Abstract

Examination of temperature variations over the past century for Europe and the Arctic from northern Norway to Siberia suggests that variations in the North Atlantic Oscillation are associated with an increase in oceanicity in certain maritime regions. A southward depression of the treeline in favour of wet heaths, bogs and wetland tundra communities is also observed in northern oceanic environments. The physiological basis for this change in ecological succession from forest to bog is discussed in relation to the long‐term effects of flooding on tree survival. The heightened values currently detected in the North Atlantic Oscillation Index, together with rising winter temperatures, and increased rainfall in many areas in northern Europe, presents an increasing risk of paludification with adverse consequences for forest regeneration, particularly in areas with oceanic climates. Climatic warming in oceanic areas may increase the area covered by bogs and, contrary to general expectations, lead to a retreat rather than an advance in the northern limit of the boreal forest. High water‐table levels are not automatically detrimental to forest survival as can be seen in swamp, bottomland and mangrove forests. Consequently, the inhibitory effects of flooding on tree survival and regeneration in northern regions should not be uncritically accepted as merely due to high water levels. Evidence is discussed which suggests that physiological and ecological factors may interact to inhibit forest regeneration in habitats where there is a risk of prolonged winter‐flooding combined with warmer winters and cool moist summers.

Received: 20 August 2001; Returned for revision: 14 November 2001; Accepted 19 April 2002

INTRODUCTION

Paludification denotes the process of bog expansion resulting from rising water tables as a consequence of peat growth. In many parts of the northern hemisphere and, in particular, in northern Europe paludification is usually accompanied by the disappearance of forest. This is such a common phenomenon that it is accepted, almost without question, as a natural occurrence that needs no further explanation. That trees do not like high water tables is a common but somewhat superficial generalization. It is true that many trees do not like growing in bogs, their deeper anchoring roots die, and the trees become unstable and subject to premature wind‐throw. However, high water tables are not universally detrimental to forest survival as can be seen in the forested wetlands that flourish in the swamps, marshes and bogs of tropical and warm‐temperate regions. The black‐water (Igapo) and white‐water (Varzea) inundation forests of the central Amazonian basin experience annual water‐level fluctuations of up to 14 m, flooding vast areas of forest for periods which can be as long as 270 days per year (Ferreira, 2000). The river swamp forests of North America’s south‐eastern coastal plain are dominated by swamp cypress (Taxodiumdistichum) and swamp tupelo (Nyssa sylvatica var. biflora) and the Sunderban marshes in the Bay of Bengal have at least 26 species of mangrove that survive daily inundation with salt water, even although only nine of the species have pneumatophores (Braendle and Crawford, 1999).

Forests are found on bogs in northern areas but these are usually in regions where the soil is frozen to a considerable depth throughout the winter period and root metabolic activity will therefore be minimal during periods of oxygen deprivation. The black spruce (Picea mariana) and tamarack (Larix laricina) are examples of trees that are notable for their ability to grow on wet peatlands in the continental cold‐winter regions of north America. When spring does arrive there is a rapid transition from frozen soils to a warm growing season. Consequently, during the summer growing season the active layer of the soil horizon will normally be aerated. However, when late spring flooding occurs on these forested peatlands it can be detrimental to growth for both P. mariana and L. laricina (Roy et al., 1999; Girardin et al., 2001).

In maritime regions, as in sub‐Arctic Québec, the Picea mariana forests have been in retreat since the inception of active peat growth from about 6000 bp. Similarly, in the western Siberian lowlands over the past 6000 years there has been a 3–400 km retreat of forest which in some areas used to extend almost to the shores of the Arctic Ocean (Kremenetski et al., 1998). Figure 1 shows the northern limit of the Boreal Forest proper with coloured lines representing the approximate modern limits for the sporadic occurrence of individual tree species north of the limit of closed forest. The areas where Holocene macrofossils have been located and carbon dated are indicated by coloured zones to the north of the current treelines. A pronounced southerly depression of the northern limit of the Boreal forest in the region of the western Siberian lowlands is due to the presence of enormous areas of bog which now cover terrain that was once covered with forest (MacDonald et al., 2000). The climate in this region of the western Siberian lowlands is influenced by the proximity of the Arctic Ocean, which, after the mid‐Holocene rise in sea level and the general climatic cooling after the hypsithermal, brought about a regime of wet cool summers, resulting in extensive replacement of forest by bog. Given the success of the tree‐form in the wetlands of warmer regions the question arises as to why the oceanic wetlands of some northern regions and, particularly north‐western Europe, generally lack significant tree cover. Possible explanations are presented in the discussion as to why wetlands vary in their tree cover and whether or not certain temperature regimes make trees less tolerant of flooding.

CLIMATIC LIMITS OF THE BOREAL FOREST

Throughout North America and Eurasia trees eventually reach a northern distribution limit which can be compared with a number of thermal indicators. However, this does not mean that in all areas the same natural processes are limiting the northern limits of boreal forests. In the past, the northern boundary of the boreal forest (the continental Arctic treeline) has tended to be considered as a purely thermal phenomenon. In North America this can be related to the median July position of the polar front which matches approximately the 10 °C July isotherm (Bryson, 1966). However, a growing season mean temperature of 6–7 °C now tends to replace this older Köppen’s rule where the limit to tree growth was considered as coinciding with the 10 °C isotherm of the warmest month of the year (Körner, 1999). This modification from a measurement indicating maximum warmth to a temperature mean for the entire growing season reflects a realization that the latitudinal and altitudinal limits to tree growth are not directly related to the ability to make a net carbon gain. Instead, the treeline is more likely to be related to the length of growing season that is needed for the production and development of new tissues (Körner, 1999). However, in any discussion between cause and effect it is necessary to remember that mean temperatures, either of warmest months or entire growing seasons, do not exist in nature and therefore should be considered only as indicators and not causal factors (Holtmeier, 2000).

Despite attempts to find a common causal basis for the position of the treeline in North America and Europe a generalized solution has proved elusive. This is not surprising given that treelines are frequently only approximate limits to the altitudinal or latitudinal distribution of scattered but upright trees (Sveinbjörnsson, 2000). The degree of scatter can vary and, in parts of Russia, the development of a mosaic of vegetation at the tundra–taiga interface is particularly noticeable, causing many Russian ecologists to doubt if there is even an approximate treeline that can be mapped with any degree of precision. The basis of the difficulty that forest ecologists have in agreeing on the position of a treeline can be seen in satellite images for North America and Siberia recording the distribution of evergreen vegetation by the Normalized Difference Vegetation Index (NDVI) as seen in May (Figs 2 and 3).

The NDVI is a simple, unit‐less index that is calculated from satellite‐measured reflectance in the red and near‐infrared regions of the electromagnetic spectrum (Tucker, 1979). It is based on the fact that mature, healthy green‐leafed vegetation exhibits strong reflectance in the infrared region and strong absorption in the red region. The NDVI ranges between values of –1 and +1, in the sense that it is mathematically impossible for it to have a value outside this range. Normally there is a strong correlation between the green leaf area index (LAI) and the NDVI, but this is not to imply that when the green LAI is zero the NDVI must also be zero. The value obtained when the LAI is zero is dependent upon the optical properties of the soil and, although it is usually close to zero, it can be negative. Snow and ice give NDVI values close to zero, while water bodies can give negative values (see Tucker et al., 1975; Holben et al., 1980).

