Editing IPCC:AR6/WGII/Chapter-2 (section)

=== 2.4.3 Observed Changes in Key Biomes, Ecosystems and Their Services ===

<div id="h2-9-siblings" class="h2-siblings"></div>

<div id="2.4.3.1" class="h3-container"></div>

<span id="detection-and-attribution-for-observed-biome-shifts"></span>
==== 2.4.3.1 Detection and Attribution for Observed Biome Shifts ====

<div id="h3-15-siblings" class="h3-siblings"></div>

Attribution for biome (major vegetation form of an ecosystem) shifts is complex because of their extensive, sometimes continental, spatial scale ( [[#Whittaker--1975|Whittaker, 1975]] ; [[#Olson--2001|Olson et al., 2001]] ; [[#Woodward--2004|Woodward et al., 2004]] ). Therefore, non-climatic factors strongly influence biome spatial distributions ( [[#Ellis--2008|Ellis and Ramankutty, 2008]] ).

The most robust attribution studies use data from many species, individual locations with minimal confounding factors, particularly observed recent LULCC, and scale up by analysing multiple locations across a large zone between biomes, providing multiple lines of evidence ( [[#Hegerl--2010|Hegerl et al., 2010]] ; [[#Parmesan--2013|Parmesan et al., 2013]] ). Multivariate statistical analyses aid attribution studies by allowing the assessment of relative weights among multiple factors, including variables related to climate change ( [[#Gonzalez--2012|Gonzalez et al., 2012]] ). However, drivers often have strong, significant interactions with one another, complicating quantitative assessment of the strength of individual drivers ( [[#Parmesan--2013|Parmesan et al., 2013]] ). In these cases, manipulative experiments are critical in assessing attribution to the drivers of climate change.

Certain biomes exhibit a relatively stronger relationship to climate; for example, Arctic tundra generally has a distinct ecotone with boreal conifer forest ( [[#Whittaker--1975|Whittaker, 1975]] ). In these areas, attribution of biome shifts to climate change are relatively straightforward, if human LULCC is minimal. However, other biomes, such as many grassland systems, are not in equilibrium with climate ( [[#Bond--2005|Bond et al., 2005]] ). In these systems, their evolutionary history ( [[#Keeley--2011|Keeley et al., 2011]] ; [[#Strömberg--2011|Strömberg, 2011]] ; [[#Charles-Dominique--2016|Charles-Dominique et al., 2016]] ), distribution, structure and function have been shaped by climate and natural disturbances, such as fire and herbivory ( [[#Staver--2011|Staver et al., 2011]] ; [[#Lehmann--2014|Lehmann et al., 2014]] ; [[#Pausas--2015|Pausas, 2015]] ; [[#Bakker--2016|Bakker et al., 2016]] ; [[#Malhi--2016|Malhi et al., 2016]] ). Disturbance variability is an inherent characteristic of grassland systems, and suitable ‘control’ conditions are seldom available in nature. Furthermore, due to the integral role of disturbance, these biomes have been widely affected by long-term and widespread shifts in grazing regimes, large-scale losses of mega-herbivores and fire suppression policies ( [[#Archibald--2013|Archibald et al., 2013]] ; [[#Malhi--2016|Malhi et al., 2016]] ; [[#Hempson--2017|Hempson et al., 2017]] ). It is necessary to conduct climate change attribution on a case-by-case basis for grasslands; such assessments are complex as direct climate change impacts from either inherent variation within disturbance regimes or directional changes in background disturbances are difficult to separate (detailed in Sections 2.4.3.2.1; 2.4.3.2.2; 2.4.3.5). Confidence in assessments is increased when the observed trends are supported by a mechanistic understanding of responses identified by physiological studies, manipulative field experiments, greenhouse studies and lab experiments (Table SM2.1).

<div id="2.4.3.2" class="h3-container"></div>

<span id="global-patterns-of-observed-biome-shifts-driven-by-climate-change"></span>
==== 2.4.3.2 Global Patterns of Observed Biome Shifts Driven by Climate Change ====

<div id="h3-16-siblings" class="h3-siblings"></div>

<div id="2.4.3.2.1" class="h4-container"></div>

<span id="observed-biome-shifts-predominantly-driven-by-climate-change"></span>
===== 2.4.3.2.1 Observed biome shifts predominantly driven by climate change =====

<div id="h4-15-siblings" class="h4-siblings"></div>

AR5 and a meta-analysis found that vegetation at the biome level shifted poleward latitudinally and upward altitudinally due to anthropogenic climate change in at least 19 sites in boreal, temperate and tropical ecosystems from 1700 to 2007 ( [[#Gonzalez--2010|Gonzalez et al., 2010]] ; [[#Settele--2014|Settele et al., 2014]] ). In these areas, temperature increased to 0.4°C–1.6°C above the pre-industrial period ( [[#Gonzalez--2010|Gonzalez et al., 2010]] ; [[#Settele--2014|Settele et al., 2014]] ). Field research since the AR5 detected additional poleward and upslope biome shifts over periods of 24–210 years at numerous sites (described below), but were not directly attributed to anthropogenic climate change as the studies were not designed or conducted properly for full attribution assessment.

