Ancient woodlands: modern threats


Author for correspondence:
Oliver Rackham
Tel: +44 1223 360144

I. Introduction

The term ‘ancient woodland’ originated in England, where for over 1000 yr trees have occurred not as continuous tracts of forest but as isolated wood-lots, typically of less than a square kilometre, forming islands among farmland, grassland, and heath. Some of these have arisen at various dates on former agricultural land, heathland, industrial wasteland, or deserted settlements: how this happens was the subject of classic successional studies such as those by A. S. Watt (1925, 1934). Others have existed for hundreds of years: these constitute ancient woodland (e.g. Rackham, 1975, 1990, 2000; Thomas, 1998).

The term is not equivalent to ‘virgin forest’ or ‘wildwood’, in the sense of ecosystems that have escaped human intervention: English ancient woods have all been managed and exploited for centuries or millennia (Rackham, 1977), but they are natural vegetation and are not artefacts. Natural processes, here as all over the world, interact with human activities in complex and subtle ways (Peterken, 1996). The relation between ancient woods and early Holocene wildwood is not straightforward: many ancient woods, on archaeological evidence, were open land in Roman times (c. 1800 yr ago) or earlier.

Ancient woods are valued as being among the most complex and diverse of ecosystems, and as having plants and animals that do not occur in woods of recent origin (Peterken, 1993). These features depend partly on continuity (some may be inherited from wildwood times) and partly on interaction with human activities such as woodcutting and management of herbivores. The value of ancient woods is not only ecological: social, spiritual, archaeological, and economic values need to be considered, although they are not dealt with in this article. Timber and game production may also be involved, and sometimes conflict with the other values (Rackham, 2003, Chapter 33). The identification, protection, and management of ancient woodland was a conservation objective in the 1980s and 1990s. The then Nature Conservancy published 1 : 50 000 maps of England and Wales identifying woodland that had existed since AD 1600 (Kirby et al., 1998).

Ancient woodland tends to be characterized by the following.

  • 1A regime of frequent felling and regrowth, often termed ‘coppicing’. This exploits the natural property of many trees to regenerate by stump sprouts or root suckers – a property inherited from their evolutionary past, long before people invented axes.
  • 2Isolation is not the result of recent fragmentation (as it often is in countries other than England) but has been a feature of woodland for long enough to become an integral feature of the woodland ecosystem.
  • 3In England there are often many tree communities, each of a few species (limewood (Tilia), hornbeam-wood (Carpinus), etc.), forming a mosaic within each wood-lot (Peterken, 1993; Rackham, 1992, 2003). Eastern North America tends to have fewer tree communities each of many species.
  • 4The flora contains characteristic ancient-woodland plants, which do not easily migrate from wood to wood (e.g. Peterken & Game, 1984; Bossuyt et al., 1999; Rackham, 2006, Chapter 12) which tend to be either clonal plants (e.g. in England Anemone nemorosa) which do not often reproduce by seed, or ant-dispersed (e.g. Melica uniflora).
  • 5Besides shade-bearing plants, there is also a large component of coppicing plants, not adapted to continuous shade, which flower in abundance every time the wood is cut down (e.g. Primula elatior) or germinate from buried seed at each felling (e.g. Euphorbia amygdaloides). (Some coppicing plants, including these two, are also ancient-woodland plants (Rackham, 2006).)
  • 6There are permanent open areas with woodland-grassland plants.

Medieval England had two kinds of wild tree-land: woodland, meaning islands of forest, and wood-pasture, land that combined trees with grazing animals, often in the form of scattered trees among grassland (Rackham, 1989, 1990). Similar distinctions between forest and treed grassland (often called savanna) run through many parts of the world from Finland (Hæggström, 1998) to Tasmania and from California to Japan (cf. Manning et al., 2006). In savanna some environmental, zoological or cultural factor allows trees to grow but not forests; it is often difficult to distinguish ‘natural’ from cultural savannas (Grove & Rackham, 2001, Chapter 12).

These properties, illustrated by ancient woods in England, apply in varying degrees to ancient woods elsewhere; but in other countries one encounters both differences on the ground and different uses of the terms. The term ‘woodland’ has been taken to other parts of the world and applied to other ecosystems: Americans and Australians use it for the denser kinds of savanna. The definition used here is that in woodland most of the ground vegetation consists of shade-tolerant or shade-evading plants, whereas in savanna the plants between the trees are not shade-adapted.

Other countries also have a contrast between ancient and recent woodland (e.g. for Brazil, see Barlow et al., 2007). Recent forest, often overlying hedges, walls, terraces, and other cultivation remains, covers huge tracts in Europe and North America, and is abundant even in Australia and Japan. It differs from ancient woodland, partly for reasons related to succession and colonization, and partly because a period as farmland or heath has permanently altered the soil characteristics (Rackham, 2006, Chapter 12). Recent woodland may retain some of the species of the previous nonwoodland vegetation; it lacks forest species which either are poorly dispersed or do not readily colonize.

Ancient woodland does not necessarily involve ancient trees. In England it seldom contains upstanding ancient trees, although often there are ancient coppice stools, multi-stemmed trees whose massive bases result from centuries of felling and regrowth. These resemble the lignotuberous bases of some American savanna trees and some Australian eucalypts, resulting from centuries of fire and regrowth. Many of the world's upstanding ancient trees are not in forests but in wood-pastures and savannas.

This article is founded on the large amount of information that is available for England, but goes on to consider threats in a world-wide context. I distinguish among past threats, no longer going on, continuing threats, and potential threats, which may arise in the future. The reader is warned against expecting that the future will be merely an extrapolation of the recent past.

II. What is meant by threats?

This paper is about ancient woodland in the wide sense of areas of forest and savanna that have been in existence on the same sites for several hundred years. They are not necessarily static: they have their own dynamics, driven by such influences as fire, windblow, or the longevity of individual species.

The normal dynamics of many forests and savannas include long-established human activities. In most countries actions such as periodic felling (as in England) or periodic burning (as in aboriginal Australia) have become part of the normal dynamics that maintain their ecosystems. Ancient woods are threatened by withdrawal of ‘traditional’ activities as well as by introduction of new activities. Their conservation should not be confused with the restoration of wildwood, nor with the future development of plantation forestry.

Threats are continuing or expected disturbances to ecosystems that interfere with normal dynamics. Thus windblow is not a threat, but is a repeated (though rare) event to which ecosystems respond (e.g. Burslem et al., 2000; Batista & Platt, 2003). Dutch elm disease, produced in European and North American Ulmus species by the fungi Ophiostoma ulmi and Ophiostoma novo-ulmi, is a marginal example. In Europe it is probably a normal part of the dynamics of elms, going back to the Elm Decline at the start of the Neolithic period (Rackham, 2006, Chapters 5 and 17); however, its effects have been aggravated by people moving various strains of the fungi around the globe.

