Welcome to Online Research .....

Most important, effective research paper collection in the world

Biodiversity: Climate change and the ecologist

Wilfried Thuiller1

The evidence for rapid climate change now seems overwhelming. Global temperatures are predicted to rise by up to 4 °C by 2100, with associated alterations in precipitation patterns. Assessing the consequences for biodiversity, and how they might be mitigated, is a Grand Challenge in ecology.
BiodiversityClimate change and the ecologistPANORAMIC IMAGES/GETTY
Alpine ecosystem. Species in mountain habitats are especially sensitive to climate change.
How serious is climate change compared with other factors affecting biodiversity?
Very — but it tends to act over a longer timescale. The ecological disruption wrought by climate change is generally slower than that caused by other factors. Such factors include habitat destruction through changes in land use; pollution, for example by nitrogen deposition; the invasion of ecosystems by non-native plant and animal species (biotic exchange); and the biological consequences of increased levels of carbon dioxide in the atmosphere (Fig. 1, overleaf). In the short-to-medium term, human-induced fragmentation of natural habitat and invasive species are particular threats to biodiversity. But looking 50 years into the future and beyond, the effects of climate are likely to become increasingly prominent relative to the other factors.
Figure 1: The main factors, or 'drivers', affecting biodiversity.
Figure 1 : The main factors, or 'drivers', affecting biodiversity. Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, or to obtain a text description, please contact npg@nature.comThis summary of the relative effects by the year 2100 is a composite derived from calculations carried out for 12 individual terrestrial and freshwater ecosystems by O. E. Sala et al. (Science 287, 1770–1774; 2000). Overall, changes in land use constitute the main estimated impact on biodiversity, but the pattern varies considerably for different ecosystems. According to Sala and colleagues' calculations, climate change will have the strongest effect on Arctic, alpine and boreal ecosystems, whereas biotic exchange (that is, invasion by non-native species) will exert its main influence in lakes.
High resolution image and legend (33K)

What are the effects of climate change?
Most immediately, the effects are shifts in species' geographical range, prompted by shifts in the normal patterns of temperatures and humidity that generally delimit species boundaries. Each 1 °C of temperature change moves ecological zones on Earth by about 160 km — so, for example, if the climate warms by 4 °C over the next century, species in the Northern Hemisphere may have to move northward by some 500 km (or 500 m higher in altitude) to find a suitable climatic regime. Higher temperatures are likely to be accompanied by more humid, wetter conditions, but the geographical and seasonal distribution of precipitation will change. Summer soil moisture will be reduced in many regions such as the Mediterranean basin, thus increasing drought stress. Overall, the ability of species to respond to climate change will largely depend on their ability to 'track' shifting climate through colonizing new territory, or to modify their physiology and seasonal behaviour (such as period of flowering or mating) to adapt to the changed conditions where they are.
What about the effect of atmospheric gases?
Carbon dioxide is, of course, known as one of the main drivers of the greenhouse effect, and so of increasing temperatures. But it is also essential for green-plant photosynthesis. Increased atmospheric CO2 results in an increase in photosynthesis rates (through CO2 fertilization), which could potentially balance the effect of temperature increase. This has the largest effect in regions where plant growth is limited by the availability of water, and will probably alter the competitive balance between species that differ in rooting depth, photosynthetic pathway or 'woodiness', as well as the subterranean organisms associated with them. Likewise, an increase of anthropogenic atmospheric nitrogen deposition affects nitrogen-limited regions (temperate and boreal forests, and alpine and Arctic regions) by conferring a competitive edge on plants with high maximum growth rates.
Which ecosystems are we talking about?
All of them, but climate change will affect them in different ways. For example, in marine ecosystems the possible consequences include increased thermal stratification (in which temperature differences separate water layers), reduced upwelling of nutrients, decreased pH and loss of sea ice. These changes will influence the timing and extent of the spring bloom of phytoplankton, and so the associated food chain (krill to fish to marine mammals and birds). On the terrestrial side, deserts, grasslands and savannahs in temperate regions are likely to respond to changes in precipitation and warming in various ways. Mediterranean-type ecosystems, which occur worldwide and are characterized by shrublands, are especially sensitive, as increased temperature and drought favour development of desert and grassland. In tropical regions, CO2 fertilization — in which plants absorb carbon from the atmosphere — and altered patterns of naturally occurring fires will have a strong influence. On tundra, low-growing plants are especially important as habitats for other organisms: their poleward movement will have an ecosystem-wide impact. Finally, species living on mountains are particularly sensitive to changed conditions, in that migration upwards can occur to only a limited extent.
How do biologists monitor changes in biodiversity?
Long-term observations and re-surveys of previously sampled sites are traditional approaches. In certain areas, natural-history societies have long recorded the seasonal time of appearance (of flowers, for instance, or migratory birds), or species' ranges. Such data sets are then viewed against measured variations in temperature or precipitation. Another approach is the re-survey of sites sampled 50 or 100 years previously. Species' identities and abundances are then compared with changes in such external factors as climate or land use. The drawback of both approaches lies in distinguishing a true cause from a correlation.
Do experimental studies help?
Monitoring programmes can be complemented by research in microcosms or, for example, on existing plots of grassland or forest. In these experiments, temperature, precipitation and even CO2 concentration can be manipulated, and such work often reveals unexpected responses arising from the complex interplay of different factors. But for obvious reasons these experiments are difficult to carry out on large spatial and temporal scales.
What responses to climate change are actually documented?
In the Northern Hemisphere, the range of terrestrial plants and animals has shifted, on average, 6.1 km per decade northward or 6.1 m per decade upwards, with advance of seasonal phenomena by 2.3–5.1 days per decade over the past 50 years. These changes are significantly correlated with measured changes in temperature and precipitation. The relationships are correlative in essence, but are too robust, numerous and consistent to be random or to have arisen from other factors (such as natural climatic variability or land-use change). Similarly, the remarkable increase in the plant diversity of some high-elevation peaks in Switzerland over the past 100 years, owing to the upward shift of species that traditionally inhabited lower elevations, can be attributed to changed climate regimes.
Is there a consistent global picture?
We can only guess that patterns such as these are likely to be global in compass, but to differing extents. The two poles are probably being most affected, because the greatest changes in temperature and precipitation are occurring there. By contrast, biodiversity in the equatorial belt is likely to suffer more immediately from deforestation and land degradation. Most of the detailed quantitative studies come from the Northern Hemisphere, or from well-studied 'hotspots' of biodiversity such as the Cape Floristic Region in South Africa. Even in these regions, it is difficult to disentangle the effects of climate change from those of other factors. And we have little or no data on vast swaths of territory in South America, Africa and Asia.
Do climate change and other factors interact?
They do. A notable example concerns invasive species: change in climate can trigger change in biodiversity by creating opportunities for previously innocuous alien species by enhancing their reproductive capacity, their survival and their competitive power against the native flora and fauna. The dispersal of many species, including microorganisms, has been immeasurably increased by the globalization of human economic activity and trade. A combination of climate change, species invasions and reduced areas of natural habitat is likely to promote biotic homogenization in biodiversity hotspots in particular, and to foster unpredictable interactions between plants, animals and microorganisms.
How do ecologists set about forecasting the impacts of climate change on biodiversity?
Experimental studies are informative, but can rarely be generalized. Another approach is to combine ecological modelling with various scenarios of climate change. For example, statistical 'niche-based' models are used to determine the environmental conditions that currently account for species' distributions, and the results can be compared with models of future climate and patterns of land use to predict where these conditions will occur in the future. Validations are usually done by modelling past distributions (as, for instance, surmised for plants from a pollen database). These models don't take into account biological factors such as competition and evolutionary history, but have produced forecasts claiming that 15–37% of natural species will be 'committed to extinction' by 2050. An alternative is 'process-based' modelling, which aims to predict species distributions on the basis of resource allocation, demography or competition. They are theoretically more robust than niche-based models, but require much more ecological knowledge and data.
What are the uncertainties behind forecasting?
All too many, starting with projections of climate change. It is no easy matter to accurately reflect complex interactions (such as those between the ocean and atmosphere), and account for different scenarios of greenhouse-gas emission. There is also our cruel lack of knowledge about the response of biota to rapid climate change. Few, if any, of the most popular models explicitly deal with migration, the dynamics at the trailing edge of shifting populations, species interactions, the interaction between the effects of climate and land use, and the direct effects of changes in atmospheric CO2 and nitrogen deposition. At a basic level, ecologists are still debating the respective influence of interspecific competition and random events in shaping animal and plant communities. And different models tend to provide different predictions of species distribution or biodiversity under similar scenarios of environmental change, showing their limitations.
Can forecasting be improved?
Large-scale, long-term experiments and observations are required to provide the data to make generalization possible, and for modelling studies. Mountains lend themselves to being natural laboratories, given that research can be carried out over steep gradients to investigate the differential response of species and the influence of local adaptations. Overall, what is needed is information that, when appropriately synthesized, can be applied to determine and fine-tune the parameters to be used in process-based models. The building of global databases is a big step forward in accumulating meta-information for this purpose. These databases include compilations of genetic sequences of species (for example, GenBank), the phylogenetic relationships of species (Tree of Life, Phylocom, TreeBASE), and measures of species traits such as mode of dispersal and competitive ability (TraitNet). There is also a new generation of hybrid models of species distributions, which aim for a compromise between realism and accuracy, and complexity and simplicity. These developments are opening up new ways to address the pressing ecological questions: combining hybrid models with statistical advances in 'ensemble' forecasting promises to provide probabilistic projections (Fig. 2).
Figure 2: The probabilistic approach to forecasting biodiversity.
Figure 2 : The probabilistic approach to forecasting biodiversity. Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, or to obtain a text description, please contact npg@nature.coma, Each sub-activity produces an ensemble of projections based on subtly different initial conditions (such as factors influencing species distributions); on the class of model involved and its parametrization; and on the climate-change scenarios chosen. These ensembles are then combined to extract the possible range of outcomes and the likelihood of each occurring. Such estimates are called the 'probability density function' of the event being studied. b, An example of such a function, in which the projected range change of a given species is expressed as a probability of occurrence. In this case, there is an 80% probability that the given species will lose 20–60% of its current range. (Graphic based on M. B. Araújo & M. New Trends Ecol. Evol. 22, 42–47; 2007.)
High resolution image and legend (59K)

