

Nor can you say, as many have, that nucleation was just a matter of lords' policies, because you find different types of settlement on the manors of the same type of lord, or even the same lord. Ramsey Abbey, at Ramsey in Huntingdonshire, for instance, has nucleated villages on some parts of its estate (such as Broughton and Abbots Ripton near Huntingdon), and dispersed settlements on others (as at Cranfield in Bedfordshire).

Nor can it be said that the decision to adopt different settlement patterns was a simple reflection of natural conditions such as soils and geology because again we find different types of settlement on the same types of soil.

Our own interpretation takes into account a great variety of factors. It is likely that land came into more intensive use in the 9th century, with a growing competition for resources leading to a bewildering palimpsest of boundaries and property rights. The association of nucleated villages with open fields (in contrast to the patchwork of enclosed fields associated with dispersed settlement) suggests that the desire for a reorganisation of farmland may have played a role in some areas, particularly those whose soils were suitable for the arable cultivation found in open fields.

Moreover, with the growth of the state from the late 9th century, and the multiplication of the number of lesser aristocrats each with his own area to rule, there came an increased demand for taxes, rents and labour. Peasants may often have been forced to organise their farming - and their settlements along with it - more effectively.

By building manor houses and churches, local lords also provided a core around which peasant settlers could congregate, either voluntarily or by force - as in the case of the lord's slaves, who were often given parcels of demesne land on which to build their cottages.

The development of market opportunities also had an impact on nucleation. By the 12th century some villages were provided with market places, perhaps attracting increased numbers of settlers in some areas, while `market villages' were founded in areas of dispersed settlement.

How, then, did villages form? Was it, in each case, a single traumatic event, in which people gave up their farms and hamlets and moved to a central village in a co-ordinated move, perhaps in October after the harvest? Or did it happen gradually, over decades or even centuries? At Harlestone in Northamptonshire, where the fields formed a regular pattern but the village is more straggling, we sense that the village has been frozen during its incomplete nucleation. But only a future campaign of excavation may resolve this question.

Christopher Dyer is Professor of Medieval Social History at the University of Birmingham