Not all changes can be interpreted as threats. Thus the recent practices among English conservationists of leaving rotten logs and of burning surplus small wood, although not a historical part of coppicing, add habitats for deadwood organisms and pyrophytes without, as far as can be seen, taking anything away from the normal state of the woodland. (Deadwood insects are one of the chief features in which wildwood is known to have differed from existing ancient woodland, and to a lesser extent from wood-pasture (Whitehouse, 2006).) However, the modern proliferation of deer demonstrably takes away features without, as far as is known, adding anything.

A few threats arise from natural causes. The disappearance of pine forests in the western Scottish Highlands cannot be related to human activities such as tree-felling (forests known to have been logged tend to be the ones that still remain) and is probably a stage in the natural succession of the Holocene, forests being sustainable for only a few thousand years in a land of high rainfall and leached soils (Smout et al., 2005).

III. Destruction and fragmentation

Destruction takes two forms: digging up woods to make farmland, and felling, grubbing and poisoning woods to make forestry plantations. It happens episodically: a great ecological tragedy of the 19th century was the remarkably thorough destruction of savannas east of the Mississippi in America. In the third quarter of the 20th century c. 40% of the ancient woodland in England was converted to farmland or forestry. It was destroyed in vain: the need for more farmland was undercut by the success of plant breeders in growing more crops per hectare. Plantations on former woodland proved to be difficult to establish, and uneconomic because of competition from other countries. In the last quarter of the century an even greater disaster overtook the temperate forests of middle Chile: within 25 yr ‘native forest’ had been reduced from 21% of the land area to 7%, and ‘plantation’ increased from 5% to 36% (Echeverria et al., 2006).

Destruction may be amplified by fragmentation. If half of a 100-ha wood is grubbed out, this results in loss of species that were confined to the part destroyed, but may also result in loss of those for which 50 ha is insufficient area to sustain a viable population; if the part remaining is in the form of three fragments of 17 ha, further species may be lost. Fragmentation is often the result of recent or ongoing human encroachment: thus some of the wood-lots of Belgium (Honnay et al., 2005) or Hungary (Molnár, 1998), or the heaths of England, can be shown from historical records to be fragments of once-continuous forest or heath. However, in England woods are more in the nature of islands among farmland: many have been islands for more than 1000 yr; indeed, the Vera theory of wildwood (Vera, 2002) casts doubt on whether continuous forest ever existed in the Holocene.

Fragmentation and insularization have been studied mainly by zoologists: thus many birds or mammals require a certain area of woodland to sustain a territory (e.g. Moore & Hooper, 1975). Browsing effects may be intensified on islands in tropical artificial lakes too small for herbivorous mammals to have predators (Terborgh et al., 2006). However, direct effects on plants, although they should exist in theory (Honnay et al., 2005), are more difficult to demonstrate – perhaps because many woodland plants are long-lived perennials or clonal, with little need for seed reproduction. Even in Canada, where fragmentation occurred only 150 yr previously, survival of tree species is a more complex process than island biogeography would predict (Weaver & Kellman, 1981). In Japan, where expansion of cities continues to reduce the area of surrounding pieces of woodland, fragmentation appears to impoverish the flora of the remaining woods. Part of this, however, may be a consequence of withdrawal of management: ‘traditional’ woodland management has the effect of keeping down the dwarf bamboo (Pleioblastus chino), but in urbanized woods this is usually abandoned, so that the bamboo develops into a vigorous undergrowth, suppressing many herbaceous plants (Iida & Nakashizuka, 1995).

In England there is a general tendency (in the range 0.5–100 ha) for smaller wood-lots to have fewer species, especially of herbaceous plants (Peterken & Game, 1984); trees are less responsive to the size of the wood (Rackham, 2003). However, there is little pattern in the species that occur disproportionately more often in larger woods. Anemone nemorosa, a plant characteristic of ancient woodland which is strongly clonal and rarely reproduces by seed, is uncommmon in woods smaller than 2 ha, whereas Primula elatior, a long-lived perennial (>25 yr) also characteristic of ancient woodland, occurs in yet smaller woods. Buried-seed plants such as Hypericum species, which need to grow from seed after each felling, show a weak avoidance of small woods. Large woods tend to have a wider range of habitats, especially unshaded and disturbed habitats: thus the plants with a definite preference for big woods tend to be either woodland-grassland specialists (e.g. Succisa pratensis) or generalized farmland species (e.g. Potentilla anserina and Plantago major) (Rackham, 2003, Chapter 30). The fragmentation effect, if any, is difficult to detect among the larger losses of species from increasing shade and deer.

Direct destruction is now almost a past threat to ancient woodland in England, but is a continuing threat in many other countries. The belief still prevails that even now large areas of the world are waiting to be made into farmland: a belief that ruined many a settler in the Amazon or Montana. Unfortunately the forests are destroyed in the process of discovering that good farmland does not result. Although foresters in England have learnt the lesson that the last place in which to plant trees is where there are trees already, forests are still destroyed to make plantations in other parts of the world, such as Tasmania. I shall not cite figures for the extent of destruction: statistics, especially official statistics, are bedevilled by problems of shifting definition (where does one draw the line between forest and the well-treed farmland that is common in the tropics?) (Grove & Rackham, 2001, p. 19f).

IV. Depletion

‘Depletion’ means felling some or all of the trees but not digging them up, and letting the land remain as forest. The effects vary hugely. In England nearly all woodland ecosystems have become adapted over centuries to periodic felling and are now threatened by lack of felling: depletion is a normal part of their functioning. A 19th century triumph of technology was the conversion of the giant trees of Sequoia sempervirens in north California into roof-shingles and railway sleepers. However, these are clonal trees and extremely tenacious of life; the effect has been to produce the world's grandest coppice-woods, rings of redwoods, now 50 m high, surrounding the fire-blackened original giant stumps. Whether they will recover the ecosystems related to decay or to long-established tree canopies remains to be seen.

Another kind of depletion is those forms of modern forestry that depend not on plantations but on manipulating natural forests so that they come to consist entirely of trees of the species or form desired at the time of the change. This is less obvious than destruction, but has happened in, for example, India under British rule and since. The larger forests of France and north Switzerland are now often dominated by beech (Fagus sylvatica; Bürgi, 1998) and those of the south-eastern USA by Pinus taeda. Some in England, such as the Forests of Dean and Alice-Holt, reflect a 19th century fashion for Quercus robur (Hart, 1966; Rackham, 2006, Chapter 18); they have lost their previous historic structure and probably much of their biodiversity.

Effects of depletion depend on properties of the particular trees and other plants and animals: for instance, which trees are killed by felling, or whether unfelled woods contain old trees, or whether vigorous grasses or aggressive climbers fill the gaps. It is often claimed that selective logging in tropical forests damages the remaining trees (e.g. Pinard & Cropper, 2000), but whether this tends towards preservation (by rendering them less attractive to future loggers) or to further damage (by killing them) would depend on the properties of the individual species. Responses to various forms of depletion have been the subject of numerous studies, which emphasize their inconsistency and often the conservation value of logged forests (e.g. Laurance & Bierregard, 1997; Barlow et al., 2006; Wells et al., 2007).