What use are forecasts for conservation planning?
For all their imperfections, they are essential. For example, projections of species distributions guide the management of organisms under threat by helping to identify biological corridors for dispersal, sites for reintroduction and areas requiring protection. Lately, the conservation agenda has moved on to consider adaptation to climate change, and to test strategies such as habitat re-creation, creation of dispersal corridors and enhancing the resilience of ecosystems to changed conditions. An alternative is to identify desired future states, and then use models for 'backcasting' to identify strategies for achieving those states. Modellers need to explore how far species-distribution models can be taken to answer the crucial questions that arise from rapidly changing climate. Invasive species are a case in point. In principle, forecasts can predict the probability of an invasive species becoming established, and can incorporate early warning systems for controlling it.
How do human societies fit into this picture?
Much debate has centred on how climate change will affect human welfare through, for instance, rising sea levels and different patterns of crop production. But that well-being also depends on the diversity of organisms used for such 'ecosystem goods and services' as food, energy production and medicines. In certain parts of the world, the chain linking biodiversity, ecosystem processes, and ecosystem goods and services is likely to be broken as biodiversity is affected by altered climatic conditions and the many other factors affecting human health and well-being (Fig. 3). Here again, forecasting can be used to formulate policies that will ameliorate the consequences. For instance, forests are among the most valuable sources of ecosystem goods and services. 'Forest-gap' models can predict tree growth and biomass, the result then being used to guide forest conservation and production strategies.
Figure 3: The complex web of factors affecting human health and well-being, biodiversity and ecosystems.
Figure 3 : The complex web of factors affecting human health and well-being, biodiversity and ecosystems. Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, or to obtain a text description, please contact npg@nature.comChanges in land use through land degradation, and climate change, are the most prominent factors. Perturbation of 'ecosystem goods and services' is just one part of this bigger picture.
High resolution image and legend (45K)

What can we conclude from all this?
As outlined above, our ecological knowledge base and modelling capabilities are far from adequate: making swifter progress will depend on attracting the best scientific talent and the funds to work on these immensely intricate issues. That apart, forecasts of the consequences of climate change for biodiversity need to be couched in probabilistic terms, by stating the possible range of outcomes and estimating the likelihood of each occurring — as is now common practice in weather forecasting. That then presents the problem of recommending a particular course of action for particular circumstances. But if that step can be taken, we reach the stage at which action comes down to political will, at levels running from the global to the individual village.

