Europe Europe Europe

Forestry and Peatlands: European scale

Vulnerabilities forests

Drought and heat-induced tree mortality is accelerating in many forest biomes as a consequence of a warming climate, resulting in a threat to global forests unlike any in recorded history (8). It has been stated that tall plants with low hydraulic conductance and high leaf area are most likely to die from future drought stress, implying that tall trees of old-growth forests are at the greatest risk of loss (9). Thus, it was concluded with high certainty that today’s forests are going to be subject to continued increases in mortality rates that will result in substantial reorganization of their structure and carbon storage. As a result, today’s forests generally should be replaced by shorter and more xeric plants, owing to future warmer droughts and associated wildfires and pest attacks (9). Forest types are expected to shift northward, which may have substantial negative impacts on forest economy, and alter species composition in natural forest stands (66).

Temperate forests

Increasing temperatures, longer growing seasons, higher atmospheric CO2 concentrations, and in the north, increasing N-mineralisation, are likely to increase the potential forest productivity where summer precipitation does not decline (7,19). It is uncertain to what extent this potential can be realised as forests will increasingly face a climate to which the planted species or provenances are not adapted, which might increase their susceptibility to pests and pathogens, such as bark beetle outbreaks, which can lead to major forest die-back events particularly in Norway spruce stands (20). 

Several studies find that tree growth rates in temperate forests passed their peak in the late 20th century and that the decline in tree growth rates can be attributed to climatic factors, especially drought or heatwaves (1). Extreme climate events have had a major impact on temperate forests over the last decade (2). Many species will likely suffer from drought stress at the southern edge of their current distribution in Europe and will probably shift their range northward (18). The complex interactions between climate and forest management in determining susceptibility to extreme events make it difficult to unequivocally attribute these events to recent climate warming (3).

Vulnerabilities peatlands

Peatlands are areas with a naturally accumulated peat layer at the surface, mires are peatlands where peat is currently being formed (21). About 80 % of the peatlands in Sweden and Norway are still mires. In countries such as the Netherlands and the UK, the contribution of mires to the total peatland area is only 1 - 7 % due to widespread drainage and intensive land use (21).

Although peatlands cover only 3-4% of the earth land surface, they hold the equivalent of half of the atmosphere’s carbon. Peatland areas are decreasing worldwide. In Europe, peatlands (> 30 cm peat) cover 5.4% of the total surface area. In about 54% of these areas peat is now being formed (mires). If shallow-peat lands (< 30 cm peat) in European Russia are also taken into account, the total peatland area in Europe is almost 10 % of the total surface area. Finland is the country in Europe with most peatland (26.7 % of the land area).The extent of mire is highest in European Russia (53).

Drained and degraded peatlands are hotspots of greenhouse gas emissions (CO2, methane (CH4), nitrous oxide (N2O) and thus influence climate (22). Mineralisation of peat organic matter and the respective CO2 emissions are strongly related to drainage depth and management intensity (23). Higher temperatures will generally increase microbial peat decomposition and carbon mobilisation (24). However, change in land use is the primary driver of changes in peatland hydrology, and probably has a stronger impact than climate change (25).

Greenhouse gas emissions from degraded and agriculturally-used peatlands can be significant when compared to their total national greenhouse gas emissions, with contributions of about 5 % in Germany and Denmark, 2–3 % in the Netherlands and about 1 % in the UK (26).

Models predict considerable increases in CH4 emissions from peatlands over the coming century due to warming as long as wetland area and soil moisture conditions remain unchanged (27). However, wetland area may decrease in response to higher temperatures, leading to lower CH4 emissions (28). Peatland water tables may become significantly lower due to increased evapotranspiration and/or decreased summer precipitation. In that case, peat decomposition and the release of CO2 may be enhanced (29), and so will N2O release (30), and CH4 production and emission will decrease (31).