V. Pollution and eutrophication

In the 1970s there was great concern about the effects of air and rain pollution on trees and woodlands. This generated a body of science on the effects of acid rain, acid fog, dry deposition, and ozone on epiphytes – especially lichens on savanna and free-standing trees where many rare lichen species and those associated with ecological continuity (e.g. on ancient trees) occur. Since the 18th century sensitive lichens have disappeared from polluted regions; historic lists of the remaining lichens have been used as a proxy for the history of air pollution (e.g. Rackham, 1989, Chapter 11). This is a declining threat: in many countries gross sources of pollution have been eliminated, and sensitive lichens have begun to return (Hawksworth & Henderson, 1974–2000). It remains a continuing threat as low levels of pollution become more widespread and affect previously unaffected areas, for instance in Mongolia (Hauck, 2008).

Wood-lots surrounded by farmland are exposed to drifting dust of fertilizers, agricultural sprays, ammonia from cattle, and the dung of birds that have fed on surrounding farmland. This affects herbaceous plants and possibly mycorrhizal fungi. A main conclusion from an extensive 30-yr survey of British woods was that there has been an increase in plants characteristic of greater fertility (Kirby et al., 2005). Even weak nitrogen pollution has a marked effect on epiphytic bryophytes and lichens in the relatively clean oakwoods of western Britain (Mitchell et al., 2005). However, a local study found that weedkiller effects were significant over only the first few metres inside the wood (Gove et al., 2007).

Gross levels of air pollution can destroy the forest itself: a famous example is the Queenstown Desert in the rainforest of Tasmania, surrounding a copper smelter and now much reduced in the decades since its closure. However, it is clear from, for example, central London or Athens that most trees survive, without obvious damage, urban levels of pollution that eliminate all but pollution-specialized lichens such as Lecanora conizaeoides. The death of areas of forest in Central Europe, and the supposed ill-health of much larger areas, in the 1980s were attributed to acid rain. Public fears that the forests of Germany would disappear had great political repercussions (as had happened before in the 1850s). This Waldsterben phenomenon was attributed to gross air pollution, especially acid fumes from old-fashioned heavy industries in then Communist Europe – although moderately sensitive lichens sometimes flourish on affected trees. It was proposed that acid rain aggravates aluminium toxicity or magnesium deficiency, or inhibits mycorrhizal fungi; yet despite intensive research the causes and even the extent of the phenomenon remain inconclusive (Brüggemeier, 2004).

It is doubtful whether Waldsterben is a continuing threat in Europe. It has not spread, as anticipated, throughout Central Europe; as a killer of trees it seems largely to be confined to c. 100 km2 on the German–Czech border, separated from obvious sources of pollution by many kilometres of living forest. Dense monocultures of Picea excelsa seem to be very susceptible; marginal trees and intermingled broadleaved trees tend to survive. However, death of trees, especially planted conifers, has been reported from other countries such as Italy (Bottacci et al., 1988), and may be attributable to ozone pollution.

In Britain it has been repeatedly claimed that trees, oak and beech as well as exotic conifers, are in poor condition, as based on the density and appearance of their canopies (Innes & Boswell, 1987 and later); however, this is open to the objection that the criteria used are derived from Germany and may not be applicable elsewhere. The claim assumes that the normal state of any population of trees conforms to a preconceived definition of perfect health, but there is no evidence that this assumption was ever fulfilled in the past.

In the USA there are two areas of severe damage to conifers attributed mainly to air pollution. In the eastern mountains, montane forests of Abies balsamea, Abies fraseri and Picea rubens have been damaged or killed, apparently by acid deposition from clouds (Adams & Eager, 1991). In the mountains of California, downwind from Los Angeles and San Francisco, damage to pines is attributed to ozone (Miller et al., 1997).

VI. Fire and lack of fire

Fire itself is not usually a threat, for the ability of plant species to burn is an adaptation, not a misfortune, and most combustible ecosystems can respond to fire. (No English woodland ecosystem is combustible, except for sparse or recently felled oakwoods with fuel from dry Pteridium.) Burning was the commonest form of pre-agricultural land management (Mellars, 1976); it has affected the ecology of entire continents in North America and Australia (Adam, 1994, Chapter 9; Bowman, 1998; Jackson & Brown, 1999). The pattern of ‘natural’ fires, without human intervention, is usually impossible to ascertain.

Changes in the frequency, seasonality and intensity of fire can have profound effects on ecosystems (e.g. Catling, 1991; Burrows et al., 1995). Fire may attack previously fire-resistant woods either after logging, or through invasion by flammable grasses, or because surrounding ex-farmland has become invaded by pines or other flammable vegetation (e.g. Rackham & Moody, 1996, Chapter 10; Grove & Rackham, 2001, Chapter 13).

Suppression, which is the reaction of most European settlers, administrators and conservationists to fire, can have two consequences. Fuel may accumulate over many years, leading to an unsuppressable conflagration, more drastic than many cycles of ordinary fires. Alternatively fire-suppressing species displace the fire-tolerant, so that the fire-adapted ecosystem disappears. Thus the distinctive woods of short-needle pines inherited from pre-settlement times on drier slopes in the mountains of the south-eastern USA (Reilly et al., 2006) are being displaced by less distinctive beechwoods and maple-woods (Harmon, 1982). Australian rain-forests expand into fire-dependent forests and savannas of Eucalyptus spp. (e.g. Russell-Smith et al., 2004; Banjai & Bowman, 2006). In Tasmania the savannas, maintained by Aboriginal burning (Fletcher & Thomas, 2007), were taken over by settlers and made into incombustible pasture-land with scattered big trees with fire-scarred bases. A century and a half on, these great eucalypts, a prized feature of the landscape, are dying, and without fire there seems to be no way of replacing them.

VII. Excessive shade

All the indications are in England that woods are getting more shady and that this is bad for most wildlife. Most woodland plants are not adapted to continuous shade, but either to periodic felling or to permanent open areas, glades and edges.

Coppicing plants of ancient woods in England and many other countries respond in various ways to felling the trees every 4–30 yr. Some are perennials that flower in abundance after felling, such as the spring-leafing Anemone nemorosa or the near-evergreen Viola reichenbachiana. Many are seed-bank plants (Fig. 1), such as Euphorbia and Hypericum species, which vary randomly from one wood-lot to another, suggesting that insular evolution has played a part in the appearance of this ability (Rackham, 2006, Chapter 22).

Figure 1.

Chalkney Wood, Essex, an ancient wood in England, in the second year after felling. The trees (here principally Tilia cordata) are sprouting from the stumps. The sudden increase in light has stimulated the germination of Rubus idaeus and Silene dioica, the characteristic seed-bank plants of this particular wood-lot.

Woodland-grassland plants include perennials such as Succisa pratensis and Stachys officinalis. The permanent open areas where they occur are not a mere artefact of human interference: pollen analysis shows that the wildwood of the early Holocene and of previous interglacials had plants that do not flower in shade (Rackham, 1998).