A force to fight global warming

 Will R. Turner1, Michael Oppenheimer2 & David S. Wilcove3

Natural ecosystems and biodiversity must be made a bulwark against climate change, not a casualty of it, argue Will R. Turner, Michael Oppenheimer and David S. Wilcove.
In the tortured history of climate-change negotiations, enlightened thinking has translated into positive action all too rarely. But governments have recently seen the light on a crucial issue: they have recognized the vital role that intact natural ecosystems have in limiting the build-up of atmospheric greenhouse gases.
When delegates convene in Copenhagen next month to strengthen the UN Framework Convention on Climate Change (UNFCCC), an initiative to preserve the world's forests to store and sequester carbon will take centre stage. Reducing emissions from deforestation and forest degradation (REDD) should give developing countries the opportunity to benefit financially by preserving their forests, either through direct payments or by allowing them to market the carbon stored in uncut trees. Its backers hope that with sufficient funding REDD could substantially slow rates of deforestation, especially in the tropics.
REDD is just one of many possible ways to exploit the potential of natural ecosystems to slow climate change and lessen its effects on people. Natural habitats are a hugely valuable tool in the fight against global warming. Use them wisely and they could save many lives and vast sums of money in the decades to come. Abuse them, and much of Earth's biodiversity could be lost, along with the fight against climate change. Urgent action is needed to understand how best to exploit this promise and develop mechanisms that can be woven into the practices of governments, corporations, communities and institutions worldwide.
To achieve such an integrated approach means fighting a host of powerful short-term political and economic interests. The carbon markets created by REDD might invite corruption, as many critics suggest. Yet the rapid progress that has already been achieved in anticipation of REDD — including new financial mechanisms to ensure verified and lasting emissions reductions, and innovative remote sensing and mapping tools to support them — suggests that these challenges are surmountable1.
There are two good reasons for focusing on natural ecosystems for tackling the threats of climate change. First, forests, peatlands, oceans and other ecosystems control carbon and other global biogeochemical cycles. The oceans alone sequester about 2 gigatonnes of carbon a year. Reducing deforestation and forest degradation rates would slash global emissions by up to 1 gigatonne of carbon a year, more than the emissions of all passenger cars combined. Restoring the world's marginal and degraded lands to natural habitats could sequester an additional 0.65 gigatonnes annually.
The second reason has to do with practicality: the maintenance and restoration of natural habitats are among the cheapest, safest and easiest solutions at our disposal in the effort to reduce greenhouse-gas emissions and promote adaptation to unavoidable changes (see graphic). The basic materials already exist — so there is no need for technological development. Indeed, ecosystem restoration (for example, replanting forest on previously cleared land) may remain for several decades the only realistic large-scale mechanism for removing carbon dioxide already in the atmosphere2.

Natural protection

Environmental carbon storage is worth trillions of dollars to the world's economies, yet it is only one of nature's services. Natural ecosystems will save lives and sustain livelihoods in myriad ways as Earth's climate changes3. For example, healthy mangroves, reefs and wetlands can protect people and property in coastal and inland communities even as climate change threatens to increase tropical cyclone activity. A cyclone in Orissa, India, in 1999 would probably have killed three times as many coastal residents if mangrove forests had not buffered their villages4. Even at current storm levels, coastal wetlands in the United States provide protection against hurricanes worth an estimated US$23.2 billion a year5.
Natural ecosystems do many climate-related jobs. Mangroves, for example, store carbon, buffer against storm impacts, support fisheries and harbour diverse species. Ecosystems also support livelihoods by providing alternative sources of income and food, especially useful if climate change disrupts current sources. Such diversification is helpful for everyone, particularly for the most vulnerable countries and communities — those with the least capacity to cope with climate change.
As important as these services are, what remains to be discovered may be more valuable still. Three decades ago, few imagined that the carbon stored in natural systems would become crucial for combating climate change. Today, enzymes from the gut of a marine crustacean (Limnoria quadripunctata), a type of gribble, show promise in breaking down agricultural waste products for biofuels, potentially reducing greenhouse-gas emissions without competing for agricultural land or threatening natural habitats6. If a promising biotechnology can emerge from a common woodlouse-like creature that lives on the underside of a busy British pier, what untapped potential — the 'option value' of biodiversity — might lie in the world's wildernesses? One area where this untapped innovation could prove particularly valuable is agriculture. When changes in precipitation and temperature start to test the physiological limits of current crops, farmers could benefit from wild relatives and novel cultivars better suited to the new conditions.
The danger is that we will overlook these benefits in natural systems or, worse, lose them. Vast areas of wilderness and undeveloped land are already falling to human abuse, either directly via habitat destruction or indirectly through the effects of climate change. One-fifth of all vertebrates are now threatened with extinction7, and habitat destruction is estimated to cost $2 trillion–5 trillion annually in lost ecosystem services such as the provision of water and carbon storage, vastly more than the cost of safeguarding those services.
Halting this decline requires identifying and securing key intact ecosystems and the climate services they provide, restoring lost or degraded ones, and limiting future losses, all in partnership with the communities that need those services most. At present, climate change is seen as one problem for nature and another for people. This must stop. If human adaptation to climate change compromises biodiversity, then the loss of forests and other natural ecosystems will accelerate climate change, increasing the need for adaptation even as the planet's capacity to accommodate it diminishes. An integrated approach makes the circle virtuous: by conserving biodiversity, we decelerate climate change while increasing the adaptive capacity of people and ecosystems alike.
Climate change is seen as one problem for nature and another for people. This must stop.
There is a real possibility that Copenhagen will create a mechanism for REDD but not a means to pay for it. So the parties to the UNFCCC must initiate financial incentives immediately, engaging public and private sources of funding so that REDD can be rolled out on a global scale. Action is also needed to help governments to monitor natural and modified ecosystems for true net emissions — including those that arise through the displacement of food crops by biofuels and other land-use changes. Policies that provide benefits to, and respect the rights of, local communities are crucial for sustaining and enhancing the ability of natural ecosystems to mitigate climate change.
We also need to try to find ways to value and market the other climate services that natural habitats provide — acknowledging that such services do not exist everywhere — and to weave these benefits into the fabric of our economies. For example, residents of Quito, Ecuador, pay via their utility bills to protect upstream habitats that provide much of their fresh water. Yet, in most cases, communities and corporations are either unaware of, or ignore, the degree to which aspects of climate security, such as their water supply, depend on natural ecosystems.

Climate reserves

For centuries, people have safeguarded natural habitats as public parks and privately owned reserves for nature conservation, sustainable resource production and other goals. They must now be harnessed for the additional goal of combating climate change. National governments, including Costa Rica and the United States, have already begun to acknowledge the importance of natural ecosystems for adaptation in their submissions to the UNFCCC. They need to ensure that any agreement emerging from Copenhagen contains substantive measures to promote the conservation of these ecosystems. Parties to the UNFCCC must also develop key principles for managing and restoring climate services that can be incorporated into international treaties, such as the Convention on Biological Diversity's inland waters biodiversity programme, now under development, as well as environmental assessments by development banks. Knowledge and resources for harnessing climate services should be shared internationally, with developing countries being supported by developed countries.
The future of economies and livelihoods across the planet depends on integrating biodiversity conservation into climate-change planning. If REDD is allowed to fail and degraded lands are also not restored, it is likely to be very difficult to avoid dangerous temperature increases. If coastal and wetland ecosystems are not preserved and restored, tropical storms will become more deadly and more economically damaging. If the diversity of life in the world's wildlands and waters disappears, so do eons of natural innovation that could yield breakthroughs. Working with natural systems rather than against them would unleash a powerful, essential force for halting climate change and reducing its impacts.
See Editorial, page 251, and News Feature, page 266. For the whole biodiversity special, see http://www.nature.com/darwin.


  1. Transparency International Global Corruption Report 2009 (Cambridge Univ. Press, 2009).
  2. Hansen, J. et al. Open Atmos. Sci. J. 2, 217–231 (2008). | Article | ChemPort |
  3. Locke, H. & Mackey, B. Int. J. Wilderness 15, 7–13 (2009).
  4. Das, S. & Vincent, J. R. Proc. Natl Acad. Sci. USA 106, 7357–7360 (2009). | Article | PubMed
  5. Costanza, R. et al. Ambio 37, 241–248 (2008). | Article | PubMed
  6. Sailors' historic scourge may hold the key to bioenergy future (Press release, Univ. York, 2009); available at http://www.york.ac.uk/news-and-events/news/2009/gribbles-bioenergy
  7. IUCN 2008 IUCN Red List of Threatened Species (2008); available at http://www.iucnredlist.org.