In the 8‐km resolution images images recorded in May (Figs 2 and 3) there is a clear discontinuity between the blue area (NDVI = 0) which represents the tundra still covered in snow and ice with the already photosynthetically active evergreen forest which is shown in dark and bright green (NDVI > 0·25). This boundary is particularly clear in NDVI images recorded at the start of the growing season before deciduous vegetation has come into leaf. In North America a southern displacement of the evergreen coniferous forest can be observed in the oceanic regions, which are east and west of the Hudson Bay as well as in Labrador (Fig. 2). Similarly, in the western Siberian lowlands there is a marked southward displacement of the Boreal Forest between the rivers Ob′ and Lena due to the influence of the Arctic Ocean (Fig. 3A–C). The NDVI images also help to illustrate the scepticism with which many Russian ecologists view the concept of a northern boreal forest treeline. As shown in Fig. 1, the northern limit for boreal forest proper deviates to the south of the treeline for individual stands of trees. This extensive transition zone (forest ecotone) between intact forest and completely treeless tundra is particularly noticeable in the western Siberian lowlands (Fig. 1). A high resolution NDVI image from this transition zone is shown in Fig. 3C and reveals the mosaic of tree and bog cover that is characteristic of the forest ecotone that extends at this location north–south for approx. 600 km from the south‐eastern shore of Obskaya Guba to the east–west flowing section of the River Ob′. This mosaic is considered by Russian ecologists to be a self‐renewing cyclic process taking place over hundreds of years. Cryoperturbation causes the soil surface in localized areas to rise above the general level of the bog (Chernov and Matveyeva, 1997). In some areas this permits the active layer to dry out sufficiently to allow the re‐establishment of trees for a period until they shade the ground and cause the permafrost to rise and favour once again the growth of mosses as opposed to trees.

In addition to the problems caused by the cartographic impossibility of tracing a line through the forest ecotone, determining the possible movements of tree limits in relation to climatic oscillations on a continental basis is even more problematical. Even with the dramatic changes in vegetation that take place at the tundra–taiga interface it is difficult to make pan‐Arctic generalizations as to the probable extent of treeline movement at present, or in the foreseeable future, as the climatic history of the different regions in relation to trees is highly varied (Skre et al., 2002). In Alaska, the treeline is currently at its most northerly Holocene extent, while in north‐western Canada (Edwards and Barker, 1994) and northern Québec (Payette and Gagnon, 1985) as well as Siberia (Blyakharchuk and Sulerzhitsky, 1999) there has been a retreat since the mid‐Holocene. Where recent advances in treeline have been observed it is frequently in areas where the dominant tree species is already present in the phenotypically reduced Krummholz form and climatic warming has resulted in the emergence of vertical trees. [Note: genetically fixed Krummholz forms will have to be replaced by other populations to show a change in form; see Holtmeier, 2000.] A recent change in Krummholz growth form has been noticed both in sub‐Arctic Québec (Lavoie and Payette, 1994) and on the western shore of Hudson Bay. In this latter site, the Picea mariana treeline runs north–south, parallel to the shore and has advanced eastwards by 12 km since the late 1800s due to the development of vertical trees within already established Krummholz (Lescop‐Sinclair and Payette, 1995).

HISTORY OF PALUDIFICATION

The early Holocene warm period, usually referred to as the hypsithermal or ‘climatic optimum’, facilitated the advance of forests in the northern hemisphere several 100 km to the north of the present position of the tundra–taiga interface. The duration of this early Holocene hypsithermal interval differs from one region to another in the boreal zone as well as between the northern and southern hemispheres. In south‐western Saskatchewan there was a warm‐dry hypsithermal period between 6400 and 4500 bp (Porter et al., 1999), while in north‐west Montana there was a warm dry period that lasted from 10 850 to 4750 bp (Gerloff et al., 1995). In western Norway the hypsithermal is dated to between 8000 and 6000 bp (Nesje and Kvamme, 1991) when hazel (Corylus avellana) extended as far as the North Cape. In Russia the early Holocene saw a spread of spruce, larch, pine and tree birch to more northern latitudes than they now inhabit (Fig. 1). In Siberia, Picea obovata was farther north than at present between 8000 and 4500/4300 bp and Larix spp. were further north between 10 000 and 5000/4500 bp. Tree birches (Betula pubescens) reached the present‐day shoreline of the Barents Sea in the Bolshezemelskaya Tundra (68°N) and in the Taimyr Peninsula (72°N) between 8000 and 9000 bp. In the Yamal Peninsula the tree birch limit was near 70°N by 8000 bp (Kremenetski et al., 1998).

In Scotland recent research has shown that in the Hebrides and northern Isles (Orkney and Shetland) large areas that are now treeless, hyper‐oceanic bogs, once had an extensive cover of birch woodland with Corylus avellana, Salix spp., Populus spp. and Sorbus aucuparia extending to their coastal fringes, while more central and eastern areas may have had stands with more warmth‐demanding species (Tipping, 1994).

The ecological consequences of forest retreat are not uniform and suggest that the causes of forest disappearance may differ from one region to another. In the more continental areas forest gives way to tundra. However, this is not a universal situation. In many areas the treeline is now at latitudes where thermal conditions are adequate for tree growth but regeneration is prevented by the growth of peat. The development of oceanic conditions in these northern areas appears to have taken place as sea levels rose in the Arctic Ocean and the Hudson Bay. This early to mid‐Holocene expansion of Arctic waters altered the delicate balance between temperature and air humidity so that both in Arctic and sub‐Arctic regions bogs began to replace forest over a wide area both in northern Québec (Payette, 1984) and in the Siberian lowlands (Kremenetski et al., 1998). A marked climatic deterioration (Klitgaard‐Kristensen et al., 1998), commonly termed ‘the 8200 bp event’ (probably due to freshwater fluxes in the final de‐glaciation of the Laurentide ice sheet), appears to have been accompanied by a reduction in tree cover throughout Europe. In the Outer Hebrides (Scotland) blanket peat began its appearance between 9000 and 8000 bp (Fossitt, 1996) and in western Lewis there was a progressive replacement of trees by blanket peat which began about 7900 bp and continued with a further decline between 5200 and 4000 bp. A trend to more oceanic conditions with cool moist summers appears to have continued in many areas probably as a result of these early Holocene rises in sea level. The Arctic Ocean and the Hudson Bay would therefore have begun to exert a greater maritime influence on climate in northern Siberia and Québec, respectively. By approx. 5000/4500 bp the northern limit of tree birch in Russia was similar to its present limit and this was mirrored in the southward retreat of other tree species.