Many of the recently detected shifts are nevertheless consistent with climate change-induced temperature increases and observed in areas without agriculture, livestock grazing, timber harvesting and other anthropogenic land uses. For example, in the Andes Mountains in Ecuador, a biome shift was detected by comparing a survey by Alexander von Humboldt in 1802 to a re-survey in 2012, making this the longest time span in the world for this type of data ( [[#Morueta-Holme--2015|Morueta-Holme et al., 2015]] ; [[#Moret--2019|Moret et al., 2019]] ). Over 210 years, temperature increased by 1.7°C ( [[#Morueta-Holme--2015|Morueta-Holme et al., 2015]] ) and the upper edge of alpine grassland shifted 100–450 m upslope ( [[#Moret--2019|Moret et al., 2019]] ).

Other biome shifts consistent with climate change and not substantially affected by local land use include: northward shifts in Canada of deciduous forest into boreal conifer forest, 5 km from 1970–2012 ( [[#Sittaro--2017|Sittaro et al., 2017]] ) and 20 km from 1970–2014 ( [[#Boisvert-Marsh--2019|Boisvert-Marsh et al., 2019]] ) and of temperate conifer into boreal conifer forest, 21 km from 1970–2015 ( [[#Boisvert-Marsh--2021|Boisvert-Marsh and de Blois, 2021]] ). Research detected upslope shifts of boreal and sub-alpine conifer forest into alpine grassland at 143 sites on four continents (41 m from 1901–2018) ( [[#Lu--2021|Lu et al., 2021]] ) and at individual sites in Canada (54 m from 1900–2010) ( [[#Davis--2020|Davis et al., 2020]] ); China (300 m from 1910–2000) ( [[#Liang--2016|Liang et al., 2016]] ) (33 m from 1985–2014) ( [[#Du--2018|Du et al., 2018]] ); Nepal (50 m from 1860–2000) ( [[#Sigdel--2018|Sigdel et al., 2018]] ); Russia (150 m from 1954–2006) ( [[#Gatti--2019|Gatti et al., 2019]] ); and the USA (19 m from 1950–2016) ( [[#Smithers--2018|Smithers et al., 2018]] ) (38 m from 1953–2015) ( [[#Terskaia--2020|Terskaia et al., 2020]] ). Other upslope cases include shifts of temperate conifer forest in Canada ( [[#Jackson--2016|Jackson et al., 2016]] ) and the USA ( [[#Lubetkin--2017|Lubetkin et al., 2017]] ), temperate deciduous forest in Switzerland ( [[#Rigling--2013|Rigling et al., 2013]] ) and temperate shrubland in the USA ( [[#Donato--2016|Donato et al., 2016]] ).

In summary, anthropogenic climate change caused latitudinal and elevational biome shifts in at least 19 sites in boreal, temperate and tropical ecosystems between 1700 and 2007, where temperature increased to 0.4°C–1.6°C above the pre-industrial period ( ''robust evidence'' , ''high agreement'' ). Additional cases of 5–20 km northward and 20–300 m upslope biome shifts between 1860 and 2016, under a mean global temperature increase of approximately 0.9°C above the pre-industrial period, are consistent with climate change ( ''medium evidence'' , ''high agreement'' ).

<div id="2.4.3.2.2" class="h4-container"></div>

<span id="observed-biome-shifts-from-combined-land-use-change-and-climate-change"></span>
===== 2.4.3.2.2 Observed biome shifts from combined land use change and climate change =====

<div id="h4-16-siblings" class="h4-siblings"></div>

Research has detected biome shifts in areas where agriculture, fire use or suppression, livestock grazing, harvesting of timber and wood for fuel and other local land use substantially altered vegetation, in addition to changes in climatic factors and CO 2 fertilisation. These studies were not designed or conducted in a manner to make climate change attribution possible, although many vegetation changes are consistent with climate change. For example, a global review of observed changes in tree lines found that, globally, two-thirds of tree lines have shifted upslope in elevation over the past 50 years or more, (( [[#Hansson--2021|Hansson et al., 2021]] ).