Most surviving English ancient woods have been more or less neglected for 50–100 yr: there are no longer stages of felling and regrowth, and permanent open areas disappear. This subtracts much of the flora, and many animals, such as nightingale, dormouse, and most woodland butterflies, that depend in various ways on open areas and stages of regrowth (Barkham, 1992; Brown & Warr, 1992; Fuller, 1992; Greatorex-Davies & Marrs, 1992; Gurnell et al., 1992; Warren & Thomas, 1992).

Ancient woods have lost species with time. Recorded losses, based on historic plant lists, indicate that species with a measurable extinction rate do not die out at random through fragmentation effects, but because of increasing shade, irrespective of the size of the wood. For example, Madingley Wood near Cambridge has lost one-third to one-half of its flora since 1660 (Rackham & Coombe, 1996): most are species of wood edges and woodland grassland. If a plant species dies out from one wood its survival in the next wood is likely to be precarious and recolonization improbable.

England, from which these examples come, may be extreme, because throughout history woodland has been scarce and intensively managed and now suffers from lack of management. However, even in Japan with its history of much more forest per head of human population, there are vast areas of disused, excessively shady, deer-bitten coppice-woods.

VIII. Excessive numbers of deer and other ungulates

A world-wide problem with woodland and wood-pasture is increasing numbers of large herbivores, which exceed the historic dynamics of the ecosystem.

In England, more than 1000 yr ago, woodland (with little or no grazing) diverged from wood-pasture (grazed by cattle, sheep, pigs or deer) (Fig. 2). The two native deer – red deer Cervus elaphus and roe deer Capreolus capreolus– became confined to parks and Royal Forests, places provided for them and for fallow deer Dama dama, introduced (ultimately from south-west Asia) c. 1100. Ordinary woodland became adjusted to not having large herbivores. In the 20th century this situation suddenly changed. Red, roe and fallow deer proliferated (despite the appearance of a predator, the motor-car) and spread throughout the countryside; they were joined by exotic species, especially Muntiacus reevesi and Cervus nippon. Many ordinary woods, which in past centuries had no deer, now have more deer per hectare than parks and Forests had in the middle ages. In the 13th century King Henry III consumed, in an average year, less than 1000 deer from all the Royal Forests in England, and complained that there were too few (Rackham, 1986, Chapter 6). Motorists alone are said now to hit c. 50 000 a year (figure based on British Deer Society's 2007 estimate of 74 000 for all Britain), which is not enough to prevent the numbers from rising.

Figure 2.

Ancient wood-pasture, Staverton Park, Suffolk, England. The trees, which appear to be 400–800 yr old, have been pollarded at c. 3 m above ground. Ancient trees such as these, which are characteristic of wood-pastures, are the specific habitat of many of the lichens and invertebrate animals associated with oak. In this site the lichens are now poorly represented because of past air pollution. The heathland ground vegetation has recently become replaced by Pteridium, which introduces a fire hazard not present before.

How the medievals, without guns or cars or effective carnivores, were able to hold down the numbers of deer is unknown. Deer now live in woods and consume whatever is edible within reach; when that is gone they feed in surrounding farmland. Isolated woods sustain far more deer per wooded hectare than continuous forest.

The effect of a sudden change from no deer to artificially large numbers is severe over-grazing of susceptible plants in the woods (Peterken & Jones, 1987; Kirby et al., 2005; Tabor, 1997; Crampton et al., 1998; Cooke & Farrell, 2001; Gill, 2001; Morecroft et al., 2001; Rackham, 2003, Chapter 33; Peterken, 2005). This (especially when combined with shade) impoverishes the flora; affects low-nesting birds, which lack cover, small mammals and even invertebrates (Feber et al., 2001; Flowerdew & Ellwood, 2001; Fuller, 2001; Perrins & Overall, 2001; Stewart, 2001); makes it impractical to resume woodcutting and control shade; and, by subtracting seedlings and shoots of palatable trees such as ash (Fraxinus excelsior) and elm (Ulmus spp.), shifts the composition of the wood. Ultimately the wood is reduced to savanna-like trees plus grasses, together with distasteful herbs such as Arum maculatum.

Deer control is problematic. Fencing is effective around areas of up to 50 ha, but requires vigilant maintenance. Culling deer (as it is rarely possible to kill them all) tends to reduce damage to the fields but not to the wood. The objective of reducing browsing to an ‘acceptable level’ (whatever that may be) is usually impractical because before it is reached deer become too few (and too shy) for shooters to find it worthwhile pursuing them.

Even if deer are absent, sheep and cattle in woods may destroy the ground vegetation. The oakwoods of west Cornwall, which are not grazed, contrast with the otherwise similar ex-coppice oakwoods of south-east Wales, where sheep getting in through breaches in the boundary walls have reduced the herbaceous plants to a few grass species.

This has happened in many other countries (Rackham, 2003, Chapter 33). In the USA, the partial disappearance of native predators and the creation of a mosaic of forest and farmland has been super-favourable to deer; they are further supported by hunters, who are not content with even artificially high densities; many herbaceous and even tree species are eliminated, some apparently permanently (Augustine & Frelich, 1998; Rooney, 2001; Rooney & Waller, 2003; Webster et al., 2005). In Japan, although effects of excessive deer are less obvious, many forests show browse-lines on palatable trees and shrubs, and local proliferations of deer result in spectacularly severe bark-stripping on mountain-tops (Akashi & Nakashizuka, 1999; Yokoyama et al., 2001).

So much for woodland. In wood-pasture and savanna large herbivores, including cattle and sheep, are an integral part of the ecosystem (Humphrey et al., 1998); their withdrawal may result in infilling, with the loss of the nonshade-bearing part of the ecosystem and of veteran trees (see Section X, ‘Infilling of savanna’). Too much grazing may prevent the trees from regenerating; however, this threat should not be exaggerated, for occasional regeneration episodes, every few decades or even centuries, may be sufficient to keep the trees going.

IX. Invasive species

Homo sapiens tends to mix up all the world's biodiversity in various ways, deliberate or accidental. Relatively few introductions become invasive, but once they do attempts to exterminate them are very rarely successful, and even control imposes an indefinite burden on future human generations.

In Japan the giant bamboo Phyllostachys pubescens is said to have been introduced from China in 1746, and has been widely grown for human food. It is a clonal grass, growing 25 m high, capable of overtopping and killing a big tree. It has come to dominate the hundreds of kilometres of abandoned terrace cultivation that fringe the bases of the mountains (Isagi et al., 1997; Isagi & Torii, 1998). Its upward spread is a continuing threat to the ancient woodland above the terraces.