Transformation from even-aged plantations to an irregular forest: the world's longest running trial area at Glentress, Scotland

  1. Gary Kerr1,*,
  2. Geoff Morgan1,
  3. John Blyth2 and
  4. Victoria Stokes1
+ Author Affiliations
  1. 1Forest Research, Alice Holt Lodge, Farnham, Surrey GU10 4LH, England
  2. 27 Peel Gardens, Clovenfords, Selkirk, Scotland TD1 3LH
  1. *Corresponding author. E-mail: gary.kerr@forestry.gsi.gov.uk
  • Received September 29, 2009.


The main aim of the Glentress Trial Area has been to study the transformation of even-aged plantations to a permanently irregular structure using group selection. The Trial Area was established in 1952 when most of the plantations were 20–30 years old. The 117-ha area was divided into six Blocks and the plan was to transform the area over a 60-year period by felling and regenerating groups totalling 2 ha in each Block every 6 years. The objectives of this paper are (1) to examine the design, implementation and monitoring of the process of transformation and (2) to investigate if the data collected can be used to quantify the progress of transformation to an irregular structure. The Trial Area has been driven by a clear objective but unfortunately the management plan has not been revised and there has not been a consistent approach to record keeping. This has made it difficult to relate management interventions to the development of the forest structure. An earlier analysis claimed that transformation was almost complete; this was based on a comparison of diameter distributions of the 1990 data with an exponential regression. However, the analysis in this paper includes all the data collected between 1952 and 1990 and shows that the diameter distribution of all Blocks has been similar to an exponential since the start of the Trial. The main reason for this is that the monitoring unit has been the Block, and a spatial scale of ∼20 hectares is probably too coarse to detect the changes that are clear in aerial photographs.


The Glentress Trial Area was established in 1952 with the objective of examining the transformation of even-aged plantations to produce permanently irregular forests. The driving force behind Glentress was Professor Mark Anderson of Edinburgh University who managed to convince the Forestry Commission of the value of establishing a large-scale trial area (117 ha) at a time when normal silvicultural practice in Britain was even-aged plantations and clearfelling, and there was limited interest in alternative silvicultural systems (Hart, 1995). Anderson had been inspired by visits to continental Europe and observations of Norway spruce–European silver fir–beech (Picea abies L.-Abies alba L.-Fagus sylvatica L.) selection forests and predicted that they had a future place in forestry in Britain (Anderson, 1960).
The initial objective of management for the Trial Area was ‘to create and maintain in perpetuity a forest of irregular structure which will function primarily in a protective capacity’ (Supplementary data 1). In 1952, the Trial Area comprised 19 compartments and to facilitate management of the transformation these were grouped into six Blocks (A–F) ranging from 16.6–24.3 ha (Figure 1). Anderson (1955) specified a transformation period of ‘∼60 years’ and described the method of transformation as ‘ …on a 6 year cycle, one tenth of the block will be opened out and planted each year. At the end of the conversion period the block should then consist of groups of various ages ranging at 6 year intervals from 1-60 years, and possibly with remains of some of the original crop aged 90 years’. The planned method of regeneration was to create gaps that were between 0.01 and 0.02 ha. These were then planted with close spaced trees (>10 000 ha−1) drawing on the ideas of Anderson (1951). For example, one surviving plan for Block B in 1953 shows that 98 groups would be formed and planted with 11 150 trees of 10 species in 24 different mixtures. However, it was not been possible to sustain this approach! The main reasons were the cost and the fact that the canopy gaps quickly closed up and the planted trees did not survive. From the mid-1960s onwards, groups of 0.2 ha were used at lower elevations (<325 m above sea level) and groups of 0.1 ha were used on the upper slopes; in addition, a smaller range of species was planted.
Figure 1.
Plan of Trial Area showing significant features and Blocks A–F.
The original plan for the area (Anderson, 1955) also offers information on the periodic assessment of the area (described here in Materials and methods), the desired future species composition and regulation of stocking. With regard to species composition, Anderson stated that ‘the objective is the ultimate establishment of a main crop of climax species of irregular structure and composition, in particular Norway spruce, silver fir and beech’. The logic for aiming at this species mixture is described in greater detail in Anderson (1960). The guidance on regulation of stocking was ‘Treatment and regeneration of stands will not follow any rigid pattern, but will be haphazard. Treatment will be flexible, and it will be modified from time to time on the basis of results obtained’. The use of the word ‘haphazard’ is perhaps unfortunate but the fact that regulation of stocking was not well defined is an indication that, at this time, little was known about the use of alternative silvicultural systems to clearfelling in Britain (Paterson, 1958).
Anderson's belief that forestry in Britain required more experience in the management of irregular stands was proved correct 39 years after the establishment of the Trial Area with the formation of the Continuous Cover Forestry Group in 1991. Continuous cover is an approach to forestry that can be delivered using a range of non-clearfelling silvicultural systems (Mason et al., 1999). This change was part of an international trend towards more ‘naturalness’ in stand level management (Franklin, 1989; Mlinsek, 1996; O’Hara, 1998; Koch and Skovsgaard, 1999), which was driven by the Rio-Helsinki process and the requirements of certification (Forestry Commission, 2004; UKWAS, 2006). Guidance on the use of continuous cover has been produced for British forests (Mason et al., 1999; Kerr et al., 2002; Mason and Kerr, 2004). However, these guides mainly focus on explaining the concept and methods to transform even-aged stands. Looking ahead, the basis of management of forest areas that have been transformed is uncertain, particularly so for irregular stands. There are only a few examples of irregular stands being managed in Britain since before 1991, mainly due to interested individuals on private estates (Hart, 1995; Poore and Kerr, 2009). However, experience and information from the Glentress Trial Area have the potential to play a significant role in developing methods to manage irregular forests of conifers in the uplands of Britain.
The above changes have prompted increased interest in the Glentress Trial Area. This paper describes the existing data from the Trial Area (1952–1990) and a future one will describe changes up to 2008. Progress at the Trial Area has been described by Blyth and Malcolm (1988), Malcolm (1992), Whitney McIver et al. (1992), Blyth (1993), Hart (1995) and Wilson et al. (1999). In general, these papers are descriptive and the only significant analysis was of the 1990 data by Malcolm (1992), which was not published.
The objectives of this paper are:
  • 1 To examine the design, implementation and monitoring of the process of transformation.
  • 2 To describe the main changes in species composition and forest structure and investigate if the data collected can be used to quantify the progress of transformation to an irregular structure.