Upslope and poleward forest shifts have occurred where timber harvesting or livestock grazing has been abandoned, allowing the regeneration of trees at sites in Canada ( [[#Brice--2019|Brice et al., 2019]] ; [[#Wang--2020b|Wang et al., 2020b]] ), France ( [[#Feuillet--2020|Feuillet et al., 2020]] ), Italy ( [[#Vitali--2017|Vitali et al., 2017]] ), Spain ( [[#Ameztegui--2016|Ameztegui et al., 2016]] ) and the USA ( [[#Wang--2020b|Wang et al., 2020b]] ) as well as in mountainous areas across Europe ( [[#Cudlin--2017|Cudlin et al., 2017]] ). Intentional use of fire drove an upslope forest shift in Peru ( [[#Bush--2015|Bush et al., 2015]] ) while mainly human-ignited fires drove the conversion of shrubland to grassland in a drought-affected area of the USA ( [[#Syphard--2019b|Syphard et al., 2019b]] ). In eastern Canada, timber harvesting and wildfire drove the conversion of mixed conifer–broadleaf forests to broadleaf-dominated forests ( [[#Brice--2020|Brice et al., 2020]] ; [[#Wang--2020b|Wang et al., 2020b]] ).

Shrub encroachment onto savanna has occurred at numerous sites, particularly across the Southern Hemisphere, mainly between 1992 and 2010 ( [[#Criado--2020|Criado et al., 2020]] ). Globally, overgrazing initiates shrub encroachment by reducing grasses more than woody plants, while fire exclusion maintains the shrub cover ( [[#D’Odorico--2012|D’Odorico et al., 2012]] ; [[#Caracciolo--2016|Caracciolo et al., 2016]] ; [[#Bestelmeyer--2018|Bestelmeyer et al., 2018]] ). The magnitude of woody cover change in savannas is not correlated with mean annual temperature change ( [[#Criado--2020|Criado et al., 2020]] ); however, higher atmospheric CO 2 increases shrub growth in savannas ( [[#Nackley--2018|Nackley et al., 2018]] ; [[#Manea--2019|Manea and Leishman, 2019]] ). A global remote-sensing analysis of biome changes from all causes, including agricultural and grazing expansion and deforestation, estimated that 14% of pixels changed between 1981 and 2012, although this approach can overestimate global changes, since it uses a new biome classification system which doubles the conventional biome classifications ( [[#Higgins--2016|Higgins et al., 2016]] ). In addition to climate change, LULCC causes vegetation changes at the biome level ( ''robust evidence'' , ''high agreement'' ).

<div id="2.4.3.3" class="h3-container"></div>

<span id="observed-changes-in-deserts-and-arid-shrublands"></span>
==== 2.4.3.3 Observed Changes in Deserts and Arid Shrublands ====

<div id="h3-17-siblings" class="h3-siblings"></div>

Divergent responses to anthropogenic climate change are occurring within and across arid regions, depending on time period, location, detection methodology and vegetation type (see Cross-Chapter Paper 3). Emerging shifts in ecosystem structure, functioning and biodiversity are supported by evidence from modelled impacts of projected climate and CO 2 levels. While observed responsiveness of arid vegetation productivity to rising atmospheric CO 2 ( [[#Fensholt--2012|Fensholt et al., 2012]] ) may offset risks from reduced water availability ( [[#Fang--2017|Fang et al., 2017]] ), climate- and CO 2 -driven changes are key risks in arid regions, interacting with habitat degradation, wildfires and invasive species ( [[#Hurlbert--2019|Hurlbert et al., 2019]] ).

Widespread vegetation greening, as projected in AR4, is occurring in arid shrublands ( [[#Zhang--2019a|Zhang et al., 2019a]] ; [[#Maestre--2021|Maestre et al., 2021]] ) as a result of increases in leaf area, woody cover and herbaceous production at desert–grassland interfaces ( [[#Gonsamo--2021|Gonsamo et al., 2021]] ). Plant productivity in arid regions has increased ( [[#Fensholt--2012|Fensholt et al., 2012]] ) because of improved water-use efficiency associated with elevated CO 2 ( [[#Norby--2011|Norby and Zak, 2011]] ; [[#Donohue--2013|Donohue et al., 2013]] ; [[#Burrell--2020|Burrell et al., 2020]] ; [[#Gonsamo--2021|Gonsamo et al., 2021]] ) ( ''medium evidence'' , ''high agreement'' ), altered rainfall seasonality and amount ( [[#Rohde--2019|Rohde et al., 2019]] ; [[#Zhang--2019a|Zhang et al., 2019a]] ) ( ''robust evidence'' , ''high agreement'' ), increases in temperature ( [[#Ratajczak--2014|Ratajczak et al., 2014]] ; [[#Wilcox--2018|Wilcox et al., 2018]] ) ( ''robust evidence'' , ''high agreement'' ) and heavy grazing ( ''robust evidence'' , ''high agreement'' ), with the relative importance differing across locations ( [[#Donohue--2013|Donohue et al., 2013]] ; [[#Caracciolo--2016|Caracciolo et al., 2016]] ; [[#Archer--2017|Archer et al., 2017]] ; [[#Hoffmann--2019b|Hoffmann et al., 2019b]] ; [[#Rohde--2019|Rohde et al., 2019]] ). Woody-plant encroachment into arid shrublands is occurring with ''high confidence'' in North America ( [[#Caracciolo--2016|Caracciolo et al., 2016]] ; [[#Archer--2017|Archer et al., 2017]] ) and southern Africa ( [[#du%20Toit--2014|du Toit and O’Connor, 2014]] ; [[#Ward--2014|Ward et al., 2014]] ; [[#Masubelele--2015a|Masubelele et al., 2015a]] ; [[#Hoffman--2019|Hoffman et al., 2019]] ; [[#Rohde--2019|Rohde et al., 2019]] ), and with ''low confidence'' in central Asia ( [[#Li--2015|Li et al., 2015]] ). In North America, sagebrush steppe changes have been attributed to increases in temperature and earlier snowpack melt ( [[#USGCRP--2017|USGCRP, 2017]] ; [[#Mote--2018|Mote et al., 2018]] ; [[#Snyder--2019|Snyder et al., 2019]] ).