In the tropics, tall vigorous ‘elephant-grasses’, often with a C4 metabolism, such as Imperata cylindrica, Saccharum spontaneum, Hyparrhenia rufa, Themeda quadrivalvis and Cortaderia jubata, originally had a limited range, but people have spread them for various reasons into other countries and other ecosystems. They tend to displace native grassland, especially in savannas. They accumulate combustible matter, leading to more frequent and hotter fires than formerly, which affects the tree structure of savannas and is partly responsible for tropical forests being so easy to destroy (e.g. D’Antonio & Vitousek, 1992; Hooper et al., 2004).

Britain has been let off lightly so far. Common invasive trees are few: Acer pseudoplatanus, Quercus ilex, Rhododendron ponticum and Fagus sylvatica outside its native range. Although locally a serious problem, many woods are still free of them (e.g. Kirby & Patterson, 1992; Peterken, 2001). Similarly, the ground vegetation of ancient woods has been relatively resistant to invasive aliens. However, there is a continuing threat from cultivars and introduced look-alike species. In European gardening tradition, until recently, it has been a point of honour not to grow native plants. Gardeners grow Galeobdolon luteum, not in its native form, but as a variegated cultivar, a vigorous and aggressive clonal plant that outcompetes other species in the garden and is beginning to invade native woodland via garden throw-outs. Anxiety has arisen for the future of Hyacinthoides nonscriptus, one of the most distinctive species of British woodland, lest it be displaced or hybridized by Hyacinthoides hispanicus escaping from cultivation (Dines, 2005). In Ontario the native tree Morus rubra is threatened with hybridization by the introduced, invasive Morus alba (Burgess & Husband, 2006).

Since 1973 ‘amenity’ tree-planting in Britain has grown into an industry in which supposedly ‘native’ trees are planted in quantities never before attempted. For commercial and bureaucratic reasons many of the trees are imported from other parts of their Eurasian range, and are not identical to the native genotypes for which they are passed off (Sell, 2006). As long as this is confined to amenity plantings this may not matter much, but there is a risk that they may turn out to be invasive in native woodland.

Introductions of exotic herbivores have had disastrous impacts on ecosystems not used to them, famously deer in New Zealand which had no native land mammals (e.g. Bee et al., 2007). In Britain, besides the deer abovementioned, the grey squirrel Sciurus carolinensis was deliberately introduced from North America in 1876–1929 and has had a disastrous and irreversible impact on various aspects of forestry and woodland ecology: for example regeneration of the common and important Corylus avellana has ceased to a degree which, although it is long-lived, must compromise its future. In North America feral domestic pigs and European wild pigs, let loose in a pigless continent, have reinforced the proliferating deer in eliminating woodland herbs (Bratton, 1974).

Exotic earthworms are invasive in Australia and North America, where they displace native species (Blakemore, 1999). They profoundly affect forest ecology, especially in parts of North America that had no native earthworms; they destroy the compact leaf-litter layer of trees such as Acer saccharum, and can mix the previously stratified soil layers down to 50 cm deep (Hendrix & Bohlen, 2002).

X. Infilling of savanna

Savanna and savanna-like ecosystems occur where the environment or human cultural practices, alone or in combination, allow single trees (or clonal patches; Fig. 5) to grow but not forest. One of the classic mechanisms in savanna is for the trees to extend their root systems beyond their own canopy and thus to capture rain falling between the trees as well as on them. However, no firm distinction can be drawn between ‘natural’ savannas, determined by climate or soil, and ‘cultural’ savannas, determined by human practices such as grazing livestock and manipulating the frequency of fire.

Figure 5.

An ancient cultural landscape in middle Texas: savanna of grassland with scattered clonal patches of trees (Quercus fusiformis, Ulmus crassifolia and Sapindus drummondii), on clayey soils overlying hard limestone, with annual rainfall of 800 mm. The tree clones, called ‘motts’, tend to be on patches of shallow soil over rock. Elephants became extinct at the beginning of the Holocene. The landscape passed through successive Native American hunter-gatherer cultures; it was periodically burnt and grazed by buffalo (Bison bison) and deer. Land-survey records at the time of European settlement (c. 1845) show that the structure of the landscape was then fully established. Most of the oak motts appear to be several hundred years old. The original tallgrass prairie between the motts, dominated by the perennial grasses Schizachyrium scoparium, Andropogon gerardii and Sorghastrum nutans – what Sir Arthur Tansley would have called a plagioclimax – disappeared under continuous grazing by European livestock and was replaced by unpalatable annuals such as Gutierrezia (Amphiachyris, Xanthocephalum) dracunculoides and Croton capitatus (Diggs et al., 1999). Charred stumps and fence-stakes indicate that the last fires were about a century ago. Under reduced grazing some of the tallgrass prairie species are now returning. There are two continuing threats to what remains of the historic ecosystem: (1) infilling by Juniperus ashei, which is unpalatable but fire-sensitive and has spread out from its original habitat in fire-sheltered canyons and rocky slopes; (2) Ceratocystis fagacearum, which is slowly killing the oak motts. In this situation the term ‘wildwood’ or ‘primeval forest’ is meaningless: human intervention has been critical for the whole of the Holocene. A future juniper forest will be much poorer in species than the savanna it replaces.

All over the world forests have been getting denser and savannas have been infilling into forest, as new trees and shrubs (of the same or a different species) arise between the old trees (e.g. Archer, 1989) (Fig. 3). In middle North America most fragments of savanna that escaped 19th century destruction turned into forest (Peters, 1978).

Figure 3.

Ancient wood-pasture, Sierra de Grazalema, south Spain. The ancient pollard oaks appear to be an excellent bryophyte and lichen habitat. They are becoming infilled in the foreground with Cratægus sp. and in the background with Abies pinsapo. Abies is a rare endemic which a century ago was much less abundant and was itself a savanna tree. In the pinsapo forest are embedded dead ancient pollard oaks and pollard pinsapos; the ground vegetation has disappeared. The conservation of pinsapo has prospered, but at the expense of destroying the cultural ecosystem of which it formed part.

In west Africa, Fairhead & Leach (1998) have demonstrated that much of the tropical forest belt is less than 200 yr old: previously it had been a cultural savanna with a dense human population, who were murdered or carried off by slavers, upon which the savanna became forest. This is shown by travellers’ accounts and historical records, and by surviving savanna trees embedded in the forest.

It might be thought that where savannas infill this proves them to have been cultural. However, the boundary between savanna and forest is not necessarily static: if drought is the limiting factor, a run of wet years may allow a new generation of infill trees to become established, but not always permanently. Oak savannas in the Pindus Mountains of northern Greece, where grazing slackens, begin to infill into forest, but the new trees are very slow-growing and do not get far before their tops die back (Grove & Rackham, 2001, Chapter 12).

Adding to the world's forests at the expense of savanna is not necessarily to be welcomed (Fig. 4). Savanna ecosystems, though less well studied than forests, often have more biodiversity than the forests that replace them. Most of the world's old trees (except, notably, coppice stools) are in savanna or similar ecosystems rather than forest, where neighbouring trees reduce a tree's capacity for retrenchment as it ages. Examples are the ancient pollard trees of English parklands, the baobabs of Africa, the scattered ancient conifers of the subalpine zone in the Alps and Crete, and the famous Pinus longaeva in the mountains of Arizona. Savanna-like ecosystems are particularly rich in animals and epiphytes that are specific to ‘veteran’ trees or that require both trees and open ground.

