Forest Naturalness in Northern Europe : Perspectives on Processes , Structures and Species Diversity

Saving the remaining natural forests in northern Europe has been one of the main goals to halt the ongoing decline of forest biodiversity. To facilitate the recognition, mapping and efficient conservation of natural forests, there is an urgent need for a general formulation, based on ecological patterns and processes, of the concept of “forest naturalness”. However, complexity, structural idiosyncracy and dynamical features of unmanaged forest ecosystems at various spatio-temporal scales pose major challenges for such a formulation. The definitions hitherto used for the concept of forest naturalness can be fruitfully grouped into three dimensions: 1) structure-based concepts of natural forest, 2) species-based concepts of natural forest and 3) process-based concepts of natural forest. We propose that explicit and simultaneous consideration of all these three dimensions of naturalness can better cope with the natural variability of forest states and also aid in developing strategies for forest conservation and management in different situations. To become operational, criteria and indicators of forest naturalness need to integrate the three dimensions by combining species (e.g. red-listed-, indicatorand umbrella species) with stand and landscape level structural features that are indicative of disturbance and succession processes.


Introduction
Throughout Europe, modern forest management has resulted in drastic changes in forest ecosystems (Esseen et al. 1997, Kuuluvainen 2002).The most notable changes are decline in the amount of dead wood, disappearance of very large diameter trees, and reduction of the complexity of the tree age and size structure of the stands.Together with the landscape-level fragmentation of natural forest habitats and loss of natural disturbances, these changes have caused remarkable changes to forest structure and function (Kouki et al. 2001, Auvinen et al. 2007, Kuuluvainen 2009).
Scientists, policy makers and non-governmental organizations have all recognized the importance of saving the remaining natural forests as a means to protecting forest biodiversity.In some areas this is not enough and active restoration of lost ecosystem characteristics is needed to secure biodiversity (Angelstam andAndersson 2001, Angelstam et al. 2004a).Of high priority is the protection of the remaining forests that have largely escaped the impact by man.This also includes forests that have in the past been managed non-intensively and often have regained or still retain natural structures and processes that support high biodiversity (e.g.Uotila et al. 2002, Lilja and Kuuluvainen 2005, Lommi et al. 2009, Josefsson et al. 2009).
To develop conservation and restoration programmes and to set measurable goals, it is important to define the concept of 'natural forest'.For example, if different countries aim at maintaining a specific proportion of their forests in natural stage, then some agreement and common idea of 'natural forest' are needed.However, this is often difficult, since naturalness can be defined in various ways and be regarded as a continuous and multidimensional variable (Peterken 1996).The gradient of forest naturalness is driven by a complex set of human influences of varying intensities, often difficult to quantify (Josefsson et al. 2009).Even without human influence natural forest exhibits a range of natural variability, driven by the disturbance-succession cycle (Landres et al. 1999, Keane et al. 2009).It is even more challenging to attempt to estimate the interaction between the natural range of variability and human impact.
The available literature contains a suite of definitions and concepts of natural forests (see a review by Rouvinen and Kouki 2008).In general, virgin and primeval forests are considered as unaffected by man (Peterken 1996).However, the concept of primeval forest has little besides philosophical relevance for countries, such as Denmark and Latvia, where all forests have at some time been harvested or had previous agricultural use.Some untouched areas still exist in Sweden and Finland.For instance, along the Scandinavian mountain range in Northern Sweden high altitude natural forests still remain as a relatively unbroken chain of well-connected forests (Bryant et al. 1997).Similarly, along the Finnish-Russian border some forests have retained their natural characteristics, but are typically small and scattered (Bryant et al. 1997, Aksenov et al. 2002), and the unprotected ones are under threat of harvest (Burnett et al. 2003).
On the other hand, in Russia, which supports the greater part of undisturbed forests in Europe (MPCFE 2007), the idea of primeval forest is more relevant and pristine forest landscapes larger than 50,000 ha have been mapped in a project by Global Forest Watch (Aksenov et al. 2002).In North America, the term old-growth forest is largely used (Bergeron andHarper 2009, Wirth et al. 2009), referring to late-successional forest stands dominated by old trees and largely unaffected by recent human or natural stand-replacing disturbances.Depending on the past history of forest continuity on a particular site, the terms ancient, primary and secondary forest have been applied (Peterken 1996).