Non-native grasses are invading the sagebrush steppes (cold deserts) in North America ( [[#Chambers--2014|Chambers et al., 2014]] ) attributed to warming ( [[#Bradley--2016|Bradley et al., 2016]] ; [[#Hufft--2016|Hufft and Zelikova, 2016]] ). In the eastern semi-desert (Karoo) of South Africa, annual rainfall increases and a rainfall seasonality shift ( [[#du%20Toit--2014|du Toit and O’Connor, 2014]] ) are increasing grassiness as arid grasslands expand into semi-desert shrublands ( [[#du%20Toit--2015|du Toit et al., 2015]] ; [[#Masubelele--2015b|Masubelele et al., 2015b]] ; [[#Masubelele--2015a|Masubelele et al., 2015a]] ) causing fire in areas seldom burned historically ( [[#Coates--2016|Coates et al., 2016]] ).

Interactions of drought, warming and land management have caused vegetation mortality (see [[#2.4.4.3|Section 2.4.4.3]] ) and reduced vegetation cover in shrublands, as projected by AR4 ( [[#Burrell--2020|Burrell et al., 2020]] ). Increased heat and drought are causing the health and abundance of succulent species to decline ( [[#Musil--2009|Musil et al., 2009]] ; [[#Schmiedel--2012|Schmiedel et al., 2012]] ; [[#Aragón-Gastélum--2014|Aragón-Gastélum et al., 2014]] ; [[#Koźmińska--2019|Koźmińska et al., 2019]] ). Hot droughts, in particular, have been shown to reduce population resilience ( [[#Koźmińska--2019|Koźmińska et al., 2019]] ).

<div id="2.4.3.4" class="h3-container"></div>

<span id="observed-changes-in-mediterranean-type-ecosystems"></span>
==== 2.4.3.4 Observed Changes in Mediterranean-Type Ecosystems ====

<div id="h3-18-siblings" class="h3-siblings"></div>

Since AR5 ( [[#Settele--2014|Settele et al. (2014)]] , all five Mediterranean-type ecosystems (MTEs) of the world have experienced extreme droughts within the past decade, with South Africa and California reporting their worst on record ( ''robust evidence'' , ''high agreement'' ) ( [[#Diffenbaugh--2015|Diffenbaugh et al., 2015]] ; [[#Williams--2015a|Williams et al., 2015a]] ; [[#Garreaud--2017|Garreaud et al., 2017]] ; [[#Otto--2018|Otto et al., 2018]] ; [[#Sousa--2018|Sousa et al., 2018]] ). Climate change is causing these droughts to become more frequent and severe ( ''medium evidence'' , ''medium agreement'' ) ( [[#AghaKouchak--2014|AghaKouchak et al., 2014]] ; [[#Garreaud--2017|Garreaud et al., 2017]] ; [[#Otto--2018|Otto et al., 2018]] ; [[#Seneviratne--2021|Seneviratne et al., 2021]] ).

MTEs show a range of direct responses to various forms of water deficit, but have also been affected by increasing fire activity linked to drought ( [[#Abatzoglou--2016|Abatzoglou and Williams, 2016]] ), and interactions between drought or extreme weather and fire affecting post-fire ecosystem recovery ( [[#Slingsby--2017|Slingsby et al., 2017]] ). Responses include shifts in functional composition ( [[#Acácio--2017|Acácio et al., 2017]] ; [[#Syphard--2019a|Syphard et al., 2019a]] ), decline of vegetation health ( [[#Hope--2014|Hope et al., 2014]] ; [[#Asner--2016a|Asner et al., 2016a]] ), decline or loss of characteristic species ( [[#White--2016|White et al., 2016]] ; [[#Stephenson--2019|Stephenson et al., 2019]] ), shifts in composition towards more drought- or heat-adapted species and declining diversity (see also section 2.4.4.3) ( [[#Slingsby--2017|Slingsby et al., 2017]] .; [[#Harrison--2018|Harrison et al., 2018]] ).