Figure 4.

The Samariá National Park, Crete. The spectacular natural feature of the Samaria Gorge is also an ancient cultural landscape with remains of villages and cultivation, ancient pollard and sacred trees, churches, mills, and other evidence of 6000 yr of human occupation. It is celebrated in Cretan history and culture. It was deserted and made into a National Park in 1962, largely to protect the agrimi, Capra aegagrus creticus, now known to be a Neolithic feral goat. It is said to be traversed by a quarter of a million tourists a year. Throughout historic time the Gorge has been forested with Cupressus sempervirens, being part of the largest wild stand of this tree in Europe and including the largest individuals. Over the last 100 yr cultivation and grazing have declined and forest has expanded. Much of the increase has been of Pinus brutia, always present but now co-dominant. Because Pinus is fire-adapted but Cupressus is not, this creates the danger of a disastrous fire which would destroy the veteran cypresses and many other features of the historic landscape (Rackham & Moody, 1996; Price et al., in press). There are two pathogenic threats. The scale-insect Marchalina hellenica is killing pines not far from the Gorge. The American fungal parasite Seridium cardinale has ravaged the ‘vertical’ cultivar of C. sempervirens throughout southern Europe, but the Cretan wild tree appears to be resistant. Samariá is also one of the richest localities in Cretan endemic plants, most of which are not forest plants but confined to cliffs where they seem to be secure.

Infilling is a continuing threat to savannas, especially in higher latitudes. Ancient trees may persist for a time in forest, but then die out and have no successors. Another threat is simplification of ground vegetation through agricultural improvement or invasion by exotic grasses, as is common in Australia (e.g. Prober et al., 2005). Few English wood-pastures remain in which the grassland component and the trees both survive intact. In California, although the scattered oaks remain as in the time of the pioneer travellers, often the grassland has been replaced by an assortment of southern European annuals.

XI. Climate change

Increasing carbon dioxide in the atmosphere ought to affect ecosystems, including ancient woodlands, directly. It has already had a measurable effect on the stomatal frequency of many plants (e.g. Woodward, 1987; van Hoof et al., 2006). At the ecosystem scale, CO2-enhancement experiments (Ainsworth & Long, 2005) are not yet directly relevant. However, many studies in ‘intact’ tropical forests report increasingly rapid recruitment, growth and death of trees. This phenomenon has been documented only for the last 30 yr, so whether it is part of normal dynamics is still uncertain. However, the observations are consistent for many species over a wide area, which suggests that either increasing CO2 or increasing temperature is the cause (Phillips et al., 2004; Malhi & Phillips, 2005).

If the climate gets warmer by c. 3°C (equivalent to descending 300 m in altitude or, roughly, to moving 700 km nearer the equator), how much difference will this make to ancient woodlands? England is not a good place to demonstrate a direct effect, for all native trees extend southwards into countries with hotter summers than any that can be expected in England in the future. Reports that trees have already been damaged by heat or drought are mostly related to planted trees. Beeches on thin chalk soils have died after extreme summers such as that of 1976. Beech extends far into Italy, so it is inconceivable that its southern limit is catching up with England. But beech has been a planted tree for 300 yr, and became a fashionable tree for chalklands; people have put it on chalk soils and have still not appreciated that it does not persist on them. Natural woodland on thin chalk soils would be dominated by Fraxinus, or in prehistory probably by Tilia (Thorley, 1981).

In Scotland the endemic subspecies of Pinus sylvestris might be threatened by warmer climate, as part of a general northward retreat of boreal coniferous forests (Saxe et al., 2001); however, it is a mobile tree, with a history of colonizing moorland adjacent to pinewoods, and has room to move to higher altitudes. At its upper limit forest often passes into nonforest by a complex transition, for instance by a savanna zone of scattered old trees, so that altitudinal limits can be affected by other factors, including drought and browsing animals, as well as temperature (Kullman, 2002; Körner & Paulsen, 2004).

Woodlands threatened by global warming should be looked for on the tops of isolated mountains with no opportunity to retreat any higher, for example the ‘sky islands’ of the south-west USA. (Such populations, if found, might be taken as evidence that temperatures substantially higher than the present have not occurred within the Pleistocene.) In the Mediterranean, where there are many mountain-top endemics that might be threatened by warming, few if any are trees or plants limited to forest.

As well as effects on individual species, it has been argued that differences in phenological response (e.g. Root et al., 2003) will disrupt long-established co-adaptations between species. This is a possible threat, but its influence should not be exaggerated, as such relationships have presumably survived greater fluctuations of climate in the Pleistocene or even Holocene. In English woods tree species and ground vegetation are poorly correlated, despite different trees coming into leaf several weeks apart, for example Corylus and Fraxinus (Rackham, 2003, Chapter 6).

What amounts to an experiment in climatic warming has already taken place where woods have become embedded in the growth of cities and their urban heat islands. Ken Wood, London, is a CarpinusQuercus wood which was rural until c.1900 and is now 15 km inside a built-up area. Unlike some urban woods (e.g. Bois de Boulogne, Paris) it has escaped municipal tidying up. Plant records for this wood go back to 1629 (Kent, 1975). Changes in the wood are similar to those general in rural woods (excessive shade, loss of coppicing plants, loss of woodland-grassland plants, and loss of pollution-sensitive lichens) and are not specifically attributable to higher temperatures. Even beech – here doubtfully native – survives.

In England, in addition to general warming, there is forecast to be a reduction in rainfall and a shift in the maximum rainfall from early summer to winter (Mitchell et al., 2007). This would enhance the effects of hotter summers. It would probably result, as dry summers do now, in the slower growth of most trees. This would be bad for those forestry interests that depend on fast growth (Broadmeadow et al., 2005), but is less likely to affect woodland ecology; indeed, slow growth is associated with greater longevity.

England and Wales has a climatic gradient from west (oceanic) to east (continental). It has a parallel gradient from ancient woods mainly dominated by oak in the west to other species in the east. The latter distinction goes back to wildwood times and is unlikely to be closely linked to climate (Rackham, 2006, Chapters 14 and 15); it may not persist into the distant future, not because of climate change but rather because oak has lost its ability to grow from seed within existing woods (see Section XII, ‘Globalization of plant diseases’). Hot dry summers, however, might threaten the eastern outliers of woodland plants with an oceanic distribution. An example is Hyacinthoides nonscriptus (bluebell), a very common plant of ancient woodland and (in the west) of other habitats as well, which might be the first indicator of the effects of global warming. However, in Hayley Wood and Buff Wood (Cambridgeshire), in the most continental part of England, this plant has been observed for 60 yr: its distribution within the woods has shifted in a minor, unsystematic way, but it has not become scarcer. (In recent years it has shown a tendency to aggregate in rings round the bases of trees.) But Primula vulgaris, another oceanic species near the limit of its range, has markedly declined after hot summers, and might be cited as the first victim of global warming (Rackham, 1999, 2003, Chapter 25).