Many definitions of natural forest used in Northern Europe overlook ecologically important factors, such as the past history, natural variability and spatial scale, and thus provide inadequate interpretation of the full range of variability in forest diversity (Rouvinen andKouki 2008, Josefs son et al. 2009).Good examples are provided by the post-disturbance phases that are an essential component of natural boreal forest landscape (e.g.Kouki et al. 2001).For example, a young forest regenerating after a stand replacing fire of an old-growth forest is still a natural forest, containing e.g.large amounts of charred dead wood.However, definitions of "natural forest" that emphasize the age of forest or trees ("old-growth") tend to completely ignore such postdisturbance forests.A major scientific as well as applied challenge is to identify valid proxies for forest naturalness so that inherent ecosystem dynamics are also taken into account; at least these dynamics should be recognized when such proxies are suggested.
The specific aims of this paper were to 1) review the usage of the concept of forest naturalness and its multi-dimensionality in northern European conditions, 2) propose a framework for defining forest naturalness based on distinguishing different dimensions of naturalness, and 3) discuss the possibility to develop more robust and general definitions and indicators of forest naturalness.

Approaches to Define Forest Naturalness in Northern Europe
The hitherto taken approaches to define forest naturalness can roughly be divided into three groups: 1) structure-based concepts, 2) speciesbased concepts, and 3) process-based concepts of natural forest.In the following, we discuss and evaluate each of these approaches and their usage in the north European context.Finally we discuss the interdependence of these three approaches in defining forest naturalness.

Structure-Based Concepts of Natural Forest
Forest structures indicating long-term absence of human influence have commonly been used to define degree of naturalness (Lloyd 1999, Norén et al. 2002, Uotila et al. 2002, Lindholm 2003, PEFC Finland 2009).These structures include old trees, variation in tree species composition (especially occurrence of large deciduous trees), multi-layered and multi-aged tree canopies, dead wood of varying sizes and decay stages, as well as signs of natural disturbances (fire, wind, insects, and fungi).Such forest structural components may serve as good proxies for habitat suitability of many species dependent on forests minimally disturbed by human (Angelstam et al. 2004a, Smirnova 2004).Visible and measurable structural patterns can easily be translated also to quantitative targets for forest management (Bütler et al. 2004, Villard andJonsson 2009).As such they may also be modelled allowing for prediction of future habitat suitability for target species (Ranius andKindvall 2004, Tikkanen et al. 2007) and analysing economical trade-offs (Jonsson et al. 2006, Tikkanen et al. 2007).
Definitions required for forest inventories and assessments over large regions (national and European) and by logging companies are typically based on structural stand-level forest characteristics that can be assessed relatively quickly.These definitions stress forest characteristics such as multi-agedness and diverse size structure of forest stands, presence of several trees species and canopy layers, high amounts and continuity of dead wood and generally the lack of major human influence on structures (e.g.Norén et al. 2002, Kriteerityöryhmä 2003, Metsäntutkimuslaitos 2009, MCPFE 2007, PEFC Finland 2009, Timonen et al. 2010).For example, in the 11th Finnish national forest inventory (VMI11 2009(VMI11 -2013) ) naturalness is estimated by three independent factors: structure of tree stand, dead wood continuity, and signs of human activity.For each of these three factors, three levels are separated: (semi) natural, slightly transformed, and clearly transformed forest (Metsäntutkimuslaitos 2009).
Another example of a recent large-scale assessment is provided by the evaluation of threatened habitat types in Finland (Tonteri et al. 2008ab), where forests were classified according to site type, successional stage and dominating species.Several quality attributes related to structural (e.g.amount of dead wood and occurrence of large diameter trees) as well as functional (e.g.signs of past disturbances) properties were used in the evaluation, in addition to quantitative changes in amounts of habitat types.In this work, e.g. the amount of dead wood had clearly defined threshold values (Tonteri et al. 2008ab, Kontula andRaunio 2009).
In Denmark, where the human footprint is profound and has a very long history, and the level of naturalness of forests is at the extreme low end, some definitions do not utilize structural elements indicative of natural processes (Møller 2000).Instead, the naturalness is assessed by time period of tree-layer continuity and natural regeneration, regardless of presence of forest management.In addition, designation as a Woodland Key Habitat in Denmark is based on evidence of any habitat type, structure or species that would be difficult to maintain under conventional forestry (Kitnaes and Forfang 2001).