The late Holocene retreat of the Eurasian treeline coincides also with declining summer insolation. This, together with the cooling of Arctic waters, would have created at northern latitudes a more oceanic climate with cool moist summers which would allow permafrost levels to rise and facilitate the development of bogs. The considerable antiquity of the peat in the upper layers of bogs in the Pur‐Taz region of western Siberia (for location see Fig. 3C) has been taken to suggest that much of the peat growth took place at an early date (Peteet et al., 1998). However, comparison of methane emissions from the Holocene Optimum (5500–6000 years bp) with modern levels from forests of northern Eurasia, suggests that the area of tundra and the proportion of wetlands within the boreal forest zone is probably greater now than at any time in the past (Velichko et al., 1998). Examination of the recent growth of bogs suggests that bogs appear to engulf the forest both in the maritime regions of Canada and northern Siberia as well as in other locations, particularly near the present tundra–taiga interface. The situation is complex, however, as at this interface between forest and bog there is a cyclical alternation between tree cover and wetland vegetation (see below).

In Russia there has been a lengthy debate among ecologists as to whether climatic warming will cause an advance or retreat of the treeline (Kriuchkov, 1971). Those who assume that climatic warming causes an advance of the treeline northwards subscribe to the eventual overriding influence of climatic factors. There is, however, a powerful argument, held by those who believe that the presence or absence of trees is controlled by edaphic factors influenced by the proximity of the Arctic Ocean and the long‐term persistence of permafrost, that climatic warming in the North Siberian lowlands can cause a retreat of the treeline due to peat growth. Examination of the history of methane emissions from boreal regions would suggest that if the present trend to warmer, wetter conditions continues with longer growing seasons, then bog expansion and paludification are likely to increase (Peteet et al., 1998).

In North America recent studies have questioned whether or not temperature alone is the main controlling factor for the northern limit of forest, and suggest that the risk of natural fires may be particularly acute at the tundra–taiga interface. The frequency of forest fires makes it essential that trees retain a capacity for subsequent regeneration (Payette et al., 2000). The removal of trees, whether by fire or any other agency, will lead to rising water tables. Fire can also aid bog formation as particles of ash and carbon deposited into the soil profile can reduce drainage and therefore initiate peat growth (Mallik et al., 1984). Trees can colonize peatlands in cold climates but remain nevertheless highly susceptible to deteriorating environmental conditions (Arseneault and Payette, 1997). In a study covering the little ice‐age periods in northern Québec, it was found by these authors that black spruce continued to colonize the peatlands as long as the adjacent well‐drained sites were occupied by seed‐producing forest. This need for neighbouring forest is two‐fold. In addition to the need for a seed source there is also a requirement for shelter to reduce the damaging effects of winter snow‐drifting conditions. When neighbouring areas succumb to periodic fires then both the seed source and shelter are removed. Under these conditions the surviving trees suffer die‐back of supranival stems. The trees then finally succumb to drowning in permafrost‐induced ponds. The post‐fire degradation and disappearance of the conifer stands from the peatlands represent the ultimate stage of a positive feedback process triggered by a modification of the snow regime at the landscape scale. In this connection it is relevant to note that recent research in Siberia has shown that along the tundra–taiga boundary there is a high frequency of thunder storms suggesting an increased fire probability due to the warmer conditions of the forest increasing evapo‐transpiration from the bog (Valentini et al., 2000).

One of the most dramatic examples of boreal landscape changes from forest to bog is the deforestation of northern Scotland that took place at the beginning of the Neolithic period. As noted above, during the hypsithermal period, forest cover in Scotland extended from the mainland to the Hebrides and the northern Isles of Orkney and Shetland. Some replacement of trees by blanket bog took place before the period of Neolithic settlement (Fossitt, 1996). However, the advent of Neolithic farming in oceanic areas such as Orkney and Shetland (Bennett et al., 1992; Bunting, 1996), western Norway (Kaland, 1986) and the Hebrides (Tipping, 1994) was marked by the rapid and extensive replacement of trees by heathlands and peat bog to an extent that did not take place in more continental areas. The reasons for this apparent sudden deforestation and ultra‐sensitivity to human disturbance in oceanic areas need further examination. It may be that the grazing activities of Neolithic settlers opened up the tree canopy sufficiently in the maritime environment to allow the water table to rise and finally accomplish the paludification process and tree‐removal which had begun with the post‐hypsithermal onset of oceanic conditions.

MONITORING CLIMATIC CHANGES IN OCEANIC ENVIRONMENTS

If mild, wet winters present a threat to tree survival in the more oceanic regions of northern Europe, then current climatic trends need to be examined to determine the future viability of northern forests and whether they are likely to be negatively affected by increased paludification. It is now apparent that in relation to the growth of wetlands there is a cyclic behaviour in the climatic record of northern Europe. Ten periods of increased effective precipitation have been detected in changes in bog vegetation in northern England since 760 bc (Mauquoy and Barber, 1999). It is also possible to relate the intensity of maritime conditions with the periodic behaviour of the North Atlantic Oscillation (NAO). The NAO has emerged as the dominant varying atmospheric phenomenon in the northern hemisphere and the only atmospheric teleconnection pattern with a periodic behaviour that matches the 700‐year stable isotope record of the Greenland GISP2 ice core (White et al., 1997). In the past century there have been three periods with prolonged and significant increases in the winter (January to March) NAO index. Anomalies caused by these changes include dry winter‐time conditions over southern Europe and the Mediterranean and wetter‐than‐normal conditions over northern Europe (Hurrell, 1995; Hurrell and VanLoon, 1997). Warmer and wetter winters have also had noticeable effects on agriculture, interfering with potato harvest and the sowing of winter cereals in the autumn as well as delaying field operations in spring.

Some of the changes in climatic pattern which may result from variation of the NAO index can be integrated by examining the degree to which maritime conditions (or the converse continentality) may be changing over an extensive geographical area. One such integration is obtained with Conrad’s Index of Continentality (Conrad, 1946) where variation in temperature range is combined with a correction for latitude:

CI = [1·7A/sin (ø + 10)] – 14

where CI is Conrad’s Index of Continentality; A the average annual temperature range (°C); and ø is degrees latitude. The scale provides values approximating to nearly zero for Thorshavn (Faroe: 62°2′N, 6°4′W) and almost 100 for Verkoyansk (Russia: 67°33′E, 133°24′E).

The ecological hazards of oceanicity for some types of vegetation have previously been reviewed and related to values of Conrad’s Index averaged over the period 1961–1990 (Crawford, 2000). This present paper extends the study to cover the past century using the 0·5° gridded monthly temperature data for the years 1901–1998 (New et al., 1999, 2000) and covering Europe and the Arctic from northern Norway to Siberia. Amounts of change differ between one decade and another, and a substantial part of what appears to be two major cyclical periods with about 60 years duration is evident in the decadal means for northern regions (Fig. 4A–D). They also correlate broadly with the changes in the NAO (Fig. 5A and B; Table 1). During the warmer periods in these cycles, Conrad’s Index in northern Siberia is reduced (more oceanic), and during the colder periods Conrad’s Index is increased (more continental). The warm period from 1991 to 1998 (the most recent data available at the time of writing) has been more oceanic than any other decade in the past century. Two examples of the spatial distribution of the changing pattern of inter‐decadal oceanicity are shown in Fig. 6. The oceanic areas of Norway and Britain show low‐amplitude decadal changes in oceanicity but nevertheless, within these lower values, there is greater year‐to‐year variation than in the more continental regions as measured by the coefficient of variation (Fig. 6D).