Peat soil degradation causes land subsidence by a combination of peat oxidation and compaction after drainage. Historical subsidence, caused by drainage since medieval times, often combined with peat extraction for fuel, in coastal peatlands of the Netherlands, Germany and eastern Britain may have resulted in up to several metres of subsidence (32). In the eastern British fenlands, compaction and peat oxidation has resulted in up to 4 m of subsidence in 150 years. In Dutch managed peatlands, subsidence is ongoing at up to one centimetre per year. Under a warmer climate, peat decomposition would be even faster, particularly in drained peatlands. The need for increasingly deeper drainage enhances the upwelling of sulphate-rich brackish or salt water (33). 

Climate induced changes in precipitation will probably be an important factor altering peatland vegetation in temperate and boreal regions, with decreasing wetness during the growing season generally associated with a shift from a Sphagnum dominated to vascular plant dominated vegetation type and a general decline of C sequestration in the long term (4). Mire ecosystems (i.e. bogs, transition bogs and fens) in Central Europe face severe climate-induced risk, with increased summer temperatures being particularly important (5). Although peatlands cover only about 3% of the land surface, they hold the equivalent of half of the atmosphere’s carbon (as CO2), or one third of the world’s soil carbon stock (400-600 Pg) (6).

Global assessment changes forest carbon stock

The results of a global assessment show that climate change stimulates forest productivity on a global level (52). This is due to the fertilization effect of CO2. Global warming itself does not necessarily increase global forest carbon; it has a negative impact on global forest carbon, likely caused by more wildfires, and climatic effects like droughts. This study was based on two scenarios of climate change: a high-end scenario (analogous to the IPCC RCP8.5 scenario), and a scenario where climate change is mitigated such that a 2 °C global mean warming from pre-industrial by 2100 is not exceeded. The latter agrees with a low scenario of climate change (in between the IPCC RCP2.6 and RCP4.5 scenarios). Changes were studied from 1980-2009 to 2070-2099.

Change of carbon dynamics

Climate change will impact the major forest regions of the world in several ways. Forests productivity will change due to higher temperatures and changes in rainfall; in some parts there will be a loss of productivity, in others an increase. Forests will benefit from the higher CO2 concentration in the atmosphere that has a fertilization effect. Climate change will drive migration of forests: they will expand in some regions, and contract in others. Wildfires will increase in most of these major forest regions. Competition will change between different types of vegetation. As a result, carbon dynamics will change (52).

More wildfires versus higher productivity 

Future potential changes were assessed with models that simulate potential future forest growth and decline under different scenarios of climate change and CO2 emissions. The assessment focused on the combined effects of wildfire, climate-induced vegetation migration and productivity in relation to climate change scenarios on a global scale. Main outcomes are: (1) At the global scale mitigating climate change can be beneficial in terms of reducing the impact of wildfires, and costly since it reduces primary production and thus forest carbon; (2) The interplay between direct climate change impacts (changes in temperature and precipitation) and the fertilization effect of CO2 on the world’s forests is complex (52). 

Poleward migration of forests

Under both scenarios of climate change a poleward migration of forests was simulated: in the leading-edge of the migration, grassland and woodlands convert to forests while at the trailing edge, forests convert to shrubland, grassland, or woodland due to lower productivity or frequent fires. Large expanses of boreal forests in Canada and Russia shifted northward, especially under the high-end scenario of climate change. In the southern hemisphere, forest expanded southward in Southern Africa. Poleward migration of forests was not distinct in Western South America, where there the forest contracted along elevation gradients. In Australia, the tropical forests in the north contracted northward as they lost productivity and became woodlands; simultaneously, increased growth of trees was simulated in the woodlands in western Australia, converting those areas to forests (52).

Increase total live forest carbon stock

In these simulations, total live forest carbon stock increased dramatically and consistently under both climate change scenarios, gaining 59% and 54% under the high- and low-end scenario, respectively. The vast majority of the total live forest carbon gain was simulated to occur in the southern hemisphere: Western South America, South America, and South Asia. For Europe only small increases were projected, while Russia was projected to see a significant decline, mainly due to forest contraction and more wildfires. Both positive and negative effects are generally higher for the high-end scenario compared with the low-end scenario of climate change: a higher increase of both carbon stocks and productivity, and of the impact of wildfires (52).