Declines in plant health and increased mortality in MTEs associated with drought have been widely documented ( ''robust evidence'' , ''high agreement'' ) ( [[#2.4.4.3|Section 2.4.4.3]] ). Remote-sensing studies show drought-associated mortality in post-fire vegetation regrowth in the Fynbos of South Africa ( [[#Slingsby--2020b|Slingsby et al., 2020b]] ), reduced canopy health in forests within MTE zones of South Africa ( [[#Hope--2014|Hope et al., 2014]] ) and declines in canopy water content in the forests of California ( [[#Asner--2016a|Asner et al., 2016a]] ). Several studies reported climate-associated responses of dominant or charismatic species. High mortality in the Clanwilliam cedar tree between 1931 and 2013 occurred at lower, hotter elevations in the Fynbos of South Africa ( [[#White--2016|White et al., 2016]] ). Drought reduced growth and increased mortality of the holm oak, ''Quercus ilex'' , on the Iberian Peninsula of Spain ( [[#Natalini--2016|Natalini et al. (2016)]] . Portuguese shrublands experienced losses of many deciduous and evergreen oak species, and an increasing dominance of pyrophytic xeric trees ( [[#Acácio--2017|Acácio et al., 2017]] ). The 2012–2015 drought in California caused high-canopy foliage dieback of the giant sequoia ( ''Sequoiadendron giganteum'' ) ( [[#Stephenson--2019|Stephenson et al., 2019]] ), increased the dominance of oaks relative to pines as a result of the increased water deficit, and led to large-scale tree mortality due to interactions of drought and insect pest outbreaks ( [[#McIntyre--2015|McIntyre et al., 2015]] ; [[#Fettig--2019|Fettig et al., 2019]] ).

Species distribution or community composition changes have contributed to declines in diversity and/or shifts towards more drought- or heat-adapted species ( ''medium evidence'' , ''high agreement'' ). Two conifer species ( ''Pinus longaeva'' and ''P. flexilis'' ) shifted upslope 19 m from 1950 to 2016 in the Great Basin, USA, ( [[#Smithers--2018|Smithers et al., 2018]] ). Reduced winter precipitation caused native annual forbs to recede, resulting in long-lasting and potentially unidirectional reductions in diversity in a Californian grassland ( [[#Harrison--2018|Harrison et al., 2018]] ). More frequent extreme hot and dry weather between 1966 and 2010 caused a decline in diversity during the post-fire regeneration phase in the Fynbos of South Africa ( [[#Slingsby--2017|Slingsby et al., 2017]] ), resulting in shifts towards species with higher temperature preferences ( [[#Slingsby--2017|Slingsby et al., 2017]] ). In Italy, [[#Del%20Vecchio--2015|Del Vecchio et al. (2015)]] observed increases in plant cover and thermophilic species in coastal foredune habitats between 1989 and 2012.

In southern California, USA, areas of forest and woody shrublands are shifting to grasslands, driven by a combination of climate and land use factors such as increased drought, fire ignition frequency and increases in nitrogen deposition ( ''robust evidence'' , ''high agreement'' ) ( [[#Jacobsen--2018|Jacobsen and Pratt, 2018]] ; [[#Park--2018|Park et al., 2018]] ; [[#Park--2019|Park and Jenerette, 2019]] ; [[#Syphard--2019b|Syphard et al., 2019b]] ).

The effects of climate change on heat, fuel and wildfire ignition limits show spatial and temporal variation globally (see [[#2.3|Section 2.3.6.1]] ), but there have been a number of observed impacts on MTEs ( ''medium evidence'' , ''high agreement'' ). Climate change caused increases in fuel aridity and the area of land burned by wildfires across the western USA from 1985 to 2015 ( [[#Abatzoglou--2016|Abatzoglou and Williams, 2016]] ). Local and global climatic variability led to a 4-year decrease in the average fire return time in the Fynbos, South Africa, when comparing fires recorded in 1951–1975 and 1976–2000 ( [[#Wilson--2010|Wilson et al., 2010]] ). In Chile, [[#González--2018|González et al. (2018)]] reported a significant increase in the number, size, duration and simultaneity of large fires during the 2010–2015 ‘megadrought’ when compared to the 1990–2009 baseline.

<div id="2.4.3.5" class="h3-container"></div>

<span id="observed-changes-in-savanna-and-grasslands"></span>
==== 2.4.3.5 Observed Changes in Savanna and Grasslands ====