Examination of precipitation in Europe over the period 1986–1995 (Fig. 7A) shows the expected pattern of a decrease in moving from west to east. The corresponding precipitation anomaly map (Fig. 7B) indicates that there has been a tendency for a northward movement of the rain belts with a significant reduction in rainfall amounts in parts of the Mediterranean, western Iberian peninsula, the Balkans and Turkey. When the number of rain‐days in the year is examined (Fig. 7C) it can be seen that the highest totals are as expected in the hyper‐oceanic regions of Ireland, Scotland and western Norway. However, it is also notable that the oceanic regions of western Siberia have 210–240 rain days per annum (Fig. 7C) and that the ratio of summer to winter precipitation (Fig. 7D) shows that more rain falls in summer than winter. In these regions, both the number of rain days per annum and the amount of precipitation have increased since about 1950 compared with the earlier half of the century. The region has been showing a positive anomaly of the order of one rain‐day per month and a precipitation anomaly of about 0·1 mm per day since 1980, but this trend cannot always be extended as a generalization to other parts of Europe, which show considerable individuality in the patterns of climate change during the century (Table 1).

Given these observations it would appear that there is a sufficient corpus of information to suggest that the meteorological conditions which created the extensive bogs that have developed over the past 6000 years in western Europe and in the western Siberian lowlands have not diminished and in some areas are clearly increasing. Given that climate warming in Europe and the Eurasian Arctic is correlated with increasing oceanicity, there are therefore grounds for considering that paludification is likely to increase and be a dominant feature controlling natural plant succession both on the Atlantic seaboard and at the tundra–taiga interface in the Siberian lowlands.

PHYSIOLOGICAL HAZARDS FOR FLOODED TREES

Adaptations to flooding in plants are usually considered as either avoidance mechanisms or else tolerance adaptations. In the former, aerenchyma facilitates aeration of the inundated root. In the latter, metabolic adaptations have been found which allow some plants to endure anaerobic conditions for a length of time sufficient to overcome the period of oxygen deprivation caused by flooding. Both these aspects of flooding have been extensively discussed (Armstrong et al., 1994; Braendle and Crawford, 1999) and are by no means mutually exclusive as there are examples of flood‐tolerant species which employ both types of adaptations. Spartina anglica can survive total anoxia for more than 28 days (Crawford, 1989) and is also capable of allowing aeration of roots when growing in the lower parts of salt marshes without developing particularly large root diameters (Bouma et al., 2001). Similarly, when flooded, the common osier (Salix viminalis) is able to aerate upper adventitious roots, while deeper roots (200–300 mm) rely on anoxia‐tolerance for their survival (Jackson and Attwood, 1996). However, in most trees, particularly as they grow in size, aeration of the deeper anchoring roots is not a facility that is readily available. Trees with large trunks and deep anchoring roots represent the ultimate challenge in withstanding oxygen‐deprivation in wetland habitats.

Outside the tropics most of these ultra‐flood‐tolerant swamp forests usually have either a relatively short winter or no winter, and trees do not have the stress of preserving an extensive root system in anaerobic conditions throughout long, non‐productive periods. In the cold and cool‐temperate regions of the world, flooding is generally unfavourable for tree survival unless the ground is frozen during the period of potential inundation. The few woody species that survive in the wetlands at high latitudes are generally bushes and scrub of a limited number of genera in which Salix spp. and Alnus spp. are probably the most successful. In dry habitats alders will grow as pole trees, but when flooded, the basal buds are stimulated to develop and the bush form predominates (Crawford, 1989). The greater area of stem surface that this produces close to the ground facilitates aeration. The bush or polycormic form is therefore possibly adapted better, in terms of aeration potential, as the area of lenticel‐containing bark in the basal region of the stems is greater. These stem adaptations combined with the presence of adventitious roots as in willow, and negatively geotropic roots in nitrogen‐fixing alders, facilitate the supply of oxygen in waterlogged soils.

In any discussion of the physiology of oxygen deprivation in perennial plants it is important to distinguish between short‐term and long‐term tolerance. Species that are only tolerant of short periods of anoxia frequently accelerate glycolysis and thus overcome temporary energy shortages. By contrast, species that are able to endure long‐term anoxia of oxygen deprivation have been shown to down‐regulate their metabolism (Schlüter and Crawford, 2001). Similarly, the common osier Salix viminalis has been found to respond to decreasing oxygen supply during the growing season by reducing both shoot and root growth. When rooted cuttings were waterlogged for 4 weeks in a glasshouse, soil redox potentials quickly decreased to below zero. At depths of 100–200 mm and 200–300 mm, extension by existing root axes was halted by soil flooding, while adventitious roots from above failed to penetrate these deeper zones. After 4 weeks waterlogging, all arrested root tips recommenced elongation when the soil was drained; their extension rates exceeding those of roots that were well drained throughout the experiment (Jackson and Attwood, 1996).

Trees appear to be most at risk from flooding in oceanic climates where winters are long, wet and not particularly cold. Further north, in some of the coldest regions of the boreal forest, Picea mariana and Larix dahurica can grow on bog surfaces even when permafrost is not far below the surface. However, as noted above, in all these situations the soils are frozen for most of the winter, the roots are relatively shallow and oxygen demand is lower and supply less impeded (Crawford, 1992). An entirely different winter situation is found in northern oceanic regions such as the British Isles and parts of western Norway. Tree roots stay metabolically active and continue growing after shoot growth has ceased, and only become dormant after the onset of low temperatures (Coutts and Philipson, 1987).

Actively growing root tips of Picea sitchensis are highly susceptible to damage if waterlogged but develop some tolerance after they stop growing (Nicoll and Coutts, 1998). Experimental studies of over‐wintering trees of P. sitchensis have shown that flooding under milder winter conditions, while the roots are still growing, causes a marked decrease in root carbohydrate reserves. This does not lead to the immediate death of the root, but when carbohydrate‐depleted roots are exposed once again to air, when the water table is lowered, there is extensive die back of the root system presumably due to post‐anoxic injury (Crawford and Braendle, 1996). Death of the whole tree will not be immediate, but reduced root development will make the larger trees unstable and therefore prone to wind‐blow.