Limitations of this study

This study focused on evaluating the role of wildfire as a major disturbance regime. There is an array of disturbance regimes, however, including land cover change, logging, and insect and pathogen outbreaks. These disturbances were not included in the simulations. Also, current developed and agricultural areas, land use change, and forest management practices were left out. Besides, the study’s results are based on a single climate model; the effects of mitigation policies on the forest carbon stock may be sensitive to climate model selection (52). 

Global assessment changes peatlands carbon stock

The world’s peatlands have been a global carbon sink for millennia. The amounts of carbon they contain equal the amount of carbon in the preindustrial atmosphere. Carbon storage in peatlands is the result of the balance between carbon uptake by plants and microbial decomposition. A disturbance of this balance changes the carbon sink potential of peatlands, thus affecting the concentration of atmospheric green houses gases and, in the end, climate change.

In a warmer climate, decomposition in peatlands increases and more carbon is lost to the atmosphere. On the other hand, the accumulation of peat may increase when primary productivity of plants increases due to higher temperatures and a longer growing season length. Studies on high-latitude northern peatlands data have shown that in warmer climates, increases in plant productivity overcome increases in respiration and that these peatlands will probably become a more efficient sink if soil moisture is maintained (55). This is good news for climate change. However, peatlands at lower latitudes may tell a different story.

In a recent study, the data for northern high latitudes was improved and expanded to low latitudes and southern high latitudes. This way, a more globally distributed peatlands dataset was obtained. From these data, global carbon accumulation over the last millennium (period 850 - 1850) could be quantified. From these insights, future projections of the world’s peatlands carbon sink potential were made (54).

These future projections indicate that until 2100 carbon sink potential of the world’s peatlands slightly increases, both under a low-end and a high-end scenario of climate change (RCP2.6 and RCP8.5 scenarios). This is good news: this increase will lead to a small negative feedback to climate change. The bad news is, this negative feedback does not persist in time. After 2100, carbon sink potential decreases and the initially negative feedback shifts to a positive feedback to climate change. According to the authors of this study, this shift is a plausible and robust result of their study (54).

The shift results from the combined changes in the peatlands at different latitudes. Over time, the balance between an increasing high-latitude sink and a decreasing low-latitude sink changes such that the global sink eventually begins to decrease. At high latitudes the carbon sink increases continuously far beyond 2100. At lower latitudes, higher temperatures drive increased microbial activity and decomposition rates in the peat and surface litter, but this is not fully compensated by increases in plant productivity. As a result, carbon sequestration decreases. At mid-latitudes, peatlands gradually move into the decline phase when decomposition rises faster than net primary productivity (54).

Uncertainties in this study are partly due to assumptions that have been made. One of these assumptions is that no changes in deeper peat layers were taken into account. Only the changes in the surface accumulation rates were considered, whilst deeper peat may also warm and provide a further source of carbon release (54). According to the authors, the most important uncertainty is human impact. Human impact is likely to be the most important determinant of global peatland carbon balance over the next century. Ongoing destruction of tropical peatlands is the largest contributor at present and at current rates, the losses from this source outweigh carbon sequestration rates in natural peatlands (56). Reducing anthropogenic release of peatland carbon is the highest priority in mitigation of peatland impacts on climate change, the authors state (54).

Impacts of forests on climate change mitigation and adaptation

Forests provide multiple water and climate-related services, including precipitation recycling, cooling, water purification, infiltration and groundwater recharge. These services may be far more important, and are often underrated, when compared with traditional benefits such as food, fuel and fiber, and carbon storage. In addition, these services benefit and impact people well beyond the local or catchment scale, often far from where actual decisions on tree planting or removal are made (34).

Forests are intimately linked to rainfall and water availability

Forests contribute to atmospheric moisture and rainfall patterns over land through evapotranspiration: evaporation from soil and plant surfaces and transpiration of water by plants. On average, at least 40% of rainfall over land originates from evapotranspiration. The resulting atmospheric moisture is circulated by winds across the Earth’s continents and oceans. This cross-continental production and transport of atmospheric moisture, called “precipitation recycling”, can promote and intensify the redistribution of water across terrestrial surfaces (34).