Although being valuable from the applied perspective, using stand structures as the only proxy for naturalness has limitations.Even if thresholds of dead wood, snags, or tree ages are reached this does not guarantee that associated species are present (for recent reviews on dead wood thresholds, see Müller and Bütler 2010, and for deadwood volume and diversity, see Lassauce et al. 2011).For example, old forest in fragmented landscape in south-western Finland hosts less redlisted polypore species compared to similar forest stands in the eastern part of the country where much more natural forest remains (Tikkanen et al. 2009, Berglund et al. 2011).Thus, the species population dynamics as well as the disturbance processes that take place at landscape scale may lead to extinction of species from individual forest fragments (e.g.Paltto et al. 2006).On the other hand, the presence of some specific structural components is not necessarily a good proxy for naturalness in every site and area (Similä et al. 2006).For example, pristine forest stands at some successional stages may have relatively low volumes of dead wood and be even-aged (Shorohova and Soloviev 2002), which does not diminish their conservation value as an integral part of the natural variability of forest structure.
Besides stand-level structures also whole landscapes may have patterns that indicate naturalness (Mladenoff et al. 1993).With the aid of modern remote sensing techniques landscape patterns and their historical development can be addressed (e.g.Löfman and Kouki 2001) and modelled (e.g.Pennanen and Kuuluvainen 2002).The naturalness of landscapes at different scales can also be estimated by thresholds that estimate the functional amount of habitat for focal species (Mikusiński et al. 2001) indicative of overall diversity (Angelstam et al. 2003, 2004b, Roberge and Angelstam 2006).The definitions of forest naturalness, however, seem not to take into account landscape-level structures explicitly.

Species-Based Concepts of Natural Forest
The structure-based definitions described above are a specific case of species-based definitions: they focus on the main tree species and their characteristics.Species-based approaches have been widely used to identify forests that are of high conservation value, which intrinsically is associated with assessment of the actual naturalness of these forest ecosystems.
In Sweden, several groups of species have been listed as indicator species (signalarter) and used to indicate naturalness and other habitat qualities (Nitare 2000).In Finland, several polypore fungi species have been used to indicate the value and naturalness of old-growth forests (Kotiranta and Niemelä 1996).Red lists of endangered species of forest can be used to provide national guidelines for the management of biological diversity (Rassi et al. 2010), particularly when examined in connection with natural structural characteristics, such as dead wood (Hyvärinen et al. 2006, Tikkanen et al. 2006), or old age of the forests (Tikkanen et al. 2009).EU protected forest habitats (Council Directive 92/43/EEC of 21 May 1992 on the conservation of natural habitats and of wild fauna and flora) include natural forests representing various stages of succession with slight or no human impact (EU 2003).The selection criteria include specialist species and typical plant communities, alongside criteria based on structures such as abundance of dead wood, variable tree size distribution and a number of generations of trees.
In Latvia, Lithuania and Estonia, for conservation purposes natural forest types (Woodland Key Habitats) are defined based on specialist species (fungi, bryophytes, lichens and invertebrates) that assumedly could not survive in stands managed for timber production, (Auzins and Ek 2001).A suite of structural characteristics such as coarse woody debris, trees with hollows and woodpecker signs are also included in the indicator-based definition.In Sweden and Finland, the woodland key habitats are sites that meet criteria based on structural characteristics, and where indicator species occur or are likely to occur (Pykälä 2007, Timonen et al. 2010).

Process-Focused Concepts of Natural Forest
In the following, disturbance and successions are referred to as key ecological processes of natural forest (Korpilahti and Kuuluvainen 2002).However, earlier definitions of natural forests were based on the Clementsian static climax community concept (Clements 1932), which ignores the prevalence of disturbances and which therefore has largely been rejected by modern ecologists.However, the traditional climax forest community concept , which is still used in scientific literature in a diluted form denoting 'old-growth' (e.g.Volkov 2003), probably formed a background for ideas of proposed higher conservation value of late-successional forests.These forests undoubtedly have priority in conservation of biological diversity, but nevertheless they only represent a subset of forests with high natural values.