Trees which have an active sap flow in early spring have a need for adequate over‐wintering supplies of carbohydrate. In birches, the active upward movement of xylem sap in spring carries with it substantial quantities of soluble carbohydrates, ethanol, organic and amino acids (Crawford, 1996). This flow has been suggested as the means whereby roots of some woody plants compensate for the lack of oxygen in the soil atmosphere by exporting hydrogen to the shoot and thus maintaining the redox balance of the inundated root system (Crawford, 1996). The upward movement of hydrogen in reduced compounds as a constituent of the xylem flow is a more efficient means of exporting the oxygen debt of submerged organs than the slow downward diffusion of gaseous oxygen through more than 1 m of woody tissue. This replacement of oxygen import by hydrogen export does, however, have a metabolic cost, namely the provision of a high carbohydrate store in the over‐wintering root. In the bottomland trees of the USA it has been shown that high pre‐flood root starch concentrations are associated in those species that can tolerate prolonged inundation. The bottomland Fraxinus pennsylvanica (green ash) and flood‐tolerant Nyssa aquatica (water tupelo) store more carbohydrate in their roots and retain less in their leaves than the less flood‐tolerant Quercus alba (Gravatt and Kirby, 1998). Regeneration of bottomland forests is dependent on the trees having the facility to pursue their natural amphibious life style. The trees of bottomland forests in North America are highly dependent on intermittent periods of low water levels for regeneration. In the past it was sufficient for this to take place once every 20–30 years. However, improved river‐level regulation has removed even this amount of occasional draw‐down and now seriously threatens the regeneration of these forests in several areas such as the Mississippi Basin and the swamp forests of Louisiana. Even the recruitment of such flood‐tolerant trees as the swamp cypress (Taxodium distichum) and the swamp tupelo (Nyssa sylvatica) is prevented when there is no relief from constant flooding (Conner et al., 1981). Similarly, it is predicted that higher flood levels on the Rhine will reduce the establishment of hardwood trees species in the low‐lying sites in this river’s flood‐plain forests (Siebel et al., 1998). The Amazonian tidal swamp forest (Varzea) provides another example of the variation in response between species in response to wet and dry seasons. A study of carbohydrate levels at the end of the dry season showed that most species had high levels of starch. The exceptions were the palms and the most aquatic of the other woody species, for which the dry season was the stress period and the wet season the period when they accumulated carbohydrate reserves (Scarano et al., 1994).

In all the flood‐tolerant forests so far examined it would appear that possession of adequate carbohydrate reserves is a prerequisite for flooding tolerance. It therefore follows that the risk of carbohydrate starvation is likely to be greater if winters are warmer and wetter. First, warmer winters will delay the onset of root dormancy and, secondly, wetter winters will be likely to cause non‐dormant roots to accelerate their carbohydrate draw‐down due to the hypoxia‐induced increased glycolytic activity that normally affects plants not adapted to long‐term anoxia‐tolerance (Braendle and Crawford, 1999). Carbohydrate starvation will not only deprive plants of energy reserves and create a so‐called ‘energy crisis’, but will also reduce the ability of the tree to synthesize compounds such as ascorbic acid, which together with other anti‐oxidants are necessary for preventing tissue damage when flood‐waters subside, exposing plants to the risk of post‐anoxic injury. Carbohydrate metabolism is intimately linked with ascorbate production (Smirnoff and Wheeler, 2000) and starved roots are likely to be less able to defend themselves against post‐anoxic injury.

CONCLUSIONS

Holocene paludification and extensive forest retreat in northern and oceanic regions has taken place over the last 6000 years from northern Québec through Newfoundland and Scandinavia to Arctic Siberia. This has also been a period over which the climate has become more oceanic in these areas. It is evident from the results presented here that a notable effect of episodes of climatic warming during the last century as a whole has been to increase oceanicity in northern maritime areas. Climatic warming beyond that which has occurred during the last two decades is therefore likely to result in additional increases in precipitation, and rainfall frequency and further reduction in temperature range in northern maritime regions. Such changes can be expected to increase bog growth. Peat accumulation raises the water table (paludification) and, although there may be periodic episodes of drying out and forest re‐advance, the underlying peat will remain and can be expected to respond with renewed growth to further periods of oceanicity. The northward advance of forest that took place during the early Holocene as a response to climatic warming will therefore be unlikely to be repeated as in many areas bogs that have formed in the intervening period will hinder forest expansion. There may even be a retreat south of the tundra–taiga interface as more forest in engulfed by further bog expansion.

ACKNOWLEDGEMENTS

This research has been supported by an Emeritus Fellowship from the Leverhulme Foundation (R.M.M.C.) which is gratefully acknowledged. Research on the tundra–taiga interface has been supported by the Royal Swedish Academy of Sciences and the International Arctic Science Committee. Figure 1 was drawn with assistance from Dr B. R. Werkman. The authors thank the Distributed Active Archive Centre (Code 902.2) at the Goddard Space Flight Center, Greenbelt, MD 20771, USA for producing and distributing the Pathfinder data in their present form. The original data products were produced under the NOAA/NASA Pathfinder Program, by a processing team headed by Ms Mary James of the Goddard Global Change Data Center and the science algorithms were established by the AVHRR Land Science Working Group, chaired by Dr John Townshend of the University of Maryland. Goddard’s contributions to these activities were sponsored by NASA’s Mission to Planet Earth Program.

Fig. 1. Comparison of present and past northern limits for tree survival in northern Siberia. Present‐day distribution of the Boreal forest (brown) is based on the vegetation map produced by Grid Arendal and published by the World Wide Fund for Nature. Mid‐Holocene limits to forest trees are regional generalizations from locations of fossil remains (green, evergreen—pine and/or spruce species; red, tree—birch; purple, larch) based on Kremenetski et al. (1998). Modern limits for the northern survival of individual tree species (colours as above) are also taken from Kremenetski et al. (1998) as drawn by Callaghan et al., 2002.

Fig. 1. Comparison of present and past northern limits for tree survival in northern Siberia. Present‐day distribution of the Boreal forest (brown) is based on the vegetation map produced by Grid Arendal and published by the World Wide Fund for Nature. Mid‐Holocene limits to forest trees are regional generalizations from locations of fossil remains (green, evergreen—pine and/or spruce species; red, tree—birch; purple, larch) based on Kremenetski et al. (1998). Modern limits for the northern survival of individual tree species (colours as above) are also taken from Kremenetski et al. (1998) as drawn by Callaghan et al., 2002.

Fig. 2. Northern limits to the North American Boreal Forest: (A) proximity of oceanic influences from the Hudson Bay and Labrador Sea; (B) Normalized Difference Vegetation Index (NDVI) image recorded in May 1998 (8 km resolution) (colour scale: blue, 0; dark‐green, 0·48–0·64; yellow–orange, 0·76–0·88). Part A was generated from a Royalty Free Program ‘Mountain High Maps’ published by Digital Wisdom, Cambridge; part B from the 8‐km resolution Pathfinder data set.

Fig. 2. Northern limits to the North American Boreal Forest: (A) proximity of oceanic influences from the Hudson Bay and Labrador Sea; (B) Normalized Difference Vegetation Index (NDVI) image recorded in May 1998 (8 km resolution) (colour scale: blue, 0; dark‐green, 0·48–0·64; yellow–orange, 0·76–0·88). Part A was generated from a Royalty Free Program ‘Mountain High Maps’ published by Digital Wisdom, Cambridge; part B from the 8‐km resolution Pathfinder data set.