Forest loss and degradation reduce evapotranspiration, with important implications for rainfall thousands of kilometres downwind (35). In addition, forests affect the Earth’s surface albedo, temperature, and surface roughness, and thus also alter moisture and heat fluxes between terrestrial surfaces and the atmosphere. Large-scale deforestation may reduce rainfall in some regions by as much as 30% (36). Trees and forests also lead to more intense rainfall through the biological particles they release into the atmosphere (37). Satellite observations suggest European forests are a major influence on cloud formation (38).

The impact of deforestation on altered rainfall patterns can lead to feedback effects on remaining vegetation, reduced biomass accumulation, drought, die-off and fires (39).

Forests transport water locally and globally

Large, continuous areas of forest drive the atmospheric circulation that brings rainfall to continental interiors, according to the so-called “biotic pump theory” (40). It explains that, through transpiration and condensation, forests actively create low pressure regions that draw in moist air from the oceans, thereby generating prevailing winds capable of carrying moisture and sustaining rainfall far within continents (41). Reforestation may re-activate such pumps, returning rainfall to continental interiors (42).

Forests cool locally and globally

Forests influence local and global temperatures and the flow of heat. Individual trees can transpire hundreds of litres of water per day. Every 100-litre of water transpired equals a cooling power equivalent to two average household central air-conditioning units per day (43). Additional regional and global cooling derives from the fact that forests can increase low-level cloud cover and raise reflectivity (44).

On the other hand, forests may contribute to warming. They may stimulate the formation of clouds that trap long wave radiation beneath. Under more cloud-free skies (at high latitudes and particularly in winter) they reduce the earth’s albedo and thus contribute to local warming (45).

Tropical and, to a lesser extent, temperate forests very likely provide net regional/global climate cooling. At higher latitudes, forests may warm regional and global climate (46).

Higher CO2-levels complicate the cooling potential of forests

Vegetation has an impact on extreme high temperatures through a combination of three processes. First, at a higher COconcentration in the atmosphere the stomata in plant leaves are closed more than at lower COconcentrations. As a result, less water is being transpired by the plant, which increases the plant’s water use efficiency but also reduces the latent heat flux, and thus cooling potential, through transpiration. Second, at a higher COconcentration a plant’s leaf area increases (the “fertilization effect” of CO2), which increases the latent heat flux and thus cooling potential, and (partly) offsets the stomatal reduction of transpiration (58). Third, vegetation influences soil moisture: because of the water taken out of the soil and because more biomass leads to more wind turbulence that cools the surface and near-surface air. If the soil dries out, the strong reduction in evaporation limits the cooling potential of the soil and increases extreme air temperature. In addition to these 3 processes, evapotranspiration can also increase as a result of enhanced evaporative demand and the lengthening of the growing season (59).

As a result of this combination of processes, vegetation modulates the latent heat flux and influences surface temperature. Model simulations for a high-end scenario of climate change (RCP 8.5) show that vegetation contributes to the projected global increase of extreme high temperatures with rising CO2-concentrations. In fact, on a global scale, vegetation may account for around 13% of the projected increase of maximum temperatures by 2100 (57).

Forests regulate water supplies

Forests regulate water supplies in many ways. High altitude forests can intercept fog and cloud droplets, which may account for up to 75% of total catchment runoff (47). Where such forests have been removed, the atmospheric moisture present in clouds may move on to other locations. This could represent an important loss to local, downstream water supply (48). Forest clearing may have several, sometimes opposing, effects on water supply, however. Less trees means less water is being evaporated and more groundwater feeds as stream flow into water supplies downstream (49). Loss of tree cover promotes soil degradation that leads to reduced soil infiltration and water retention capacity (50), and in turn reduces groundwater reserves that maintain dry season base flows.