The natural range of variability in forest dynamics and properties at different scales has recently been emphasized (Landres et al. 1999, Kuuluvainen 2002, Keane et al. 2009).In Northern Europe and Russia, a view of virgin forest as uneven-aged and dynamic was presented at the very beginning of development of forest science (Shorohova et al. 2009, Brūmelis et al. 2011, Hytteborn and Verwijst 2011, Jonsson and Hofgaard 2011).The studies considered age structure as indicative of successional processes within the forest community (Dyrenkov 1984).
In Russia, Bogushevsky already in 1912 argued that a primeval forest (korennoj les) state is characterized by heterogeneity of trees in age and size, implying that the dynamics of natural forest takes place at the scale of small gaps or patches.He also warned of extrapolating results acquired from one site to all sites and tree species, thus emphasizing the uniqueness of ecological context.Morozov (1912) proposed that primeval forests are in a shifting equilibrium state that is continually changing due to multiple pathways of succession.In Sweden, the storm gap theory of virgin spruce (Picea abies (L.) H. Karst.)forest regeneration and dynamics was presented by Sernander (1936).Thus, the basic ideas of shifting mosaic steady state dynamics (sensu Bormann and Likens 1979) and mosaic-cycle of forests (Remmert 1991) were expressed already in the early literature.
The importance of natural disturbances in affecting and maintaining the spatial pattern of forest landscape was indicated in some early studies.For example, Ivashkevich (1915) described the pattern of a virgin pine (Pinus sibirica Du Tour) forest in Siberia.The main features of a virgin forest (devstvennyj les) included unevenagedness, periodicity of peaks in regeneration, very slow growth of the trees during the first 100 years after a major disturbance, and patchiness in spatial distribution.Natural disturbances (wind, fire or insect induced) were assumed to lower the predominance of coniferous forests and convert them into mixed woods.Thus, the importance of natural disturbances in affecting and maintaining the spatial landscape forest pattern was obvious in these early studies.
Based on extensive literature reviews, three main types of disturbance regimes have been distinguished for boreal forests (Angelstam 1998, Angelstam and Kuuluvainen 2004, Kuuluvainen 2009, Shorohova et al. 2009): 1) stand replacing disturbances followed by even-aged stand development (with proposed sub-types of monodominant and compositional change dynamics (Shorohova et al. 2009), 2) partial disturbances creating cohort stand dynamics and 3) fine-scale gap dynamics.Each of these stand dynamics types create specific stand structures, characterized by directional successional change and successional stages (type 1), more or less fluctuating dynamics (type 2), or a shifting steady state dynamics (type 3).Incorporating the dynamic features created by disturbance and successional processes means that deductions of naturalness from structure have to be carried out in context of the disturbancesuccessional cycle of the forest in concern.This also poses requirements to observational scale, as the full range of forest dynamics usually only take place at landscape or even regional scales.

Interdependence of the Components of Natural Forests
It becomes evident from the discussion above that forest naturalness has been and can be defined at different levels and using variable criteria, depending on the actor and the needs (see also Rouvinen and Kouki 2008).However, it appears that usually the definitions are clearly interdependent and many examples of interrelationships between forest structures, processes and species can be provided.To take an example, a natural dry Scots pine (Pinus sylvestris L.) forest, with a long history of cohort dynamics driven by recurrent surface fires, is identified by a multi-aged structure of separate cohorts (e.g.Kuuluvainen et al. 2002), i.e., the presence of the process can be deduced from the structure.This represents the well-known relationship between structure and process in forest ecosystems (Bormann and Likens 1979).The coarse woody debris profile (Stokland 2001) reflects disturbance history and represents another structure-process interrelationship, which can serve as a 'dynamic target' for conservation and restoration of coarse woody debris (Harmon 2001).
The importance of coarse wood debris for many threatened species is well known (Siitonen 2001, Jonsson et al. 2005, Tikkanen et al. 2006).The quantity of structures needed to support biodiversity can be estimated (Müller andBütler 2010, Lassauce et al. 2011).For example, it is estimated that 18 m 3 ha -1 snag volume is needed in an area of 100 ha for presence of three-toed woodpecker, Picoides tridactylus (L.) (Bütler et al. 2004).It has been suggested at least 20 m 3 ha -1 is required for threatened polypore fungi to occur, but the threshold will differ depending on forest type and availability of dead wood in the surrounding landscape (Penttilä et al. 2004, Junninen andKomonen 2011).
