A number of different criteria have, by general agreement and established practice, become accepted as a means of judging the nature conservation value of a defined area of land (= site); the following have been used in the Review.

In the lowlands of Britain where semi-natural habitats tend to be highly fragmented, the importance of a site usually increases with size of area, and the concept of the 'viable unit' embodies the view that there is a minimum acceptable size for areas which need to be safeguarded in order to maintain their conservation interest. With woodlands or lowland grasslands and heaths, many of the best sites are undesirably small in their total area. Larger sites are not always valued more highly than smaller ones if other qualities are not equal, and minimum or optimum size for a key site varies according to the type of formation. With upland grasslands and heaths, and coastlands, the problem is often the converse one of having to restrict the choice to an area of reasonable size, i.e. not too large. In practice, the extent of a key site is determined by a variety of factors such as diversity, particular interest, and 'natural' boundaries. Size can also be taken as a mark of quality in terms of area of an especially interesting habitat or vegetation type within any formation, or of numbers (i.e. population size) of species of plant or animal. With species populations, an aggregation factor is often involved, notably with colonial animals in which a large proportion of the total British population is located within a relatively small number of sites, i.e. high density may be an important feature. A high proportion of a national, or still more, a world population of a species is regarded as very important. In the case of a wood, size can also refer to the actual stature of the trees, tall well-grown specimens being preferable to the small or poorly grown.

One of the most important site attributes is variety in numbers of both communities and species, which are usually closely related and in turn depend largely on diversity of habitat. It is especially desirable to represent ranges of variation shown by important ecological gradients, e.g. catenas, altitudinal zonation, fading influence of salt spray and blown sand with distance from the sea, stages in pod-solisation of soils, and effects of aspect on biological features. Site diversity is especially related to differences in local climate and micro- climate, topography (affecting drainage, exposure-shelter, aspect), parent rocks and derived soils. Variations in land-use and management practice are often related to these primary factors and have further important effects. Diversity is also related to extent, for the number of species of both plant and animal shows a marked tendency to increase with size of area (the species/area effect), quite apart from the probability that habitat variation will also increase. Diversity is sometimes related to habitat instability and may then give management problems. Often, instability or immaturity of habitat involves serai change, and there is need to represent particularly striking examples of vegeta-tional succession, though many of these inevitably require continual or repeated intervention and management in order to preserve the early serai stages. Sometimes, however, serai changes may have to be allowed to run their course, and it is then important to ensure that earlier stages in the succession are represented elsewhere in the area, i.e. the range of variation must not become depleted overall. Conversely, diversity of an area can often be increased by appropriate management. Areas containing high quality examples of more than one major formation, e.g. woodland and peatland, have especially strong claims to key site status. Very many sites rated as important mainly for one ecosystem contain lower grade examples of another type which can be regarded as a 'bonus', giving enhanced value.

Richness of flora and fauna, i.e. number of species, is an important criterion, and is partly related to extent, but also depends greatly on environmental diversity. Species diversity on areas of similar size is generally a reflection of habitat diversity. It is, however, usual in Britain for an area of calcareous rocks and soil to support a much richer flora than an otherwise similar area of non- calcareous substrata. Thus, areas of limestone tend to be highly rated. Similarly, as the Bryophyta are as a whole a moisture loving group of plants, they tend always to be better represented in the more humid west of Britain than in the drier east. Many more species of bird are likely to occur within a square kilometre of woodland than within a comparable area of upland. In other words, species richness has to be treated as a factor of relative and not absolute importance.

It has been customary to use the term natural for vegetation or habitat which appears to be unmodified by human influence. This is a rare condition in Britain, where so much of the land surface has been profoundly altered from its original state by man's activities. Tansley gave the name semi-natural to modified types of vegetation in which the dominant and constant plant species are accepted natives to Britain, and structure of the community conforms to the range of natural types. For instance, many grasslands are semi-natural, whereas a hop- field or Sitka spruce plantation is artificial. Roadside and railway verges have a semi-natural character in floristics, but are regarded as artificial because of their linearity and setting. The distinction between natural, semi-natural and artificial cannot be rigidly defined, and the separations made in this review are somewhat arbitrary. Nature conservation interest is affected by the actual degree of modification, in both structure and species composition. An abundance or predominance of obviously introduced species usually, in fact, reduces the value of an area, though in moderation, non-indigenous species may add to diversity and interest.