Materials and methods

Site description

Glentress forest has a total area of 1140 ha and is 4 km northeast of Peebles and 34 km south of Edinburgh, Scotland (Longitudinal 3° 9′ W, Latitudinal 55° 40′ N). The Trial Area is located in the core area of the forest (389 ha) to the southwest of the highest point in the forest, Caresman Hill (561 m), and reaches down to an elevation of 240 m. The topography of the area is characterized by steep rounded hills of glacial origin. The Trial Area occupies the slopes above the Glentress Burn that drains to the south (Figure 1). The aspect of the Trial Area is therefore an arc between southeast and southwest. The underlying Ordovician sediments give rise to generally well-drained soils. At lower elevations in the Trial Area, soils are predominantly acid brown earths with surface water gleys in the main valleys, according to the classification of Kennedy (2002). At higher elevations, there is a mixture of podzols and ironpans, which become skeletal at the highest elevations. Ecological conditions vary throughout the area (Table 1) as does annual precipitation, which ranges between 1000 and 1500 mm year−1.
Table 1:
Ecological site classification (Pyatt et al., 2001) (ESC) at three locations
The ground vegetation of the lower parts of the Trial Area is typical of the site of a W17 woodland (Quercus petraeaBetula pubescensDicranum majus) according to the classification of Rodwell (1991); major components of the vegetation are creeping soft grass (Holcus mollis L.), bracken (Pteridium aquilinum (L.) Kuhn) and ferns (Dryopteris spp.) plus tufted hair grass (Deschampsia caespitosa (L.) Beauv.) in small mid-slope flush areas. As altitude increases, there is a transition to the site type that would be occupied by a W11 woodland (Q. petraeaB. pubescensOxalis acetosella); major components of the ground flora include wavy hair grass (Dechampsia flexuosa (L.) Trin.), bilberry (Vaccinium myrtillus L.) and heather (Calluna vulgaris (L.) Hull).

Original stand composition

Most of the core area of Glentress forest was planted after acquisition by the Forestry Commission in 1921. Historical records show that trees had been planted on some areas before 1921 and these include planting of European larch (Larix decidua Mill.) on Caresman Hill in 1878, a shelterbelt of Scots pine (Pinus sylestris L.) and European larch on the exposed Smithfield ridge and patches of Douglas-fir (Pseudotsuga menziesii (Mirb.) Franco) and Sitka spruce (Picea sitchensis (Bong.) Carr.) in the valley of the Glentress burn (1903). It is clear from Anderson (1955) that parts of the former were later incorporated into the Trial Area but the exact location of the plantings on Smithfield ridge and in the Glentress burn is uncertain. The main species planted by the Forestry Commission between 1921 and 1949 were Douglas-fir at lower elevations (240–320 m): Japanese larch (Larix kaempferi (Lamb.) Carr.) and European larch at mid-elevations (320–400 m) and Scots pine and Corsican pine (Pinus nigra spp. laricio (Poir.) Maire) on the upper slopes (400–560 m).


Mark Anderson was an advocate of the Check method of management (Anderson, 1953; Knuchel, 1953; Reade, 1990) and this was used to assess the Trial Area between 1952 and 1964, subsequently referred to as the ‘Anderson era’. The first assessment using this method was in October 1952 when Block A was assessed. After being marked for thinning, all stems over 12.5 cm diameter at breast height (d.b.h.) were recorded by species in 5-cm classes and a nominal volume was attributed to each diameter class using a single entry volume table (shown in Malcolm, 1971). These data were then summarized for each Block by species and diameter class for ‘the stand’, ‘yield marked’ and ‘stand after thinning’. It does not appear that any checks on actual volume removals were made and the first category was a total figure for the other two. One Block was then assessed, thinned and group planted in the autumn of every year until Block F was assessed in 1957 (Table 2). This method was then repeated with each Block being assessed and treated on a 6-year cycle so that the second assessment occurred between 1958 and 1963. The last full assessment in the Anderson era occurred in September 1964 when Block A was assessed for the third time. The field sheets from the 1952 to 1964 assessments have not survived but the typed summaries have and these were used as a data source for this study.
Table 2:
Summary of transformation plan – year of assessment and harvesting
The second set of data available from the Trial Area was recorded by John Blyth in 1983 but was restricted to Block D. The method used was essentially the same as the Check method of the Anderson era; the main difference was that it included a smaller diameter class, trees 7.5–12.4 cm d.b.h. The field sheets for this assessment have survived and the data were punched, verified and summarized.
The last comprehensive assessment of the Trial was between 1989 and 1990 and used permanently marked plots (40 × 10 m) in a stratified random design (in Block F plots were 50 × 10 m). Each Block was assessed using between 38 and 50 permanent plots. Each Block was sub-divided into smaller areas to facilitate the establishment of plots; the number of plots in each area was weighted by area. A surveying baseline was then established for each area within a Block. Perpendicular to this baseline and on a known bearing, a number of transect lines were then established from known starting points. The number of transect lines established was related to the number of plots and the shape of the area, some transect lines had two plots established on them, while others, in narrower parts of the area, had only one. The lineal distance occupied by each transect in the area was then assessed and a distance (or two) within this range was selected at random to mark the lowest point of the plot. Each plot was split into four 10 × 10 m sub-plots (five in Block F) for ease of recording. In each sub-plot, the following were recorded: (1) species and number of seedlings (<1.3 m tall), (2) species and number of saplings (≥1.3 m tall but <7 cm d.b.h.) and (3) species and d.b.h. (rounded down to nearest centimetre) of each tree (≥7 cm d.b.h.). In addition, a sketch was drawn of each plot showing crop details and vegetation types with approximate cover (on a per cent scale), plus any other features of note, such as the location of any groups of planted trees. In total, there were 198 plots of 0.04 ha and 38 of 0.05 ha, which equates to an 8.5 per cent sample of the total Trial Area of 117 ha. The original field data from this assessment have survived and were punched, verified and compared with Malcolm (1992).

Data analysis

To examine the changes in composition, the species were grouped into ‘spruce’, ‘pine’, ‘larch’, ‘broadleaves’ and ‘other evergreen conifers’ (OEC; Douglas-fir, western hemlock (Tsuga heterophylla (Raf.) Sarg.), western red cedar (Thuja plicata D. Don.), grand fir (Abies grandis Lindl.) and noble fir (Abies nobilis Lindl.)). The proportion of trees in each of these groups was then calculated for all available data. These species groups were also used to calculate the basal area for each Block at each assessment. The calculation of basal area for the data collected by the Check method used the mid-point of each diameter class, whereas that for 1990 used the diameter of each tree.
To examine the shape of the diameter frequencies, an exponential regression was fitted to the data from each Block and the whole Trial Area. The following exponential regression was fitted in Genstat (Payne et al., 1993): Graphic
In addition, the diminution coefficient (q) was calculated as Graphicwhere Y = number of trees per hectare of class i, x = mid diameter of class i, w = width of class i and a and k are constants.
This procedure for fitting of the exponential model has been commonly used in the forest science literature: the ‘standard’ approach. However, from a statistical point of view, there are flaws to the approach, although it can be defended on the basis of giving greater weight to the smaller diameter categories where the fit is sometimes more critical. The fitting of the model by least squares as a non-linear regression assumes that the errors for the data have a constant variance. This is clearly not true when the numbers of trees per hectare, per class, range from zero to several hundred. As the data to which the models are fitted are scaled counts, it is more appropriate to fit the model using a log-linear Poisson generalized linear model (Crawley, 1993). This allows for the variance being proportional to the number of observations. This second method of fitting the exponential model was also used and compared with the results of the standard approach.
The diameter frequency data were fitted using the diameter classes used in the Anderson era; the same range of diameter classes was used for each fit (mid-diameters 15–80 cm). For convenience, one 94 cm d.b.h. Douglas-fir tree in the 1990 assessment of Block E was included in the largest diameter class. The 1983 and 1990 records include data for the number of trees per hectare in the 8–12 cm d.b.h. category but, to be consistent with the earlier data, these were not included in the model fitting.
The regeneration data from the 1990 assessment were examined for each Block to determine species composition and density.