<div id="h3-19-siblings" class="h3-siblings"></div>

Savannas consist of co-existing trees and grasses in tropical and temperate regions ( [[#Archibald--2019|Archibald et al., 2019]] ). The global trend of woody encroachment reported in AR5 ( [[#Settele--2014|Settele et al., 2014]] ) is continuing ( ''robust evidence'' , ''high agreement'' , ''very high confidence'' ) (see Table SM2.1), with increases occurring in temperate savannas in North America (10–20% per decade) and tropical savannas in South America (8% per decade), Africa (2.4% per decade) and Australia (1% per decade) ( [[#O’Connor--2014|O’Connor et al., 2014]] ; [[#Espírito-Santo--2016|Espírito-Santo et al., 2016]] ; [[#Skowno--2017|Skowno et al., 2017]] ; [[#Stevens--2017|Stevens et al., 2017]] ; McNicol et al., 2018; [[#Venter--2018|Venter et al., 2018]] ; [[#Rosan--2019|Rosan et al., 2019]] ). Additionally, the forest expansion into mesic savannas reported in AR5 ( [[#Settele--2014|Settele et al., 2014]] ) is continuing in Africa, South America and Southeast Asia ( [[#Marimon--2014|Marimon et al., 2014]] ; [[#Keenan--2015|Keenan et al., 2015]] ; [[#Baccini--2017|Baccini et al., 2017]] ; [[#Ondei--2017|Ondei et al., 2017]] ; [[#Stevens--2017|Stevens et al., 2017]] ; Aleman et al., 2018; [[#Rosan--2019|Rosan et al., 2019]] ). Extreme high rainfall anomalies have also contributed to an increase in herbaceous and foliar production in the Sahel ( [[#Brandt--2019|Brandt et al., 2019]] ; [[#Zhang--2019a|Zhang et al., 2019a]] ).

New studies since AR5, using multiple study designs (experimental manipulations in lab and field, meta-analyses and modelling), attribute climate change increases in woody cover to elevated atmospheric CO 2 ( [[#Donohue--2013|Donohue et al., 2013]] ; [[#Nackley--2018|Nackley et al., 2018]] ; [[#Quirk--2019|Quirk et al., 2019]] ) and increased rainfall amount and intensity ( ''robust evidence'' , ''high agreement'' ) ( [[#Venter--2018|Venter et al., 2018]] ; [[#Xu--2018b|Xu et al., 2018b]] ; [[#Zhang--2019a|Zhang et al., 2019a]] ). Direct quantification of climate-change drivers is confounded with local LUC such as fire suppression ( [[#Archibald--2016|Archibald, 2016]] ; [[#Venter--2018|Venter et al., 2018]] ) '','' heavy grazing ( [[#du%20Toit--2014|du Toit and O’Connor, 2014]] ; [[#Archer--2017|Archer et al., 2017]] ), removal of native browsers and, specifically, loss of mega-herbivores in Africa ( ''medium evidence'' , ''medium agreement'' ) ( [[#Asner--2016b|Asner et al., 2016b]] ; [[#Daskin--2016|Daskin et al., 2016]] ; [[#Stevens--2016|Stevens et al., 2016]] ; [[#Davies--2018|Davies et al., 2018]] ). The relative importance of the climate- and non-climate-related causes of woody plants varies between regions, but there is general consensus that the impacts of climate change, specifically, increasing rainfall and rising CO 2 , are frequent and strong contributing factors of woody-cover increase ( ''robust evidence'' , ''high agreement'' ).

Extensive woody-cover increases in non-forested biomes is reducing grazing potential ( [[#Smit--2015|Smit and Prins, 2015]] ) as well as changing the carbon stored per unit of land area ( [[#González-Roglich--2014|González-Roglich et al., 2014]] ; [[#Puttock--2014|Puttock et al., 2014]] ; [[#Pellegrini--2016|Pellegrini et al., 2016]] ; [[#Mureva--2018|Mureva et al., 2018]] ) and the hydrological characteristics ( [[#Honda--2016|Honda and Durigan, 2016]] ; [[#Schreiner-McGraw--2020|Schreiner-McGraw et al., 2020]] ). Woody-cover encroachment also reduces biodiversity by threatening fauna and flora adapted to open ecosystems ( [[#Ratajczak--2012|Ratajczak et al., 2012]] ; [[#Smit--2015|Smit and Prins, 2015]] ; [[#Pellegrini--2016|Pellegrini et al., 2016]] ; [[#Andersen--2019|Andersen and Steidl, 2019]] ).

The global extent of grasslands is declining significantly because of climate change ( ''medium confidence'' ). In temperate and boreal zones, where about half of tree lines are shifting, they are overwhelmingly expanding poleward and upward, with an accompanying loss of montane and boreal grassland ( ''robust evidence'' , ''high agreement'' ) whereas tropical tree lines have been generally stable ( ''medium evidence'' , ''medium agreement'' ) ( [[#Harsch--2009|Harsch et al., 2009]] ; [[#Rehm--2015|Rehm and Feeley, 2015]] ; [[#Silva--2016|Silva et al., 2016]] ; [[#Andela--2017|Andela et al., 2017]] ; [[#Song--2018|Song et al., 2018]] ; [[#Aide--2019|Aide et al., 2019]] ; [[#Gibson--2019|Gibson and Newman, 2019]] ). The Eurasian steppes experienced a 1% increase in woody cover per decade since 2000 ( [[#Liu--2021|Liu et al., 2021]] ) and inner Mongolian grasslands in China experienced broad encroachment as well ( [[#Chen--2015|Chen et al., 2015]] ). Climatic drivers of woody expansion in temperature-limited grasslands, particularly alpine grasslands, are most frequently attributed to warming ( ''robust evidence'' , ''high agreement'' , ''high confidence'' ) ( [[#D’Odorico--2012|D’Odorico et al., 2012]] ; [[#Hagedorn--2014|Hagedorn et al., 2014]] ), an increase in water and nutrient availability from thawing permafrost ( ''medium evidence'' , ''high agreement'' ) ( [[#Zhou--2015b|Zhou et al., 2015b]] ; [[#Silva--2016|Silva et al., 2016]] ) and rising CO 2 ( ''medium evidence'' , ''medium agreement'' ) ( [[#Frank--2015|Frank et al., 2015]] ; [[#Aide--2019|Aide et al., 2019]] ). Interactions of LULCCs such as land abandonment, grazing management shifts and fire suppression with climate change are contributing factors ( [[#Liu--2021|Liu et al., 2021]] )