This is a criterion which rates differently according to the formation concerned. For instance, unmodified vegetation is probably most consistently found in upland grasslands and heaths and coastlands whereas the whole of the chalk grassland is in some degree anthropogenic, and it is doubtful if any truly natural woodland remains in this country. Some wetlands have been much disturbed through peat-cutting or other activity, but natural processes of succession over a long period have subsequently restored a nearly original character to the vegetation and habitat, e.g. the Norfolk Broads. In general, this is a difficult criterion to apply: for one thing it is often not easy to judge accurately the degree of modification (especially since the nature of the truly natural ecosystem is often largely a matter of conjecture) and, for another, the realistic view of conservation nowadays results in a high value being placed upon some entirely artificial habitats. The bulk of ecosystems considered in the Review is semi-natural, but insofar as they are identifiable at all, the types least modified by man tend to be rated highly. Naturalness is perhaps of more concern to botanists than zoologists. It is a condition which management sometimes seeks to restore, and is often closely linked to rarity and fragility, i.e. its importance is partly that of scarcity value and dwindling or threatened habitat.

To many people, one of the most important purposes of nature conservation is to protect rare or local species and communities. Rarity on the national scale has been given particular weight in the setting up of non-statutory ' species reserves' by bodies such as the Royal Society for Protection of Birds and local Naturalists' Trusts. In the present review, more emphasis has been given to the inclusion of rare communities, habitats or groups of species, and individual rare species have tended to be regarded as a bonus on sites selected for other reasons. The aggregation of several or many rare species to form a group within a single site, as in a plant refugium, is regarded as an important feature and has influenced the choice of certain key sites. Other things being equal, however, the presence of even one rare species on a site gives it higher value than another comparable site with no rarities.

The rare species which have received particular attention in the present choice of key sites are vascular plants, bryo-phytes, lichens, birds, mammals, Lepidoptera and dragon-flies. Lack of knowledge or interest has led to the relative neglect of other groups, though some consideration has been given to fish, weevils and spiders. A recent examination of the status of rare British vascular plants (defined as species known to occur now in not more than 15 l0-km grid squares of the Atlas of the British Flora, 1962) has given a more objective means of assessing needs and achievements in conserving this group. Comparable data for other groups are mostly lacking at present, but distribution mapping now in progress should in time remedy the deficiency.

Some species tend to be rare because they have extremely specialised habitat requirements, others have become rare because they are the focus of some direct human pressure, including collecting, or suffer indirectly by man's destruction of their habitat. Rarity of species is often obviously related to rarity of habitat, which again links with extent, but many rare species are relict, i.e. they have a discontinuous distribution, with a great many absences from apparently suitable localities, resulting from historical processes which have contracted and fragmented their range. A few rare species are recent arrivals which have not had a chance to spread, and others still (especially birds) are 'fringe' species which could apparently spread, but are at the limits of their climatic environment. Rare species and communities are often thus of great ecological and biogeographical significance, and their conservation is considered to be important. It is essential to understand as far as possible what makes a species rare, since this can affect management needs.

Rarity of habitat and community is closely connected with fragility, though it sometimes depends on the chance occurrence of unusual environmental conditions, singly or in combination, e.g. serpentine is an uncommon rock-type generally, whilst limestone is comparatively rare at high elevations.

This criterion is complex but essentially it reflects the degree of sensitivity of habitats, communities and species to environmental change, and so involves a combination of intrinsic and extrinsic factors. Some ecosystems, such as certain serai vegetation types and associated animals, are inevitably unstable and ephemeral, and may require continuous management to maintain them in a desired state. Vegetational climaxes tend to be more inherently stable, but the natural, climatic types are necessarily more fragile than the biotic types. Intrinsic sensitivity to change varies considerably according to the organisms involved, e.g. during climatic shift, vegetation has a certain inertia of response, whereas certain insects may react very rapidly. Different species within the same taxonomic group can also vary widely in their resilience to adverse conditions.