Species composition

The changes in composition for the different species groups for all available data are shown in Figure 2. The general trends for the Trial Area are as follows.
  • The percentage of spruce increased at a rate of 0.6 per cent year−1 from 19.7 to 26.8 per cent during the Anderson era, after this, the rate of increase more than doubled to 1.4 per cent year−1 with the result that spruce was the dominant species group in 1990.
  • The proportion of pine increased slightly in the Anderson era but then declined to less than one-third of its original level.
  • Larch was the most common species group during the Andersen era but by 1990, it had declined to less than half its original level.
  • The presence of OEC has been low (6–9 per cent) but stable over the 38 years.
  • Broadleaves have increased from being barely present in 1952–1957 to being a minor component (1 per cent) in 1990.
Figure 2.
Changes in composition of species groups between 1952 and 1990.
The trends in each of the six Blocks relate to the initial planting of the area between 1922 and 1949 and the implementation of the transformation plan. The main points to note for each Block are summarized below.

Block A

In general, this has good soils and is at a relatively low elevation. This explains the very low proportion of spruce in the Anderson era, which subsequently increased dramatically from 2 to 34 per cent. Block A also has the most OEC throughout the 38-year period having considerably more than Blocks B and E, which would be ranked second and third for this species group.

Block B

This block is in a more elevated position than Block A and has a greater range of soil types. The changes in species groups follow the general trends above, with the exception that the proportion of spruce was lower and those of pine and larch was higher than average during the Anderson era.

Block C

This Block lies to the north of Block B and has the greatest altitudinal range of the six Blocks, sharing some of the most exposed sites with Block D near the top of Caresman Hill. This is reflected in the initially high proportion of spruce; however, this increased further still after the Anderson era. All other species groups were more poorly represented than in the Trial Area: the proportion of pine was low and there were no OEC present until 1990.

Block D

The inclusion of the 1983 assessment has filled a void of information between 1964 and 1990. This confirms that there was not a steady increase in the proportion of spruce between these dates as it declined between 1983 and 1990. Larch was more prominent in Block D compared with other Blocks for most of the Anderson era. The decline in the representation of larch was steep, from 75 per cent in 1955 to 7 per cent in 1983; however, by 1990, it had increased to 21 per cent.

Block E

This Block occupies a mid-slope position in the Trial Area and soils are generally good. In general, the trends in species composition are similar to that for the Trial Area as a whole with the exception that spruce was at a higher proportion than in other Blocks and pine and larch were lower.

Block F

This Block is at the most western part of the Trial Area and has a mixture of soils, most of which are freely draining, and lies at a relatively low elevation. Species group trends for spruce, larch and broadleaves closely follow those for the Trial Area, with the exception that the proportion of OEC was low being <1 per cent throughout the 38-year period.
A more detailed analysis of the change in species between 1952 and 1990 as well as the composition of the regeneration in 1990 are shown in Supplementary data 2.

Basal area

The changes in basal area in the Trial Area between 1952 and 1990 are shown in Figure 3. It is clear from the data that during the Anderson era, there was a great deal of variation between each of the Blocks. At the first assessment, the range of basal area between individual Blocks was 19.5 m2 ha−1 (25.4 (Block A)–5.9 (D)); this subsequently increased to 23.3 m2 ha−1 at the second assessment (31.2 (B)–7.9 (D)) and then decreased in 1990 to 9.7 m2 ha−1 (31.3 (D)–21.6 (B)). The assessment of Block D in 1983 indicates that the basal area of this Block was low until relatively recently. Over a period of 22 years between 1961 and 1983, the basal area of Block D only increased by 7.5 m2 ha−1. Figure 3 also shows the basal area changes between the different species groups, which generally follow the patterns described for the number of trees.
Figure 3.
Changes in basal area between 1952 and 1990 (the number in brackets is the total basal area at the date of measurement).

Diameter frequency regressions

The results for fitting the non-linear regression and the generalized linear model to the data are shown in Table 3. In general, the non-linear regression approach tended to underestimate the value of q relative to the Poisson log-linear approach. The rest of the results concentrate on the non-linear regression so that the values produced can be compared with those obtained from other studies.
Table 3:
Regression parameters from standard fitting (columns 2–5) and generalized linear model (columns 6–9)
The diameter frequency regressions fitted using non-linear regression are shown in Figure 4. In general, all regressions indicate that the shape of the diameter frequency data is similar to an exponential. For the individual Blocks, in all but three cases, the per cent variance accounted for was >90 per cent and in all but five cases, it was >95 per cent. The lowest per cent variance accounted for was Block B in 1990. For the Trial Area as a whole, the per cent variance accounted for was >98 per cent for all three full assessments.
Figure 4.
Diameter distributions and fitted exponential regressions (a) 1952–1957, (b) 1958–1963 and (c) 1990.
The diminution quotient q reflects the slope parameter a in the exponential regression; q describes the ratio of trees in one diameter class to that in the next smallest class. In general, the value of q declined from an initial value of 2.07 for the Trial Area at the first assessment to 1.46 in 1990. In 1990, the range of q for all Blocks except B was 1.4–1.6. There was a consistent decline in the value of q in Blocks A, C and E. In Block F, it showed little change and ranged between 1.46 and 1.49 for all three assessments. In Blocks B and D, the values of q were similar in the first and last assessment but increased at the second assessment.
In general, lower values of the parameter k were recorded as the Trial Area developed (Table 3). This was the case for Blocks A, B, C and E and the Trial Area as a whole. In Blocks D and F, the initial value of k was much lower than in the other Blocks and subsequently the value increased. The range of k in the six Blocks also showed a sharp reduction; in 1990, it was ∼10 per cent of the range in 1952–1957.