Remote sensing shows overall increasing trends in both the annual maximum Normalized Difference Vegetation Index (NDVI) and annual mean NDVI in global grassland ecosystems between 1982 and 2011 ( [[#Gao--2016|Gao et al., 2016]] ). Multiple lines of evidence indicate that changes in grassland productivity are positively correlated with increases in mean annual precipitation ( [[#Hoover--2014|Hoover et al., 2014]] ; [[#Brookshire--2015|Brookshire and Weaver, 2015]] ; [[#Gang--2015|Gang et al., 2015]] ; [[#Gao--2016|Gao et al., 2016]] ; [[#Wilcox--2017|Wilcox et al., 2017]] ; [[#Wan--2018|Wan et al., 2018]] ). Increasing temperatures positively impact grassland production and biomass, especially in temperature-limited regions ( [[#Piao--2014|Piao et al., 2014]] ; [[#Gao--2016|Gao et al., 2016]] ). However, it is expected that grasslands in hot areas will decrease production as temperatures increase ( ''limited evidence'' , ''low agreement'' ) ( [[#Gang--2015|Gang et al., 2015]] ) ''.'' Nevertheless, grassland responses to warming and drought are being ameliorated by increasing CO 2 and associated improved water-use efficiency ( [[#Roy--2016|Roy et al., 2016]] ). For example, in a cool temperate grassland experiment, warming led to a longer growing season and elevated CO 2 further extended growing by conserving water, which enabled most species to remain active longer ( ''medium evidence'' , ''medium agreement'' ) ( [[#Reyes-Fox--2014|Reyes-Fox et al., 2014]] ).

<div id="2.4.3.6" class="h3-container"></div>

<span id="observed-changes-in-tropical-forest"></span>
==== 2.4.3.6 Observed Changes in Tropical Forest ====

<div id="h3-20-siblings" class="h3-siblings"></div>

Overall declines of tropical forest cover ( [[#Kohl--2015|Kohl et al., 2015]] ; [[#Liu--2015|Liu et al., 2015]] ; [[#Baccini--2017|Baccini et al., 2017]] ; [[#Harris--2021|Harris et al., 2021]] ), with declines more than triple the gains ( [[#Harris--2021|Harris et al., 2021]] ) have been driven primarily by deforestation and land conversion ( ''robust evidence'' , ''high agreement'' ) ( [[#Lewis--2015|Lewis et al., 2015]] ; [[#Curtis--2018|Curtis et al., 2018]] ; [[#Assis--2019|Assis et al., 2019]] ). In opposition to this general trend, expansion of tropical forest cover into savannas and grasslands has occurred in Africa, South America and Australia ( [[#Marimon--2014|Marimon et al., 2014]] ; [[#Baccini--2017|Baccini et al., 2017]] ; [[#Ondei--2017|Ondei et al., 2017]] ; [[#Stevens--2017|Stevens et al., 2017]] ; Aleman et al., 2018; [[#Staver--2018|Staver, 2018]] ; [[#Rosan--2019|Rosan et al., 2019]] ).

Specific examples of climate change-driven range shifts of tropical deciduous forests upslope into alpine grasslands have been documented in the Americas ( [[#Chacón-Moreno--2021|Chacón-Moreno et al., 2021]] ; [[#Jiménez-García--2021|Jiménez-García et al., 2021]] ) and Asia ( [[#Sigdel--2018|Sigdel et al., 2018]] ). However, tree line behaviours are diverse. A study in Nepal recorded that the tree line fomed by ''Abies spectabilis'' had been stable for more than a century, while the upper limit of large shrubs ( ''Rhododendron campanulatum'' ) had been advancing ( [[#Mainali--2020|Mainali et al., 2020]] ). In both the Andes ( [[#Harsch--2009|Harsch et al., 2009]] ) and Himalayas ( [[#Singh--2021|Singh et al., 2021]] ), most tree lines have been stable, leading ( [[#Rehm--2015|Rehm and Feeley, 2015]] ) to postulate a ‘grass ceiling’ that has been difficult for trees to penetrate. The tree line shifts that have occurred are probably driven by interactions between changing land use (e.g., fire suppression) and climate changes such as increased rainfall, warming and elevated CO 2 (via CO 2 fertilisation or increases in water-use efficiency) ( ''medium evidence'' , ''medium agreement'' ) ( [[#Cernusak--2013|Cernusak et al., 2013]] ; [[#Huang--2013|Huang et al., 2013]] ; [[#Van%20Der%20Sleen--2015|Van Der Sleen et al., 2015]] ; [[#Yang--2016|Yang et al., 2016]] ).