Fig. 3. Northern limits to the Eurasian Boreal Forest: (A) proximity of oceanic influences from the Arctic Ocean; (B) Normalized Difference Vegetation Index (NDVI) image recorded in May 1998 (8 km resolution) (colour scale: blue, 0; dark green, 0·48–0·64; yellow–green, 0·76–0·88); (C) detail of the transition zone between forest and tundra as seen in the Normalized Vegetation Index recorded in May 1998 (1 km resolution) (colour scale: blue, 0; blue–green, 0·11–0·25; dark‐green, 0·26–0·40; bright green, 0·41–0·64). It is considered (see text) that this mosaic represents a self‐renewing cyclic process taking place over hundreds of years as patches of forest develop on land that dries out after being raised by frost‐heave and then reverts again to bog as tree‐cover cools the underlying ground. Part A was generated from a Royalty Free Program ‘Mountain High Maps’ published by Digital Wisdom, Cambridge; part B from the 8‐km resolution Pathfinder data set; part C from the 1‐km AVHRR global land data set distributed by the EROS Data Center, Distributed Active Archive Center (EDC DAAC), at the US Geological Survey’s EROS Data Center in Sioux Falls, South Dakota.

Fig. 3. Northern limits to the Eurasian Boreal Forest: (A) proximity of oceanic influences from the Arctic Ocean; (B) Normalized Difference Vegetation Index (NDVI) image recorded in May 1998 (8 km resolution) (colour scale: blue, 0; dark green, 0·48–0·64; yellow–green, 0·76–0·88); (C) detail of the transition zone between forest and tundra as seen in the Normalized Vegetation Index recorded in May 1998 (1 km resolution) (colour scale: blue, 0; blue–green, 0·11–0·25; dark‐green, 0·26–0·40; bright green, 0·41–0·64). It is considered (see text) that this mosaic represents a self‐renewing cyclic process taking place over hundreds of years as patches of forest develop on land that dries out after being raised by frost‐heave and then reverts again to bog as tree‐cover cools the underlying ground. Part A was generated from a Royalty Free Program ‘Mountain High Maps’ published by Digital Wisdom, Cambridge; part B from the 8‐km resolution Pathfinder data set; part C from the 1‐km AVHRR global land data set distributed by the EROS Data Center, Distributed Active Archive Center (EDC DAAC), at the US Geological Survey’s EROS Data Center in Sioux Falls, South Dakota.

Fig. 4. Changes of annual mean temperature, Conrad’s Index and annual temperature range in the Siberian Arctic (lat 60–85°N, long 10–180°E) and of annual mean temperature in the northern hemisphere temperature compared with the 1961–1990 means—all curves Lowess smoothed. (A) Decadal changes of annual mean temperature; (B) decadal changes of Conrad’s Index anomaly; (C) decadal changes of annual temperature range; (D) annual mean temperature anomalies from the 1961–1990 means for the northern hemisphere, 1901–1998. Temperature data from the Climatic Research Unit 0·5° gridded 1901–1995 Global Climate Dataset (New et al., 1999, 2000).

Fig. 4. Changes of annual mean temperature, Conrad’s Index and annual temperature range in the Siberian Arctic (lat 60–85°N, long 10–180°E) and of annual mean temperature in the northern hemisphere temperature compared with the 1961–1990 means—all curves Lowess smoothed. (A) Decadal changes of annual mean temperature; (B) decadal changes of Conrad’s Index anomaly; (C) decadal changes of annual temperature range; (D) annual mean temperature anomalies from the 1961–1990 means for the northern hemisphere, 1901–1998. Temperature data from the Climatic Research Unit 0·5° gridded 1901–1995 Global Climate Dataset (New et al., 1999, 2000).

Fig. 5. North Atlantic Oscillation (NAO) anomalies for 1900–2000 (grey columns with superimposed lowess line in black: data from Hurrell, 1995) compared with anomalies of annual mean temperature (°C) (green) and annual temperature range (°C) (red) for (A) a restricted part of the western Siberian lowlands (lat 62·5–67·5°N, long. 62·5–67·5°E, showing a significant relationship between NAO and the annual mean temperature. The temperature range anomaly is typically of opposite sign to the mean anomaly. (B) Spain (lat 36–44°) showing weak relationships between NAO index and mean temperature but a stronger relationship between NAO and annual temperature range.

Fig. 5. North Atlantic Oscillation (NAO) anomalies for 1900–2000 (grey columns with superimposed lowess line in black: data from Hurrell, 1995) compared with anomalies of annual mean temperature (°C) (green) and annual temperature range (°C) (red) for (A) a restricted part of the western Siberian lowlands (lat 62·5–67·5°N, long. 62·5–67·5°E, showing a significant relationship between NAO and the annual mean temperature. The temperature range anomaly is typically of opposite sign to the mean anomaly. (B) Spain (lat 36–44°) showing weak relationships between NAO index and mean temperature but a stronger relationship between NAO and annual temperature range.

Fig. 6. Decadal variations in Conrad’s Continentality Index (CI) in Europe (lat 11–66°N, long 34–72°E). (A) Distribution of CI for 1991–1998. (B–D) CI anomalies with respect to the 1961–1990 mean: (B) CI anomaly for 1941–1950; (C) CI anomaly for 1991–1998; (D) distribution of coefficients of variation (CV) of Conrad’s Index during the 9‐year period 1990–1998. Colour scale intervals chosen to maximize contrasts.

Fig. 6. Decadal variations in Conrad’s Continentality Index (CI) in Europe (lat 11–66°N, long 34–72°E). (A) Distribution of CI for 1991–1998. (B–D) CI anomalies with respect to the 1961–1990 mean: (B) CI anomaly for 1941–1950; (C) CI anomaly for 1991–1998; (D) distribution of coefficients of variation (CV) of Conrad’s Index during the 9‐year period 1990–1998. Colour scale intervals chosen to maximize contrasts.

Fig. 7. Distribution of precipitation amount, frequency and seasonality in Europe: (A) total annual precipitation (mm) during 1986–1995; (B) annual precipitation anomaly 1986–1995 with respect to the 1961–1990 30‐year mean; (C) rain‐days, 1986–1995; (D) the ratio of summer (April–September) to winter (October–March) precipitation. Temperature and precipitation data from the CRU 1901 to 1995 Global Climate Dataset (New et al., 1999, 2000).

Fig. 7. Distribution of precipitation amount, frequency and seasonality in Europe: (A) total annual precipitation (mm) during 1986–1995; (B) annual precipitation anomaly 1986–1995 with respect to the 1961–1990 30‐year mean; (C) rain‐days, 1986–1995; (D) the ratio of summer (April–September) to winter (October–March) precipitation. Temperature and precipitation data from the CRU 1901 to 1995 Global Climate Dataset (New et al., 1999, 2000).

Table 1.