For all the reasons noted above, transpiration, interception, evaporation, infiltration and groundwater recharge, tree cover can either store or recycle substantial amounts of water downwind, providing a positive impact on (and protection of) the local catchment, thereby moderating floods (34).

Mixed species forests are more effective in regulating water supplies and moderating floods than monocultures. Through variation in rooting depth, strength and pattern, different species may aid each other through water uptake, water infiltration and erosion control (51). 

Adaptation strategies - Forests

Climate change is expected to have a diverse range of effects on forests, such as changes in distribution of tree species, effects on forest productivity, increased risk of storms, fires, insect pests, and drought (10). Evidence of such impacts has already been reported for scots pine (11) and beech forests (12).

Climate change-induced productivity and species suitability changes 
in Europe have been assessed (13). This was done for several tree species (oaks, beech, spruce, scots pine) under the A1B scenario of climate change (for three different combinations of regional climate models and general circulation models). With respect to these changes, the value of alternative forest management strategies to compensate for negative effects has been assessed. These alternative strategies are: (a) reduce the rotation length of harvesting and (b) adapt species composition to those species that are expected to perform better under a changed climate (30 years from now).

Reductions in rotation length decrease the time the timber crop is at risk (14), limit the top height reached, reducing windthrow risk (15), and generally reduce uncertainty, allowing another more suitable species to be replanted. A change in species composition avoids risks associated with specific species, as for example, windthrow and bark beetles in Norway spruce (16) or drought-intolerant species, but also to spread risk in general by using more species, the so-called insurance hypothesis (17).

Results show that the European forest system is very inert and that it takes a long time to influence the species distribution by replacing species after final felling. By 2070, on average about 36 % of the area expected to have decreased species suitability will have changed species following business as usual management. Alternative management, consisting of shorter rotations for those species and species planting based on expected trends of climate change (looking 30 years ahead), will have increased this species transition to 40 % (13).

Northern Europe is projected to show the highest production increases under climate change and can also adapt its species distribution faster. The forest in Northern Europe is generally distributed rather equally over the age classes, and lowering the rotation length makes a considerable additional amount of area available for harvesting, resulting in fast adaptation. Besides, in Northern Europe only a few species show a lower suitability due to climate change. Southwest Europe is expected to face the greatest challenge by a combination of a predicted loss of production and a slow rate of management alteration under climate change. In Southwest Europe, the age class distribution is less favourable for adaptation with a large share of relatively young stands, and the suitability due to climate changes affects almost all (nine out of 10) species (13).

Although alternative management can increase the rate at which species with declining suitability are replaced, it comes at the cost of a lower increment. Generally, the species with declining suitability are productive coniferous species like spruce, whereas the species to plant are mostly broadleaves with lower increment rates. Moreover, increased final felling leads to a temporary reduction in increment due to the increased replacement of mature forest by low-productive young forest. At longer timescales, the increased regeneration effort might lead to higher increment rates under adaptive management (13),

The results show that forest management cannot keep pace with the projected change in species suitability. As a consequence, larger parts of the forest will be exposed to climate regimes they are not adapted to, thereby increasing the risk of large-scale mortality events due to for example drought and pests (13). 

Adaptation strategies - Peatlands

Drained peat soils used for agriculture can be large greenhouse gas sources. By rewetting these soils and thus restoring the wetlands, significant agricultural emissions are avoided, and soil carbon can be sequestered and protected. Rewetting of peat soils is not a quick win contribution to the reduction of greenhouse gases emissions, however (60).

With respect to rewetting peat soils, the net impact on the emission of greenhouse gases depends on the balance between CO2 and CH4 (the impact of N2O, another greenhouse gas emitted by these soils, has been shown to be negligible, compared to the other two (61)). CO2 is emitted when peat soils are drained and exposed to the air. CH4 is formed under anaerobic conditions, so when peat soils are rewetted again. Clearly, wetland restoration comes with a biogeochemical compromise (62): while flooded wetlands have the potential to sequester carbon, the highly reduced (anaerobic) conditions can result in significant CH4 emissions (63). What’s more, CH4 is a much stronger greenhouse gas than CO2! Wetlands contribute around 30% of all anthropogenic and natural CH4 sources (65).