Regeneration data have been examined using the above species groups and are presented in Table 4. In general, sapling and seedling regeneration was dominated by spruce. In four of the Blocks, spruce accounted for >70 per cent of the saplings present and in the two Blocks with lower percentages, some of the regeneration had unfortunately been assessed as ‘mixed conifer’. In three of the Blocks, spruce was >70 per cent of the seedlings present. In the other three Blocks, either mixed conifer was present (Block E) or the density of seedlings was low (Blocks A and B). The most abundant species as saplings (present in >10 plots) were Norway spruce (105 plots of 236), Sitka spruce (74), Douglas-fir (25), Japanese larch (23), European larch (19), sycamore (14) and western hemlock (14). The most abundant species as seedlings (present in >7 plots) were Sitka spruce (50 plots), Norway spruce (45), beech (8), Douglas-fir (8) and sycamore (8). The density of saplings varied between 139 ha−1 in Block B and 687 ha−1 in Block C (Table 4). The density of seedlings was much lower and ranged between 27 ha−1 in Block B and 324 ha−1 in Block E.
Table 4:
Summary of species composition and density of regeneration in 1990


The first objective of this paper is to examine the design, implementation and monitoring of the process of transformation. For any long-term trial area to fulfil its potential, it is important to have well-defined objectives and clear consistent record keeping that is appropriate for the subject being investigated. A strength of the Glentress Trial Area is that the initial intention to ‘create and maintain in perpetuity a forest of irregular structure …’ (Anderson, 1955) has not been compromised (Supplementary data 1). However, a major weakness of the Trial Area is that it has lacked consistent record keeping of the transformation plan and its implementation. The position would have been better if the initial plan to use the Check method had been continued between 1965 and 1990. A small but important point is that the Check method requires information to be recorded on volume removals; although thinnings were marked separately to the remaining stand, there was no independent monitoring of whether these trees or others were actually removed. Unfortunately, the Check method was abandoned in 1964 and the failure to replace this with any system of recording is a significant problem. In this study, it has been impossible to relate management and interventions to the structures that had been produced by 1990. For example, it has not been possible to evaluate the success of changing the size of the regeneration groups or explain the different structure of Block B compared with the other Blocks. Although not apparent in this study, a lot of data were recorded in the Trial Area between 1964 and 1988. However, the main driver for this was the production of 35 BSc (Honours) dissertations and 5 doctoral theses (Wilson et al., 1999). It is a shame that this effort, or a proportion of it, could not have been deployed to record relevant information from the Trial Area in a systematic manner. However, despite these problems, a significant success is that the Trial Area has survived. This is mainly due to the interest and enthusiasm of a small number of key individuals at Edinburgh University and commitment from the Forestry Commission.
The original intention of Professor Anderson was to create a Norway spruce–European silver fir–beech forest at Glentress; the justification for this is described in Anderson (1960). However, the plan for the area (Anderson, 1955) does not include this as a specific objective. At one point, the plan states ‘It must be kept in mind that the objective is the ultimate establishment of a main crop of climax species of irregular structure and composition, in particular Norway spruce, silver fir and beech’. However, the first principle of management acknowledges that species composition must be the best adapted to the natural conditions of the area (Supplementary data 4). Hindsight has shown this to be wise; European silver fir and beech have been preferred browse for deer and sheep and it was not possible to plant them without incurring high protection costs. The choice of species has therefore been under continuous review and a wide range of species has been planted in the Trial Area (Supplementary data 2). Analysis of the 1990 data shows that the most abundant species are Sitka spruce and Norway spruce and the main concern about the future species composition is the domination of spruce. Malcolm (1992) proposed limits on the different species groups depending on altitude, for example 60 per cent spruce above 325 m and 37 per cent below. However, no mechanism was put in place for implementing this species composition and it may no longer be desirable because of climate change (Green and Ray, 2009).
It should be clear from the above that the process of implementing and monitoring the process of transformation has not produced sufficient information to explain changes in the structure of the forest in the Trial Area. The transformation has been dominated by group felling and regeneration of areas selected on an annual basis. No evidence has been found that basal area has ever been used as a method of stocking control and very little information has been recorded on how the matrix of the stands have been managed. These lessons are being taken forward in the management of the Trial Area: a management plan has been agreed and starting in 2010, a new planning and recording process will operate. It will also involve an attempt to develop methods of stocking control that ensure that the felling of groups, thinning of the matrix stand and regeneration are in balance.
The second objective of this paper is to describe the main changes in species composition and forest structure and investigate if the data collected can be used to quantify the progress of transformation to an irregular structure. The 1990 assessment of the Trial Area has resulted in two previous attempts to quantify progress towards irregularity. The first of these, and most comprehensive, was by Malcolm (1992). His analysis examined the diameter/frequency data and diameter/volume data, with volume estimated using a local volume table (shown in Malcolm, 1971), and compared them with an ‘ideal’ distribution generated from an even-aged yield table of Norway spruce from Hamilton and Christie (1971). The basis of the comparison was the assumption that each age class of an irregular structure constitutes a miniature even-aged stand (Assmann, 1970, p446) and can therefore be used to generate stocking relationships for uneven-aged stands (Osmaston, 1968, p. 103). The main conclusion of Malcolm's work was that ‘overall (and in most blocks) the inverse-J type of stem number curve was being approached’. Whitney McIver et al. (1992) and Wilson et al. (1999) have published parts of this analysis and both support the ‘just approaching irregularity’ conclusion. The second analysis was by Blyth (1993) who uses two measures of progress towards transformation. The first, in common with Malcolm (1992), was through diameter frequency distributions and the second was by examination of the area regenerated by group felling from aerial photographs. The first part of the analysis concludes that ‘the change in the curves from a typical even-aged bell shape towards a reverse-J is clearly seen’. In the second part of the analysis, he comments that 1993 is a convenient time to assess the progress of transformation as it is theoretically the end of the seventh 6-year cycle. However, due to some missed years (Table 2), it was considered that the transformation was 60 per cent complete. An area analysis of the 1992 aerial photographs showed that only 45 per cent of the total area had been regenerated (Figure 5 shows aerial photographs from 1968, those from 1992 cannot be relocated). The reasons for the difference between 45 and 60 per cent were attributed to excessive shading of the small groups and the impacts of initially sheep and latterly roe deer. As a result, Blyth (1993) forecast that the transformation would be complete by 2033, some 20 years after the initial plan.
Figure 5.
Aerial photograph of the Trial Area taken on 25 April 1968.
Analysis of all data available for the Trial Area shows that comparison with an exponential regression as a measure of the success of transformation is flawed. The main reason for this is that the diameter frequency regressions from the period 1952–1964 are as close to an exponential, in terms of variance accounted for, as those from 1990. The assumption in all the previous published work on the Trial Area that the early diameter distributions are normally distributed is wrong. The fact that the initial diameter distributions have a negative exponential distribution can probably be explained by two main factors. Firstly, each of the Blocks consisted of a number of stands with very different characteristics in terms of species, planting year, site and initial growth. Grouping these areas into circa. 20 ha units and then attempting to detect a change from even-aged stands to a more intimate mix of trees of different sizes are probably too coarse a spatial scale to detect meaningful changes. Secondly, historical records indicate that the Trial Area incorporated areas planted as far back as 1868 and 1903 (Anon, 1952; Anderson, 1955; Malcolm, 1992). The presence of large trees at the start of the Trial Area was significant, particularly for Blocks C, D, E and F (Supplementary data 3). Evidence for these large trees originating from planting before 1921 is provided by the fact that Blocks A and B, which were the first to be planted in the 1920s, have reasonably good growing conditions but did not contain any trees >43 cm in 1952–1957.
Although there is little, if any, evidence of transformation in the diameter frequency distributions, there are some interesting patterns in the number and size of trees over the 38-year period. The general pattern was for q and the parameter k to decrease with time. The value of q is large when the number of small trees is high compared with the larger trees and vice versa. Hence, it is not surprising that in the period 1952–1964, the values of q were relatively high as the majority of trees were small. The changes between the early period of the Trial Area and 1990 can be explained by the growth and development of these small trees and the removal of others in group felling. The range of values of q for all Blocks except B in 1990 is 1.4–1.6. These compare well with the values of 1.58 for ‘all species’ and 1.68 for ‘spruce only’ obtained by Malcolm (1992). These findings are significant as they are the first estimates of q from managed forests in the uplands of Britain. The main practical implementation of this information is that q can be used to generate target diameter distributions to guide thinning of stands managed using continuous cover as described by Kerr (2002).
The value of k is related to q and it is clear that in the early phase of the Trial Area, the large number of small trees was the main factor explaining relatively high values of k in Blocks A, B, C and E. The different patterns in Blocks D and F may in part be attributable to the increased density of trees up to 1990; for Block D, the increase is from 131 to 670 trees ha−1 and for Block F 338 to 461 trees ha−1, all other Blocks except C show a reduction in tree density over the period (Figure 4). The value of k is related to the development of saplings into trees. Many authorities refer to this as ‘ingrowth’ (Roesch et al., 1989), although in some cases, the point at which change is measured can be a tree as large as 16 cm d.b.h. (Poore and Kerr, 2009). The main point of interest is the extent to which the value of k can be used to guide the necessary recruitment levels to sustain uneven-aged stands at Glentress, i.e. by comparing the values of the parameter k (Table 3) with the densities of saplings (Table 4). In Blocks B, C and E, the density of saplings is greater than the value of k and in the other three Blocks, although the density of saplings is lower than k, there are at least 250 saplings ha−1. This may indicate that there is enough regeneration but it gives no clue as to whether the regeneration is in the correct spatial location, i.e. in the areas that have been group felled.
There is no information available on the source of the regeneration, i.e. planted trees or natural regeneration. Because of the written information that exists about the transformation plan, it would be safe to assume that the majority of regeneration has resulted from planted trees. However, Sitka spruce was dominant as saplings and seedlings and this suggests that natural regeneration as been playing some part in determining the species, as it has recently started to demonstrate profuse regeneration in upland forests in Britain (Nixon and Worrell, 1999). The main areas of concern about the regeneration are that there were considerably fewer seedlings than saplings and the dominance of spruce. Because of factors such as browsing, vegetation competition and harvesting damage, there should be more seedlings than saplings. This may indicate that there has been a reduction in effort to plant trees following group felling. Records support this and indicate that the practice followed from the 1980s onwards, the period in which natural regeneration was first observed in the Trial Area, was to only plant if there was no natural regeneration 2–3 years after felling.
The Glentress Trial Area is unique in that it is a large contiguous area of forest in which transformation to an irregular structure has been studied since 1952. In general, there have been three main ways of collecting data for the study of transformation: (1) randomized experiments, (2) growth and yield plots and (3) comparison of managed stands. The use of randomized experiments is challenging because of the need for a large number of plots, each of which is big enough to capture the variability of the transformed stand. Only a few of these studies are reported in the literature, possibly the best known is the Bartlett density study in the White Mountains of New Hampshire, USA (Gove et al., 2008; Leak and Gove, 2008). This started in 1964 and compared 12 treatments replicated four times; however, each of the plots was only 0.13 ha with a 15-m buffer between treatments. The use of growth and yield plots is common in forest science and there are a number of studies that use data from plots that have an irregular or uneven-aged structure (Lundqvist, 1993; Lahde et al., 2001). However, reports of growth and yield plots being used to study transformation are rare. The most common type of study that has been reported in the literature to study transformation is the comparison of managed stands. For example, Schuler (2001) reports results from 21 stands that have compared four different stand treatments since 1951, with a total area of 280 ha. However, it is more common for a smaller number of stands to be compared over a shorter period of time (Meyer, 1952; Leak and Filip, 1977; Schutz, 1997; Schutz, 2001; Lundqvist, 2004; Sterba, 2004; Schwartz et al., 2005; Yoshida et al., 2006; Noguchi and Yoshida, 2007; Neuendorff et al., 2007; Falk et al., 2008; Janowiak et al., 2008).