Another prediction is an increase in the frequency of extreme events: a return, maybe, to the unstable weather that existed in the Little Ice Age before the coming of instrumental recording (Grove & Rackham, 2001, Chapter 8). In lower latitudes drought, induced for example by El Niño, has been held responsible for fires in otherwise incombustible rain-forest – which, although a rare event, is not necessarily outside normal dynamics (van Nieuwstadt & Sheil, 2005).

An increase in storms is likely to be wholly beneficial, through counteracting too much shade and through fallen branches adding to the habitat for deadwood animals and fungi. However, it challenges the assumption that the normal state of a tree is upright. Future woods, like those of south and east England after the storm of 1987, may be full of uprooted, horizontal, living trees.

XII. Globalization of plant diseases

Homo sapiens tends to mix up all the world's plant parasites and pathogens. Native pathogens, such as the Armillaria basidiomycetes which are ubiquitous in ancient woodland in Europe, usually keep within normal dynamics, although they may kill exotic trees in gardens and plantations. When introduced into other countries they often eliminate nonadapted trees and other plants, subtracting species after species from forests and savannas.

For instance, in Ohio (USA), which I visited in 2003, Castanea dentata has been removed by the east Asian ascomycete Endothia (Cryphonectria) parasitica, most elms by the Eurasian ascomycete Ophiostoma spp., Cornus florida by the ascomycete Discula destructiva of unknown origin, most red oaks by the ascomycete Ceratocystis fagacearum of unknown origin, and on the border of the state Abies fraseri by the European insect Adelges piceae; ashes, the commonest remaining trees, are about to be destroyed by the Asian insect Agrilus planipennis. This has happened within 100 yr: how much will be left if such introductions continue for another 100 yr?

In Japan, Pinus densiflora, one of the commonest trees and one of the most significant to the historic culture, has largely vanished from much of the country as a result of the introduction of the nematode Bursaphelenchus xylophilus from North America. It is replaced by very different forests of evergreen and deciduous broadleaves (Fujihara, 1996). Less obvious was the appearance in Europe of the North American oak mildew fungus Microsphaera alphitoides in c. 1900; in Britain this may have had a profound effect on the behaviour of Quercus robur, which ceased permanently to regenerate within most existing woodland (Rackham, 2003, Chapters 17 and 32).

Root-attacking oomycetes of the genus Phytophthora have become aggressive pathogens, rapidly evolving to attack new hosts (Brasier, 1999). An ecological tragedy of the 20th century was the introduction of Phytophthora ‘cinnamomi’, apparently a widespread minor tropical pathogen, to south-west Australia, a region of peculiar climate with hundreds of endemic species, nearly half of which turn out to be mortally susceptible (Podger, 1972; Podger et al., 1996).

Introduced plant diseases are not new. Vine-growing in Europe was nearly destroyed in the 19th century by the introduction of three American parasites: powdery mildew Uncinula necator, the aphid Phylloxera infestans (Daktulosphaira vitifoliae), and downy mildew Plasmopara viticola (Large, 1940). The risk has multiplied with the rise in the tree-planting industry, especially with the practice of growing young trees in one country and exporting them to another.

The introduction of exotic trees to a country may be followed, after an interval, by the appearance of their pathogens, which may then – sometimes after a period of genetic adaptation – go on to attack related native trees. Thus the commercial growing of the American Pinus contorta in Sweden is thought to be a threat to the native P. sylvestris (Ennos, 2001).

Pathogens may even be deliberately introduced. Beekeepers in Greece were recently subsidized to spread the scale-insect Marchalina hellenica around forests of Pinus spp., and even to introduce it to Abies cephalonica on which it did not naturally occur, in order to use the exudations of affected trees to make cheap honey – even though hive-bees themselves are threatened by the globalization of pests such as the east Asian mite Varroa destructor.

Lessons are not learnt. Once a pathogen has been established it can rarely if ever be eliminated by counter-measures: Endothia has been rendered partially innocuous in Europe by the unexpected appearance of a fungal virus, but attempts to disseminate this virus in North America have had much less effect (Anagnostakis, 1978; Davelos & Jaresz, 2004).

XIII. Interactions

Threats may reinforce each other. The effects of the 1960–1980 epidemic of Dutch elm disease in England were reinforced (for woodland elms) by deer eating the very palatable regrowth shoots. Disturbance or depletion of forests, which might not matter much in itself, often admits invasive species, especially elephant-grasses and other fire-promoters, which interfere with recovery. Ozone pollution increases in hot summers.

Global warming may allow parasites and invasive aliens to extend their range. It may already have been partly responsible for the proliferation of deer. In the past numbers were limited by starvation in cold winters. Without cold winters deer find plenty to eat outside the wood, helped by the fashion for autumn-sown cereals.

Negative interactions, where one threat counteracts another, seem to be less common. An example is the destruction of ancient woods in England in the 1960s in order to plant Picea excelsa and Picea sitchensis on the sites. Foresters then were obsessed with fast growth and did not realize that spruces would not prosper in dry summers. In eastern England many such plantations have failed and the ancient woods have recovered: an example is Potton Wood near Cambridge, where most of the spruces died in summer 2005.

Fragmentation and loss of connectivity sometimes protect forests by impeding the spread of pathogens (Holdenrieder et al., 2004).

XIV. Importance of different threats

The above 10 influences are outside normal dynamics, or in the case of climate change it is widely believed that normal dynamics have been exceeded. Their significance varies from country to country and from species to species. In England the most immediate threat to native woodland used to be destruction; now it is excessive numbers of deer.

Some of the threats could in theory be reduced by management actions within the woods themselves, although this may be so expensive or labour-intensive, or so much at odds with other human interests, that in practice it is achievable only on particular high-value sites: for example reducing shade or controlling deer. Others can only be combated, if at all, as a small part of national or even global conservation strategies, for example climate change or introduced pathogens.

Effects of these influences vary, both with the properties of trees and ecosystems and with the particulars of human activities, in ways that defy generalization or prediction. A forest dominated by Pinus sylvestris or Fagus sylvatica may be destroyed, or converted into something very different, by felling, but not one dominated by Tilia spp. or Fagus grandifolia, which survive felling. Unfortunately these critical properties – relationships of trees to woodcutting or fire, clonal properties, ability to grow from seed or sprouts – are seldom to be found in Floras or forestry books, and often have to be ascertained by going into the field and looking. Similarly, herbaceous species tend to be subtracted, not in accordance with any top-down prediction, but randomly according to their individual properties. Fallow deer are a threat to Primula elatior but not to Mercurialis perennis.