Also the presence of species can indicate a past history of natural processes.On fire scarred and charred pine stumps lichens of the genus Hypocenomyce may be present for centuries after the fire.Yet their presence indicates a certain level of continuity in forest fires.However, the presence of structural indicators of natural processes does not always mean that biodiversity will be high.For example, a large forest stand with abundant dead wood in intensively managed Southern Swedish landscape does not host as high biological diversity as present in intact forest landscapes of Russia, since few forests in Fennoscandia have escaped management (e.g.Kouki et al. 2001, Löfman andKouki 2001).In fragmented landscapes where stands with sufficient quality are in short supply, demographic and environmental stochasticity can lead to local species extinctions (Lande et al. 2003) and immigration/emigration and regional stochasticity can lead to extinction of metapopulations at a landscape scale (Hanski 1991).Thus, the landscape scale is important in inventory of natural forests, if used to assess biological diversity (Angelstam et al. 2004b, Penttilä et al. 2004, Rouvinen and Kouki 2008).
Estimates can be made of expected species richness in a landscape based on species-area relationships (MacArthur and Wilson 1967, Tikkanen et al. 2009, Wallenius et al. 2010).Thus, on a landscape scale, the relationships between species diversity and extent of remaining natural forest can be tested.Population sizes of target species can also be estimated for landscapes by building habitat suitability models (Tikkanen et al. 2007).Forest habitat demands have been, for example, determined for capercaillie, Tetrao urogallus L. (Suchant and Braunisch 2004); black grouse, Tetrao tetrix L. (Angelstam 2004); hazel grouse, Bonasa bonasia (L.) (Jansson et al. 2004); Siberian jay, Perisoreus infaustus (L.) (Edenius et al. 2004) and Siberian flying squirrel, Pteromys volans (L.) (Reunanen et al. 2004) and these have been used to define management targets (Tikkanen et al. 2007, Villard andJonsson 2009).It then seems reasonable to expect that national forest inventory data collected on proportion of area of natural forests and structures known to enhance biological diversity would be related to the existing population sizes of specialist species (cf.Hottola et al., in prep.).

Problems in Definitions and Data Compatibility
At the European level quantitative and qualitative information of the state and management of forests is collected nationally using Ministerial Conference on the Protection of Forests in Europe MCPFE (UNECE/FAO) criteria and indicators (MCPFE 2007).The MCPFE Criterion 4 on maintenance, conservation and appropriate enhancement of biological diversity in forest ecosystems includes the indicators tree species composition, regeneration, naturalness, introduced tree species, deadwood, genetic resources, landscape pattern, threatened forest species and protected forests.
In the MCPFE ( 2007) inventory, forests undisturbed by man are defined as showing natural forest features, such as natural tree composition, occurrence of dead wood, natural age structure and natural regeneration processes, the area of which is large enough to maintain its natural characteristics, and no known human intervention or long enough ago to have allowed the natural species composition and processes to have become re-established.At the other end of the scale, plantations are stands of introduced species or intensively managed stands of indigenous species, with one or two species, even age and regular spacing.Semi-natural forests are considered as those that are not natural, nor plantations.Modified natural forests are a subclass of seminatural forests that are near natural.They show characteristics of forests undisturbed by man, such as natural forest dynamics, but which have clear indication of human activities.
As the definition of naturalness lack thresholds, interpretation of the definition varies between countries.For example, using the MCPFE definitions, Latvia has reported that the proportion of natural forest area is 0.5% of total forest area, while 7.4% was reported by its neighbour Estonia (MCPFE 2007).Similar differences between neighbours also are apparent for Finland (3.8%) and Sweden (17.6%).It seems unlikely that such large differences in estimated area of natural forests undisturbed by man would exist if assessed by standard methods using the same definitions.Given that Estonia, Latvia and Lithuania reported high coverage of modified natural forest (63%, 79% and 74%, respectively), it seems odd that Fennoscandian countries appear to lack this subclass of semi-natural forests.Thus, standardization of the definition of natural forest with threshold values for structures is clearly needed.