  1. The fact that the Glentress Trial Area has survived over such a long period of time is testament to the interest and enthusiasm of a small number of people at the University of Edinburgh combined with organizational commitment from the Forestry Commission. A second reason for the success of the Trial Area is that it has had well-defined objectives that have been amended to take account of policy changes, without losing sight of the initial motivations for setting up the Trial.
  2. Unfortunately, there has not been a consistent approach to record keeping, particularly with reference to the transformation plan and monitoring the development of the forest structure. There is a clear lesson here for others considering the establishment of any long-term extensive forest science experiment. Clear, consistent, appropriate and robust record keeping is essential or, as has been the case with this study, it is difficult to relate causes to any effects that may be apparent in the data.
  3. Previous analyses of data from the Trial Area that have concluded that the structure is approaching irregularity are flawed. This study has shown that at the spatial scale of ∼20 hectares few, if any, changes are apparent. However, it is clear from aerial photographs that the Trial Area has been under active transformation for a 38-year period.
  4. The main method and focus of transformation at Glentress has been group felling followed by regeneration. It is surprising that there has been no attempt to develop any methods of stocking control to ensure that the felling of groups, thinning of the matrix stand and regeneration are in balance. Development of appropriate methods of stocking control is a clear priority for the future development of the Trial Area.

Supplementary data

Supplementary data 14 are available at Forestry Online.


Scottish Forestry Trust to the 1990 assessment of the Trial Area.

Conflict of Interest Statement

None declared.


This paper is dedicated to Professor Mark Anderson (1985–1961) whose vision and determination made the Glentress Trial Area a reality. A large number of people from Edinburgh University and the Forestry Commission have been involved in the Trial Area throughout its history and there are probably too many of them to name. However, the following people have had a significant input into the development of the Trial Area and deserve acknowledgement: Douglas Malcolm, Mark Wright (who collected the 1989/90 data), Helen Whitney McIver, Ted Wilson and C.J. Taylor. In addition, we would like to thank the following for their help in the production of this paper: Bill Mason, Hamish Mackintosh, George Gate, Tom Jenkins, Christina Tracey, Helen McKay, Anna Duckett, Steve Mitchell and two anonymous referees. The Scottish Forestry Trust funded the 1990 assessment of the Trial Area and we are grateful for their support of a PhD study that is examining changes at Glentress between 1990 and 2008.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/2.5/), which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.