Matrix of Pearson correlations, r (n = 98), between anomalies of the NAO index and of the northern hemisphere annual mean temperature with annual mean temperature and annual temperature range of four European regions—Siberian lowlands (lat 60–70°, long 65–85°), British Isles (lat 49–61°, long –10° to –3°), Turkey (lat 36–44°, long 27–32°) and Spain (lat 36–44°, long –9° to 4°)

   NAO index anomaly Northern hemisphere mean temperature anomaly 
Siberia Mean temperature anomaly  r 0·384 0·543 
  P 0·000096 0·0000000074 
 Temperature range anomaly r –0·277 –0·321 
  P 0·0057 0·0013 
British Isles Mean temperature anomaly r –0·290 0·0799 
  P 0·0038 0·43 (n.s.) 
 Temperature range anomaly r 0·482 0·413 
  P 0·00000049 0·000024 
Turkey Mean temperature anomaly r –0·198 0·192 
  P 0·051 (n.s.) 0·058 (n.s.) 
 Temperature range anomaly r 0·170 –0·0177 
  P 0·095 (n.s.) 0·86 (n.s.) 
Spain Mean temperature anomaly r 0·167 0·571 
  P 0·10 (n.s.) 0·00000000083 
 Temperature range anomaly r 0·0629 0·0827 
  P 0·54 (n.s.) 0·42 (n.s.) 
   NAO index anomaly Northern hemisphere mean temperature anomaly 
Siberia Mean temperature anomaly  r 0·384 0·543 
  P 0·000096 0·0000000074 
 Temperature range anomaly r –0·277 –0·321 
  P 0·0057 0·0013 
British Isles Mean temperature anomaly r –0·290 0·0799 
  P 0·0038 0·43 (n.s.) 
 Temperature range anomaly r 0·482 0·413 
  P 0·00000049 0·000024 
Turkey Mean temperature anomaly r –0·198 0·192 
  P 0·051 (n.s.) 0·058 (n.s.) 
 Temperature range anomaly r 0·170 –0·0177 
  P 0·095 (n.s.) 0·86 (n.s.) 
Spain Mean temperature anomaly r 0·167 0·571 
  P 0·10 (n.s.) 0·00000000083 
 Temperature range anomaly r 0·0629 0·0827 
  P 0·54 (n.s.) 0·42 (n.s.) 

References

ArmstrongW, Braendle R and Jackson MB.
1994
. Mechanisms of flood tolerance in plants.
Acta Botanica Neerlandica
 
43
:
307
–358.
ArseneaultDand Payette S.
1997
. Landscape change following deforestation at the Arctic tree line in Quebec, Canada.
Ecology
 
78
:
693
–706.
BennettKD, Boreham S, Sharp MJ and Switsur VR.
1992
. Holocene history of environment, vegetation and human settlement on Catta Ness, Lunnasting, Shetland.
Journal of Ecology
 
80
:
241
–273.
BlyakharchukTAand Sulerzhitsky LD.
1999
. Holocene vegetational and climatic changes in the forest zone of western Siberia according to pollen records from the extrazonal palsa bog Bugristoye.
Holocene
 
9
:
621
–628.
BoumaTJ, Nielsen KL, van Hal J and Koutataal B.
2001
. Root system topology and diameter distribution of species from habitats differing in inundation frequency.
Functional Ecology
 
15
:
360
–369.
BraendleRand Crawford RMM.
1999
. Plants as amphibians.
Perspectives in Plant Evolution and Systematics
 
2
:
56
–78.
BrysonRA.
1966
. Air masses, streamlines, and the boreal forest.
Geographical Bulletin
 
8
:
228
–269.
BuntingMJ.
1996
. The development of heathland in Orkney, Scotland – pollen records from Loch‐of‐Knitchen (Rousay) and Loch‐of‐Torness (Hoy).
Holocene
 
6
:
193
–212.
ChernovYIand Matveyeva NV.
1997
. Arctic ecosystems in Russia. In: Wielgolaski FE, ed. Polar and alpine tundra. Amsterdam: Elsevier,
361
–507.
ConnerWH, Gosselink JG and Parrondo RT.
1981
. Comparison of the vegetation of three Louisiana swamp sites with different flooding regimes.
American Journal of Botany
 
68
:
320
–331.
ConradV.
1946
. Usual formulas of continentality and their limits of validity.
Transactions of the American Geophysical Union
 
27
:
663
–664.
CouttsMPand Philipson JJ.
1987
. Structure and physiology of Sitka spruce roots.
Proceedings of the Royal Soociety of Edinburgh
 93B:
131
–144.
CrawfordRMM.
1989
. Studies in plant survival. Blackwell Scientific Publications, Oxford.
CrawfordRMM.
1992
. Oxygen availability as an ecological limit to plant distribution.
Advances in Ecological Research
 
23
:
93
–185.
CrawfordRMM.
1996
. Whole plant adaptations to fluctuating water tables.
Folia Geobotanica and Phytotaxonomica
 
31
:
7
–24.
CrawfordRMM.
2000
. Ecological hazards of oceanic environments.
New Phytologist
 
147
:
257
–281.
CrawfordRMMand Braendle R.
1996
. Oxygen deprivation stress in a changing climate.
Journal of Experimental Botany
 
47
:
145
–159.
EdwardsMEand Barker ED.
1994
. Climate and vegetation in Northeastern Alaska 18,000 Yr Bp–present.
Palaeogeography, Palaeoclimatology and Palaeoecology
 
109
:
127
–135.
FerreiraLV.
2000
. Effects of flooding duration on species richness, floristic composition and forest structure in river margin habitat in Amazonian blackwater floodplain forests: implications for future design of protected areas.
Biodiversity and Conservation
 
9
:
1
–14.
FossittJA.
1996
. Late quaternary vegetation history of the Western Isles of Scotland.
New Phytologist
 
132
:
171
–196.
GerloffLM, Hills LV and Osborn GD.
1995
. Postglacial vegetation history of the Mission Mountains, Montana.
Journal of Paleolimnology
 
14
:
269
–279.
GirardinMP, Tardif J and Bergeron Y.
2001
. Radial growth analysis of Larix laricina from the Lake Duparquet area, Quebec, in relation to climate and larch sawfly outbreaks.
Ecoscience
 
8
:
127
–138.
GravattDA and Kirby CJ.
1998
. Patterns of photosynthesis and starch allocation in seedlings of four bottomland hardwood tree species subjected to flooding.
Tree Physiology
 
18
:
411
–417.
HolbenB, Tucker CJ and Fan C‐J.
1980
. Spectral assessment of soybean leaf area and leaf biomass.
Photogrammetric Engineering and Remote Sensing
 
46
:
651
–656.
HoltmeierF‐K.
2000
. Die Höhengrenze der Gebirgswälde. Solingen: Verlag Natur and Wissenschaft.
HurrellJW.
1995
. Decadal trends in the North Atlantic Oscillation: regional temperatures and precipitation.
Science
 
269
:
676
–679.
HurrellJWand Van Loon H.
1997
. Decadal variations in climate associated with the north Atlantic oscillation.
Climatic Change
 
36
:
301
–326.
JacksonMBand Attwood PA.
1996
. Roots of willow (Salix viminalis L.) show marked tolerance to oxygen shortage in flooded soils and in solution culture.
Plant and Soil
 
187
:
37
–45.
KalandPE.
1986
. The origin and management of Norwegian coastal heathlands as reflected by pollen analysis. In: Behre K‐E, ed. Anthropogenic indicators in pollen diagrams. Rotterbal: Balkema,
19
–36.
Klitgaard‐KristensenD, Sejrup HP, Haflidason H, Johnsen S and Spurk M.
1998
. A regional 8200 cal. yr BP cooling event in northwest Europe, induced by final stages of the Laurentide ice‐sheet deglaciation?
Journal of Quaternary Science
 