The MCPFE inventory indicator list includes number of threatened forest species.Such lists are aimed on a national level to direct forest management towards conservation of biological diversity.However, the lists of threatened species differ between countries due to biogeographical differences, past legacy of industrial forestry, and level of research conducted.Thus, it remains unclear how numbers of threatened species could be related to forest naturalness.We suggest that the assessment of naturalness of forests could include indicator lists of forest specialist species.The EC forest specialist bird indicator list (EEA 2004) is a step in this direction, but the species included in this list are mostly widespread species that can successfully utilize urban and recreation forest, such as blackbird (Turdus merula L.), Great Tit (Parus major L.) and Wren (Troglodytes troglodytes (L.)).The EC specialist forest bird list does not include Capercaillie, Three-toed woodpecker and Black woodpecker (Drycopus martius (L.)), which have more specific habitat demands for old coniferous forest (Virkkala and Rajasärkkä 2007).Woodpecker species, many of which have become extinct or form relict populations in some northern Europe countries, have in Poland been shown to be good/sensitive indicators of overall bird diversity (Mikusiński et al. 2001).The woodpecker species also differ in preferred habitat, eg.coniferous versus deciduous forest.Thus, separate lists for biogeographical regions are needed, and also for different forest types (eg.coniferous, deciduous and mixed), as habitat preference differs largely between species.In this respect differentiation of mean dead wood volumes by forest type (coniferous, mixed, deciduous) is also needed to establish relationships with population sizes of specialist species using different habitats.

An Approach Based on Three Dimensions of Forest Naturalness
As becomes evident from the previous discussion, forest naturalness is a dynamic entity encompassing different dimensions at any given point of time.Thus, there is a quest for approaches that are able to integrate different dimensions of forest naturalness.Ideally, a definition of natural forests should incorporate the idea of natural and human-induced variability of ecosystems while addressing the main dimensions of naturalness.These dimensions include, as discussed above, forest structure, processes and species composition (Angelstam 1998, Kuuluvainen 2002, Rouvinen and Kouki 2008).Although ecologically linked, their combinations can vary somewhat aid in developing strategies for forest management.For example, in the boreo-nemoral region (Sjörs 1963), which has undergone a long legacy of industrial forestry and land-use conversion to agriculture, it might be expected that the forests with highest value for conservation of biological diversity are mesic woodland with an abundance of deciduous species (Lõhmus et al. 2005).
Woodland key habitats, which are forests that rank the highest in naturalness within the region, generally lack the pre-industrial structure of the past primeval forest (Ericsson et al. 2005, Jönsson et al. 2009).Thus, in the south boreal region, the appropriate strategies for management might be aimed to ensure temporal and spatial continuity of old deciduous trees, particularly aspen (Kouki et al. 2004).It would be equally important to retain a larger component of coniferous .An illustration of four potential states and combinations of forest naturalness in the structurespecies-processes space (naturalness increases with the axes).Forest 1 represents a natural forest landscape with all natural species and processes fully incorporated (see Fig. 2a).Forest 2 represents an isolated small forest fragment with still rather natural structures but too small to sustain viable populations or natural disturbance and successional dynamics (see Fig. 2b).Forest 3 could represent a case where the natural process has been introduced but the structures and species have not yet been rehabilitated (see Fig. 2c).Forest 4 is an artificial forest with very low degree of naturalness on any of the axis (see Fig. 2d).
independently within wide limits due to the variability of human impact, and natural disturbance and successional dynamics (Fig. 1).Criteria covering forest structure, processes and species composition can be derived to quantitatively rank naturalness of woodlands within this framework (Trass et al. 1999).Table 1 attempts to provide qualitative thresholds for the different dimensions of naturalness in European forests.Primeval forests undisturbed by man show clear signs of natural disturbance, such as fire and windthrow, which are reflected by forest structures (eg.dead wood and large diameter trees), as well as support the natural species in viable populations (Table 1, see forest 1 in Fig. 1 and Fig. 2a).A small coniferous forest stand in a fragmented landscape might still contain abundant natural structures, but the full ranges of succession processes would be limited by fragment size, and the species pool would be affected by reduced connectivity (see forest 2 in Fig. 1 and Fig. 2b), leading to risk of local and eventually regional extinction (Hanski 1991, Lande 2003).