It is hardly possible to pick out one threat as more serious than another. However, the most widespread threat in the foreseeable future is probably the spread of pathogens. This suddenly and randomly subtracts species after species from ecosystems; it is often an overwhelming influence, turning one ecosystem irreversibly into a quite different ecosystem. (It also threatens plantation forestry, as the world's cellulose supply becomes dangerously dependent on monocultures of a few genotypes of tree. What if, instead of the relatively local P. densiflora, the nematode had attacked Pinus radiata, one of the world's most widespread plantation trees?)

In Britain the tree-planting movement has been so successful that it begins to threaten native woods. Planting is institutionalized and has fallen into the hands of bureaucrats and contractors who plant as many trees as they can before the end of the financial year. Most of what is offered as ‘native’ trees comes from other countries with more reliable seed production and cheaper labour. They may not be native trees but look-alike varieties or even other species in the same genus, whose invasiveness is not known. Importing millions of container-grown trees involves importing thousands of tons of foreign soil, which creates excellent conditions for introducing nematodes or Phytophthoras. It is a superhuman task to expect the Customs to search millions of plant-pots for a microscopic organism when they do not know in advance what to look for. Under present regulations any of the world's pathogens is at liberty to enter Britain provided it does so via some other European Union country.

XV. Conclusions

The world's biological history can be divided into three unequal periods. During the Age of Evolution (the first few thousand million years) it seems, at least from this distance in time, that changes in environment – apart from the occasional asteroid crash – were slow. Not only animals but plants (other than species with exceptionally long generation times, such as the clonal Sequoia sempervirens) could catch up with them by evolution.

This came to an end with the Pleistocene c. 2 million yr ago, when climate lost its apparent stability and climate change replaced evolution as the driving force in plant ecology. Evolution was too slow to catch up with climate, at least for trees and woodland plants, most of which are perennials with long generation times. Although trees can show adaptations to factors such as frost within only a few generations (e.g. Saxe et al., 2001), plants tend now to be mismatched to their environments: they make do with the environments that accidents of history have thrust them into. In the Mediterranean, for example, most deciduous trees and shrubs shed their leaves in winter which would otherwise be their normal growing season. The present regime of hot dry summers has existed for only a small part of each glacial cycle, and only a few woody plants, for example Anagyris foetida, are adapted to summer dormancy.

In the Holocene (the last 12 000 yr) the Age of Unstable Climate was overtaken by the Age of Humanity. The human species has been altering the world's mainland forests ever since the last glaciation. The concept of ‘virgin forest’, meaning forest unaltered by humanity, recedes in the face of archaeological discoveries about the extent and pervasiveness of human activity. Homo sapiens, even in small numbers, acts at a distance by exterminating some animals and increasing the numbers of others, and by manipulating the frequency of fire. It can be argued that all mainland forests, in this interglacial, have been transformed by people exterminating the giant herbivores (the ‘living bulldozers’, super-elephants and their like) of the Pleistocene (e.g. Owen-Smith, 1988; Brook & Bowman, 2004). Most forests, except on remote islands, have had to come to terms with human activities. Even the solitudes of the Amazon are not primeval but result from the destruction of human cultures (Roosevelt, 1980). Evidence of such activity has been slow to emerge, for archaeological surveys are difficult in dense vegetation, but that does not mean that it never happened.

Conservationists have a tradition of striving to make all vegetation under their control conform to some preconceived standard, which changes both with advancing (or receding) knowledge and with changing priorities and fashions. Inconstant objectives have been a destructive factor in long-standing nature reserves or national parks (e.g. Chase, 1986). In the 1990s there was a fashion for preserving or ‘restoring’ wildwood, usually by abandoning historic management and doing nothing: success would be judged by how closely the result approximated to the then prevailing theory of what wildwood was like (e.g. Hambler & Speight, 1995). This encounters three problems.

(1) There is not enough information to allow us to determine what wildwood was. Even in well-investigated England it is still possible to debate whether the pre-Neolithic vegetation was forest or something like savanna (Rackham, 1998; Vera, 2002; Birks, 2005; Mitchell, 2005). Whoever wins the argument, the answer will hardly be robust enough to be a basis of reconstruction.

(2) Changes can be irreversible. In most of England, pre-Neolithic forests (or savannas?) were dominated by Tilia cordata, which is an ancient-woodland plant: it is difficult to exterminate, but once exterminated does not recover lost ground. Conversely, it is rarely practicable to remove introduced species such as A. pseudoplatanus, S. carolinensis, M. reevesi or C. fagacearum. Withdrawing human activity does not re-create what would now be there if the activity had never happened.

(3) Time has moved on, and maybe even evolution has moved things on. Whatever wildwood in England was once like, woodland ecosystems have become adjusted over centuries to periodic felling and lack of grazing animals. Wood-lots are islands, and different woods have adjusted themselves in different ways. Each wood-lot now has its own individual guild of seed-bank plants (Rackham, 2003, Chapter 26), most of which do not adapt to a sudden change to lack of felling combined with excessive grazing.

Restoring wildwood is a credible objective of woodland conservation only in favourable circumstances, for example on remote islands with a short history of human contact. Biological conservation is part of conservation in general, and should be linked to other kinds such as archaeological conservation (e.g. Macinnes & Wickham-Jones, 1992). Conservationists should study the history of each individual site or landscape and consider what makes it special and different from other sites: what are its peculiar and rare features that it is their business to perpetuate.

There is a need to maintain archives of the present or recent state of woodland as a basis against which to measure future changes. The current ethos of ecological research favours designing observations with the narrow objective of confirming (or rarely disconfirming) a preconceived hypothesis (Crogan, 2005). However, ecology is not a science like physics where observations can be repeated at will and investigators’ notes thrown away after the results are published. Early articles in ecological journals are often of more permanent value for the primary data collected by past ecologists than for the authors’ conclusions. Without them we would know little of what English woods were like before the proliferation of deer, or how elms behaved before the 20th century epidemics of Dutch elm disease. I am disturbed by the recent practice, among some journals, of relegating the primary data (‘Supplementary Material’) to a website whose long-term continuation is impossible to predict.


This article is based on a lecture to the Cambridge Philosophical Society on 29 October 2007. The material has come from visits and discussions in many parts of the world. I am especially grateful to John Birks, Susan Bratton, the late David Coombe, Alain Desbrosse, Katsue Fukamachi, Dick Grove, the late Jean Grove, Simon Leatherdale, Daniel Lunney, Jennifer Moody, Toru Nakashizuka, Tim Nevard, George Peterken, Frank Podger, the late John Rishbeth, Christopher Smout, Hiroshi Tanaka, Edmund Tanner, the late Max Walters, Hap Wotila, and Vicki Ziegler.



II.What is meant by threats?573
III.Destruction and fragmentation573
V.Pollution and eutrophication574
VI.Fire and lack of fire575
VII.Excessive shade575
VIII.Excessive numbers of deer and other ungulates576
IX.Invasive species577
X.Infilling of savanna578
XI.Climate change578
XII.Globalization of plant diseases580
XIV.Importance of different threats581