13
:
165
–169.
KörnerC.
1999
. Alpine plant life. Berlin: Springer Verlag.
KremenetskiCV, Sulerzhitsky LD and Hantemirov R.
1998
. Holocene history of the northern range limits of some trees and shrubs in Russia.
Arctic and Alpine Research
 
30
:
317
–333.
KriuchkovVV.
1971
. Woodland communities in tundra, possibilities for its establishment and dynamics.
Ekologiya (Soviet Journal of Ecology)
 
6
:
9
–19 (in Russian).
LavoieCand Payette S.
1994
. Recent fluctuations of the lichen‐spruce forest limit in subarctic Québec.
Journal of Ecology
 
82
:
725
–734.
Lescop‐SinclairKand Payette S.
1995
. Recent advance of the arctic treeline along the eastern coast of Hudson Bay.
Journal of Ecology
 
83
:
929
–936.
MacDonaldGM, Velichko AA, Kremenetski CV, Borisova OK, Goleva AA, Andreev AA, Cwynar LC, Riding RT, Forman SL, Edwards TWD, Aravena R, Hammarlund D, Szeicz JM and Gattaulin VN.
2000
. Holocene treeline history and climate change across northern Eurasia.
Quaternary Research
 
53
:
302
–311.
MallikAU, Gimingham CH and Rahman AA.
1984
. Ecological effects of heather burning. 1. Water infiltration, moisture retention and porosity of surface soil.
Journal of Ecology
 
72
:
767
–776.
MauquoyDand Barber K.
1999
. A replicated 3000 year proxy‐climate record from Coom Rigg Moss and Felecia Moss, the Border Mires, northern England.
Journal of Quaternary Science
 
14
:
263
–275.
NesjeAand Kvamme M.
1991
. Holocene glacier and climate variations in western Norway – evidence for early Holocene glacier demise and multiple neoglacial events.
Geology
 
19
:
610
–612.
NewM, Hulme M and Jones P.
1999
. Representing twentieth‐century space‐time climate variability. Part I. Development of a 1961–1990 mean monthly terrestrial climatology.
Journal of Climate
 
12
:
829
–856.
NewM, Hulme M and Jones P.
2000
. Representing twentieth‐century space‐time climate variability. Part II. Development of 1901–1996 monthly grids of terrestrial climate.
Journal of Climate
 
13
:
2217
–2238.
NicollBCand Coutts MP.
1998
. Timing of root dormancy and tolerance to root waterlogging in clonal Sitka spruce.
Trees—Structure and Function
 
12
:
241
–245.
PayetteS.
1984
. Peat inception and climatic change in northern Québec. In: Karlén W, ed. Climatic change on a yearly to millennial basis. Dordrecht: Reidel,
173
–179.
PayetteSand Gagnon R.
1985
. Late holocene deforestation and tree regeneration in the forest‐tundra of Quebec.
Nature
 
313
:
570
–572.
PayetteS, Bhiry N, Delwaide A and Simard M.
2000
. Origin of the lichen woodland at its southern range limit in eastern Canada: the catastrophic impact of insect defoliators and fire on the spruce‐moss forest.
Canadian Journal of Forest Research
 
30
:
288
–305.
PeteetD, Andreev A, Bardeen W and Mistretta F.
1998
. Long‐term Arctic peatland dynamics, vegetation and climate history of the Pur‐Taz region, western Siberia.
Boreas
 
27
:
115
–126.
PorterSC, Sauchyn DJ and Delorme LD.
1999
. The ostracode record from Harris Lake, southwestern Saskatchewan: 9200 years of local environmental change.
Journal of Paleolimnology
 
21
:
35
–44.
RoyV, Bernier PY, Plamondon AP and Ruel JC.
1999
. Effect of drainage and microtopography in forested wetlands on the microenvironment and growth of planted black spruce seedlings.
Canadian Journal of Forest Research
 
Revue Canadienne De Recherche Forestiere
 
29
:
563
–574.
ScaranoFR, Cattanio JH and Crawford RMM.
1994
. Root carbohydrate storage in young saplings of an Amazonian tidal varzea forest before the onset of the wet season.
Acta Botanica Brasilica
 
8
:
129
–139.
SchlüterUand Crawford RMM.
2001
. Long‐term anoxia tolerance in leaves of Acorus calamus L. and Iris pseudacorus L.
Journal of Experimental Botany
 
52
:
2213
–2225.
SiebelHN, Van Wijk M and Blom C.
1998
. Can tree seedlings survive increased flood levels of rivers?
Acta Botanica Neerlandica
 
47
:
219
–230.
SkreO, Baxter R, Crawford R, Callaghan TV and Fedorcov A.
2002
. How will the tundra‐taiga interface respond to climate change?
Ambio
  Special Report No. 12, Tundra‐Taiga Treeline Research, 37–46.
SmirnoffNand Wheeler GL.
2000
. Ascorbic acid in plants: biosynthesis and function.
Critical Reviews in Biochemistry and Molecular Biology
 
35
:
291
–314.
SveinbjörnssonB.
2000
. North American and European treelines: external forces and internal processes controlling position.
Ambio
 
29
:
388
–395.
TippingR.
1994
. The form and fate of Scotland’s woodlands.
Proceedings of the Society of Antiquaries of Scotland
 
124
:
1
–54.
TuckerCJ.
1979
. Red and photographic infrared linear combinations for monitoring vegetation.
Remote Sensing of Environment
 
8
:
127
–150.
TuckerCJ, Miller LD and Pearson RL.
1975
. Shortgrass prairie spectral measurements.
Photogrammetric Engineering and Remote Sensing
 
41
:
1175
–1183.
ValentiniR, Dore S, Marchi G, Mollicone D, Panfyorov M, Rebman C, Kolle O and Schulze E‐D.
2000
. Carbon and water exchanges of two contrasting central Siberia landcape types: regenerating forest and bog.
Functional Ecology
 
14
:
87
–96.
VelichkoAA, Kremenetski CV, Borisova OK, Zelikson EM, Nechaev VP and Faure H.
1998
. Estimates of methane emission during the last 125,000 years in Northern Eurasia.
Global and Planetary Change
 
17
:
159
–180.
WhiteJWC, Barlow LK, Fisher D, Grootes P, Jouzel J, Johnsen SJ, Stuiver M and Clausen H.
1997
. The climate signal in the stable isotopes of snow from Summit, Greenland: results of comparisons with modern climate observations.
Journal of Geophysical Research—Oceans
 
102
:
26425
–26439.

Author notes

1Plant Science Laboratory, Sir Harold Mitchell Building, St Andrews University, St Andrews KY1 6AJ, 2Edinburgh University Institute of Cell and Molecular Biology, Daniel Rutherford Building, King’s Buildings, Mayfield Road, Edinburgh EH9 3JH and 3Scott Polar Research Institute, Lensfield Road, Cambridge CB2 1ER, UK

Comments

0 Comments