Certain physical conditions besides climate may give an extrinsic disposition towards fragility, e.g. gravitational instability or a delicate balance in water table, but on the whole it is imminence of human impact, representing 'threat', which forms the main second element in this criterion. Virtually all natural and semi-natural habitats are sensitive to human impact of one kind or another, but there are geographical differences in vulnerability. Fragility rating for a particular ecosystem or site may also increase as land-use pressures intensify and spread. For instance, the great blanket bog ' flows' of east Sutherland and Caithness are easily damaged, but remain relatively safe unless there is increased interest in exploiting these peatlands for fuel or forestry. Some sites have escaped destruction largely by chance, e.g. certain chalk grasslands which, though fairly stable under the traditional grazing management regime, are extremely vulnerable to agro-economic trends. The nature conservation value of many important sites is therefore largely dependent on freedom from radical change in the established land-use pattern.

Fragility is thus a dual concept, but in practice the different elements have usually to be taken together. Fragile sites are usually highly valued, in that they so often represent ecosystems which are highly fragmented, dwindling rapidly, difficult to recreate, or perhaps threatened with total disappearance. Other criteria such as rarity obviously tend to enter this evaluation of survival risks. There are, however, cases where fragility is such that viability is also extremely doubtful, even under favourable conditions of management.

Fragility also applies to species of plant and animal, and especially includes relict or fringe species which maintain a foothold under marginal or suboptimal conditions; it is thus again linked with the criterion of rarity. A good example of a fragile species is the reintroduced large copper butterfly at Woodwalton Fen, for this would clearly die out rapidly but for careful management. Many of the rare British breeding birds, such as red kite, avocet and marsh warbler, are essentially fragile species, in that a small increase in adverse environmental pressure could easily tip the scales against their chances of survival.

The most fragile ecosystems and species have high value, but their conservation may be difficult and often requires relatively large resources.

While key sites, especially the 'living museum' kind, are usually chosen as the best examples of particular ecosystems, their quality may be determined by features which are in some degree unusual. This is valid but it is also necessary to represent the typical and commonplace within a field of ecological variation, insofar as this contains habitats, communities and species which occur extensively or commonly. Sites sometimes have to be selected for their characteristic and common habitats, communities and species, and it is then necessary to overlook the absence of special or rare features. This criterion links particularly with research needs for experimental areas, in which homogeneity may be a desirable feature, for a sufficiently extensive stand of a particular vegetation type is sometimes required, e.g. for plot replication is randomised treatments. The ordinary as well as the unusual attributes sometimes both occur within the same site, and a great many sites rated highly on other criteria take adequate account of typical and commonplace features. By definition, unusual communities or ecosystems may have only a few available samples, whereas there may be a much wider choice of those which are typical or common, and the actual selection may have to be somewhat arbitrary, or influenced by non- scientific factors.

The extent to which a site has been used for scientific study and research is a factor of some importance. The existence of a scientific record of long-standing adds considerably to the value of a site, and can elevate its rating above that of a site comparable in intrinsic quality, but about which little or nothing is known. For instance, the importance of Wicken Fen, Cambridgeshire, is enhanced considerably by the large body of biological and ecological data collected over several decades, giving a picture of the processes which mould and change the nature of the ecosystem in time. The detailed stratigraphical and pollen analytical studies made at Cors Goch Glan Teifi show a classical developmental sequence from lake to raised mire and give this site an importance which could not be accorded simply from an examination of its present surface features. In some cases, sites form the location of long-term studies and experiments whose value would be seriously damaged or destroyed if these study areas were no longer available.