A secondary forest stand that developed on an alluvial floodplain previously used for hay cutting in the mid 1900's in eastern Europe might support high biological diversity of woodpeckers, due to an abundance of structural elements, such as large diameter aspen (Populus tremula L.) with age close to its maximum, and CWD.Thus, a stand of pioneer species such as aspen can develop natural structural features over a relatively short period of time (Bergeron and Harper 2009), but it may not rank high in naturalness on the dimension of processes.Natural processes can be reintroduced in a landscape, for example by fire.However, if there are no large diameter dead trees that typically provide large amounts of dead wood for several specialist postdisturbance species, the initial post-fire succession may fail to attract several early-successional species (Muona and Rutanen 1994, Hyvärinen et al. 2005, 2006).Nevertheless, such a forest would rank high on the process scale (see forest 3 in Fig. 1 and Fig 2c).As an extreme case, a planted and thinned pine (see forest 4 in Fig. 1 and Fig. 2d) or spruce stand would be lacking in structures, processes and species.
We argue that information on the dimensions of naturalness of stands at variable scales can forests that rank highest on the processes dimension, as they likely have a higher capacity to improve on the naturalness scale over a longer period.Both of those aims should involve management targets (e.g.amounts of CWD and large diameter trees) at stand and landscape scales.In the northern boreal region, where there are still large expanses of forest with minimal human disturbance, the aim of management could be to maintain natural disturbance regimes over large landscapes.In regions with less impact from industrial forestry, the aims can thereby be more ambitious and at a larger scale (Angelstam et al. 2004b).Quantitative criteria for ranking of degrees of naturalness (Table 1) are clearly needed.Criteria related to specific quantitative variables such as amounts of dead wood and presence of indicator species are promising.However, these barely cover the conceptual framework that is required to sufficiently address the naturalness that encompasses several aspects in forest ecosystems.As shown above (Fig. 1), evaluating and ranking of the structures, species and processes components separately, as opposed to a summed score, can be used to set more specific and representative priorities in practical conservation.This should enable the use of limited resources of conservation more cost-effectively.A generally applicable method for the whole boreal region seems an unattainable goal at the moment.Due to inherent differences in forest ecosystems in different regions any quantitative criteria and indicators need to be elaborated for individual biogeographic regions at landscape and stand scales, and also for different forest types.Acknowledging this diversity would allow setting conservation performance targets at long-term, regional and management levels, which will differ depending on ambition and the past history of forest use (Angelstam et al. 2004b).
Fig.1.An illustration of four potential states and combinations of forest naturalness in the structurespecies-processes space (naturalness increases with the axes).Forest 1 represents a natural forest landscape with all natural species and processes fully incorporated (see Fig.2a).Forest 2 represents an isolated small forest fragment with still rather natural structures but too small to sustain viable populations or natural disturbance and successional dynamics (see Fig.2b).Forest 3 could represent a case where the natural process has been introduced but the structures and species have not yet been rehabilitated (see Fig.2c).Forest 4 is an artificial forest with very low degree of naturalness on any of the axis (see Fig.2d).

Fig. 2 .
Fig. 2. Examples of states and combinations of forest naturalness in the structure-species-processes -space (see Fig. 1).(a).Pristine taiga forest in the Dvina-Pinega natural forest massif in the Archangelsk region; this is one of the largest intact boreal forest areas in northern Europe.All natural processes, structures and species are intact and present.Corresponds to forest 1 in Fig. 1. (b).Susimäki is a small (50 hectares) protected forest fragment in southern Finland.Many of the natural structures are still present, but the species pool has deteriorated and the full range of natural disturbance and successional dynamics cannot take place in such a small area.Corresponds to forest 2 in Fig. 1. (c).An example of forest where the process has been introduced (fi re and post-fi re succession with deciduous trees), but where the structures and species do not yet correspond to a natural state.A restored site using cuttings and fi re in the Evo area in southern Finland (for details see Vanha-Majamaa et al. 2007).Corresponds to forest 3 in Fig. 1. (d).A planted and thinned young pine forest in southern Finland scores low in all three dimensions of forest naturalness (see forest 4 in Fig. 1).(Photos by Timo Kuuluvainen).

Table 1 .
Assessment of naturalness based on components of structures, species and processes.