35 years of continuous CO2 and CH4 measurements in the Sacramento-San Joaquin River Delta (USA) showed that restored wetlands effectively sequestered carbon and halted soil carbon loss associated with drained agricultural land uses. It takes some time, however, before this change from carbon source (drained land) to carbon sink (restored wetland) becomes effective in terms of climate change mitigation: restored wetlands do not begin to accrue greenhouse gas benefits until nearly a half century, and become net sinks from the atmosphere after a century. This is related to the fact that after rewetting, the peat soils start to emit CH4 as a result of anaerobic decomposition of carbon. CH4 being a much stronger greenhouse gas, it takes some time for the CO2 accumulation to dominate over this CH4 effect. The latter effect may turn restored wetlands into net sources of greenhouse gases to the atmosphere over decadal timescales (64). The good news in this US study: over time, the cumulative removal of CO2, an extremely long-lived green house gas, from the atmosphere vastly outweighs the short-lived CH4 warming effect (60).

The study shows that land use change from agriculture on drained, degraded peat soils to freshwater, deltaic restored wetlands will result in a net greenhouse gases benefit over multi-decadal timescales, while accreting soil and sequestering carbon from the atmosphere into the ecosystem. Along with climate benefits, these ecosystem services have the potential to halt and reverse soil subsidence and protect the fragile hydrological network of wetland ecosystems. The message to policymakers and planners: take measures that promote the long-term restoration of these kinds of systems to maximize climatic benefit (60).


The references below are cited in full in a separate map 'References'. Please click here if you are looking for the full references for Europe.