This criterion should not, however, be over-rated. It is less important than the intrinsic features of the sites themselves, for in time the differences in amount of information about sites will tend to disappear, though there may well remain differences in historical value which are directly related to intrinsic site features (e.g. in the completeness of a stratigraphic sequence). Recorded history has not therefore been used as a criterion on its own, though there are instances, e.g. Kerloch Moor, Kincardineshire, where it gives added value to a typical ecosystem, and may with passage of time advance the claims of a site to key status. Where the research has revealed classical features of the site, it points to an important intrinsic feature, but research which is classical in the sense of revealing or extending ecological principles can also give added value to an area.

In the event of two sites representing a certain formation being of equivalent intrinsic value, contiguity of one site with a highly rated example of another formation is regarded as conferring superior quality. Where practicable, and without lowering the standards of selection, it has been felt desirable to include within a single geographical area as many as possible of the important and characteristic formations, communities and species of a district. Clearly, there are few areas where anything approaching a comprehensive representation could be made, and these are mainly in northern and upland districts, where fragmentation of semi-natural habitats is least. Such areas as the New Forest, Hampshire; the Isle of Rhum, Inverness-shire; Durness, Sutherland; Foinaven and Meall Horn, Sutherland; Inver-polly and Knockan, Ross-Sutherland; and Cairngorms, Inverness- shire, Banffshire, Aberdeenshire, illustrate the point. This criterion is obviously related to those of size and diversity. There is also a practical convenience in having two different key sites within a single geographical area.

Certain sites could, through appropriate management or even natural change, eventually develop a nature conservation interest substantially greater than that obtaining at present. Sometimes a site once known to be of exceptional quality has deteriorated seriously in recent years through adverse treatment. This is especially true of certain woodlands which were spoiled by war-time timber extraction, and of some mires which have dried out through burning and/or draining. In such cases, it is sometimes probable that in time, and with suitable management - which depends partly on availability of adequate resources - the former quality of the ecosystem can be restored. When other high-quality examples of the ecosystems concerned cannot be found to take the place of those which have deteriorated, there is good reason for choosing the latter as key sites in the hope that restoration can be achieved through appropriate management. The potentiality for regeneration of high quality mire ecosystems in peat workings is shown by the Norfolk Broads and Moorthwaite Moss, Cumberland, and it is hoped that the interest of Shapwick Heath, Somerset, and Thorne Waste, Yorkshire, will recover in some degree.

Similarly, when a particular ecosystem has been lost altogether, or when no viable examples remain, it may be best to attempt to re-create an example de novo, starting either from some quite different kind of formation, or from one with a closer relationship to that desired. As an instance, it would be possible to produce a woodland of desired type from a grassland or an area of scrub. If, as has been suggested, projected estuarine barrage schemes make provision for the development of completely new freshwater and other wildlife areas, it may be that sites of high quality will come to exist in places where conservation interest is at present quite different or merely negligible. Artificial reservoirs and flooded gravel workings are numbered among the high-quality open water sites, and the importance of other artificial habitats is an indication of the possibilities for creating sites of nature conservation interest. In many instances it would be advantageous to link potential value with the previous criterion, and to choose land contiguous with or part of a key site selected for its existing values.

There is finally the awkward philosophical point that different kinds of organism do not rate equally in value because of bias in human interest, as regards numbers of people concerned. There is no disputing that, for instance, birds as a group attract a great deal more interest in the public generally than do spiders or beetles. Similarly, colourful wild flowers and rare orchids arouse more enthusiasm than toadstools or minute liverworts. While science may view all creatures as equal, therefore, pragmatism dictates that in nature conservation it is realistic to give more weight to some groups than others. This view is supported by the fact that knowledge of the distribution and numbers of the 'popular' groups is often much greater than for obscure groups. The Review has thus given a good deal of weight to ornithological interest (apart from pest species such as the woodpigeon) and many wetlands and coastal sites have been rated highly for their concentrations of wildfowl, waders and seabirds. Nevertheless, within the limitations of available knowledge, an attempt has been made to ensure that the less popular groups of organisms are adequately represented in the key site series.