  1. Charru et al. (2010); Silva et al. (2010), both in: IPCC (2014)
  2. Ciais et al. (2005); Witte et al. (2011); Kasson and Livingston (2012), all in: IPCC (2014)
  3. Allen et al. (2010a), in: IPCC (2014)
  4. Limpens et al. (2008), in: IPCC (2014)
  5. Essl et al. (2012), in: IPCC (2014)
  6. Limpens et al. (2008); Frolking et al. (2011); Page et al. (2011), all in: IPCC (2014)
  7. IPCC (2014)
  8. Van Mantgem et al., 2009; Phillips et al., 2009; Williams et al., 2013; Reichstein et al. 2013, a.o., all in: McDowell et al. (2015)
  9. McDowell et al. (2015)
  10. Hanewinkel et al. (2013); Reyer et al. (2014); Nikulin et al. (2011); Carvalho et al. (2011); Robinet and Roques (2010); Allen et al.( 2010), all in: Schelhaas et al. (2015)
  11. Rigling et al.( 2013), in: Schelhaas et al. (2015)
  12. Kint et al. (2012); Penuelas et al. (2007); Bolte et al. (2014), all in: Schelhaas et al. (2015)
  13. Schelhaas et al. (2015), in: Schelhaas et al. (2015)
  14. Schelhaas (2008); Schelhaas et al. (2010), both in: Schelhaas et al. (2015)
  15. Schelhaas (2008), in: Schelhaas et al. (2015)
  16. Seidl et al. (2009), in: Schelhaas et al. (2015)
  17. Folke et al. (1996), in: Schelhaas et al. (2015)
  18. Rabasa et al. (2013); Saltré et al. (2014), both in: Bosela et al (2016)
  19. Lindner et al. (2010), in: Hölzel et al. (2016)
  20. Schlyter et al. 2006; Bolte et al. (2010), both in: Hölzel et al. (2016)
  21. Joosten and Clarke (2002), in: Hölzel et al. (2016)
  22. Blodau (2002); Frolking and Roulet (2007); Limpens et al. (2008); Oleszczuk et al. (2008); Finkelstein and Cowling (2011); Couwenberg (2011); Joosten et al. (2012), all in: Hölzel et al. (2016)
  23. Beetz et al. (2013); Leiber-Sauheitl et al. (2013); Schrier-Uijl et al. (2014), all in: Hölzel et al. (2016)
  24. Ise et al. (2008), in: Hölzel et al. (2016)
  25. Hölzel et al. (2016)
  26. Cannell et al. (1999); Van den Bos (2003); Drösler et al. (2008), (2011); Verhagen et al. (2009); Joosten (2010); Worrall et al. (2011); Nielsen et al. (2013), all in: Hölzel et al. (2016)
  27. Walter et al. (2001); Van Huissteden and Van den Bos (2006); Bohn et al. (2007); Meng et al. (2012), all in: Hölzel et al. (2016)
  28. Melton et al. (2013), in: Hölzel et al. (2016)
  29. Laine et al. (2009), in: Hölzel et al. (2016)
  30. Goldberg et al. (2010), in: Hölzel et al. (2016)
  31. Laine et al. (2007), (2009), both in: Hölzel et al. (2016)
  32. Godwin (1978); Borger (1992); Verhoeven (1992); Hoogland et al. (2012), all in: Hölzel et al. (2016)
  33. Hoogland et al. (2012), in: Hölzel et al. (2016)
  34. Ellison et al. (2017)
  35. Aldrich and Imberger (2013); Debortoli et al. (2016), both in: Ellison et al. (2017)
  36. Lawrence and Vandecar (2015); Spracklen and Garcia-Carreras (2015), both in: Ellison et al. (2017)
  37. Morris et al. (2014); Sheil (2014), both in: Ellison et al. (2017)
  38. Teuling et al. (2017), in: Ellison et al. (2017)
  39. Brienen et al. (2015); Duffy et al. (2015), both in: Ellison et al. (2017)
  40. Makarieva and Gorshkov (2007), in: Ellison et al. (2017)
  41. Makarieva et al. (2013a); Makarieva and Gorshkov (2007); Nobre (2014); Sheil and Murdiyarso (2009), all in: Ellison et al. (2017)
  42. Sheil and Murdiyarso (2009), in: Ellison et al. (2017)
  43. Teuling et al. (2010); Zaitchik et al. (2006), both in: Ellison et al. (2017)
  44. Ban-Weiss et al. (2011); Heiblum et al. (2014), both in: Ellison et al. (2017)
  45. Lee et al. (2011); Li et al. (2015), both in: Ellison et al. (2017)
  46. Bala et al. (2007); Chapin et al. (2000); Lee et al. (2011), all in: Ellison et al. (2017)
  47. Bruijnzeel et al. (2011), in: Ellison et al. (2017)
  48. Bruijnzeel (2004), in: Ellison et al. (2017)
  49. Andréassian (2004); Bosch and Hewlett (1982); Farley et al. (2005); Stickler et al. (2013), all in: Ellison et al. (2017)
  50. Lal (1996); Nyberg et al. (2012); Zimmermann and Elsenbeer (2008), all in: Ellison et al. (2017)
  51. Reubens et al. (2007), in: Ellison et al. (2017)
  52. Kim et al. (2017)
  53. Tanneberger et al. (2017)
  54. Gallego-Sala et al. (2018)
  55. Loisel et al. (2014); Charman et al. (2013); Yu (2011), all in: Gallego-Sala et al. (2018)
  56. Page et al. (2002); Moore et al. (2013), both in: Gallego-Sala et al. (2018)
  57. Lemordant and Gentine (2019)
  58. Mccarthy et al. (2007); Wullschleger et al. (2002), both in: Lemordant and Gentine (2019)
  59. Frank et al. (2015), in: Lemordant and Gentine (2019)
  60. Hemes et al. (2019)
  61. McNicol et al. (2016), in: Hemes et al. (2019)
  62. Hemes et al. (2018a); Hoper et al. (2008); Petrescu et al. (2015), all in: Hemes et al. (2019)
  63. Bridgham et al. (2013); Dean et al. (2018), both in: Hemes et al. (2019)
  64. Hemes et al. (2018a), in: Hemes et al. (2019)
  65. Feldman et al. (2018); Nisbet et al. (2016); Poulter et al. (2017), all in: Hemes et al. (2019)
  66. Takolander et al. (2019)