Increases in productivity of tropical forests ( [[#Gatti--2014|Gatti et al., 2014]] ; [[#Brienen--2015|Brienen et al., 2015]] ; [[#Baccini--2017|Baccini et al., 2017]] ), Africa and southeast Asia ( [[#Qie--2017|Qie et al., 2017]] ) have been attributed to elevated CO 2 ( ''robust evidence'' , ''medium agreement'' ) ( [[#Ballantyne--2012|Ballantyne et al., 2012]] ; [[#Brienen--2015|Brienen et al., 2015]] ; [[#Sitch--2015|Sitch et al., 2015]] ; [[#Yang--2016|Yang et al., 2016]] ; [[#Mitchard--2018|Mitchard, 2018]] ). The rates of these increases have been slowing down in the central Amazon ( [[#Brienen--2015|Brienen et al., 2015]] ; [[#de%20Meira%20Junior--2020|de Meira Junior et al., 2020]] ) and Southeast Asia ( [[#Qie--2017|Qie et al., 2017]] ). In contrast, the carbon sink (and hence the rate of biomass gain) in intact African forests was stable until 2010 and has only recently started to decline, indicating asynchronous carbon sink saturation in Amazonia and Africa, the difference being driven by rates of tree mortality ( [[#Hubau--2020|Hubau et al., 2020]] ). At the global level, [[#Hubau--2020|Hubau et al. (2020)]] argue that the carbon sink associated with intact tropical forests peaked in the 1990s and is now in decline.

Declines in productivity are most strongly associated with warming ( [[#Sullivan--2020|Sullivan et al., 2020]] ), reduced growth rates during droughts ( [[#Bennett--2015|Bennett et al., 2015]] ; [[#Bonai--2016|Bonai et al., 2016]] ; [[#Corlett--2016|Corlett, 2016]] ), drought-related mortality ( [[#Brando--2014|Brando et al., 2014]] ; [[#Zhou--2014|Zhou et al., 2014]] ; [[#Brienen--2015|Brienen et al., 2015]] ; [[#Corlett--2016|Corlett, 2016]] ; [[#McDowell--2018|McDowell et al., 2018]] ), fire ( [[#Liu--2017|Liu et al., 2017]] ) and cloud-induced radiation limitation ( ''robust evidence'' , ''high agreement'' ) ( [[#Deb%20Burman--2020|Deb Burman et al., 2020]] ) ''.'' Increases in the frequency and severity of droughts and shorter tree residence times due to increases in growth rates caused by elevated CO 2 may be additional interactive factors increasing tree mortality ( [[#Malhi--2014|Malhi et al., 2014]] ; [[#Brienen--2015|Brienen et al., 2015]] ). Vulnerability to drought varies between tree species and sizes, with large, older trees at the highest risk of mortality ( [[#McDowell--2018|McDowell et al., 2018]] ; [[#Meakem--2018|Meakem et al., 2018]] ). Mortality risk also varies between forest types, with seasonal rainforests appearing to be the most vulnerable to drought ( [[#Corlett--2016|Corlett, 2016]] ).

Lianas (long-stemmed woody vines) generally negatively impact trees, significantly reducing the growth of heavily infested trees ( [[#Reis--2020|Reis et al., 2020]] ). Lianas would benefit from climate change and disturbance ( [[#LingZi--2014|LingZi et al., 2014]] ; [[#Hodgkins--2018|Hodgkins et al., 2018]] ). The extent of their suitable niche can increase ( [[#Taylor--2016|Taylor and Kumar, 2016]] ), thereby decreasing forest biomass accumulation ( ''robust evidence'' , ''high agreement'' ) ( [[#van%20der%20Heijden--2013|van der Heijden et al., 2013]] ; [[#Fauset--2015|Fauset et al., 2015]] ; [[#Estrada-Villegas--2020|Estrada-Villegas et al., 2020]] ).

Climate change continues to degrade forests by reducing resilience to pests and diseases, increasing species invasion, facilitating pathogen spread ( [[#Malhi--2014|Malhi et al., 2014]] ; [[#Deb--2018|Deb et al., 2018]] ) and intensifying fire risk and potential dieback ( [[#Lapola--2018|Lapola et al., 2018]] ; [[#Marengo--2018|Marengo et al., 2018]] ). Drought, temperature increases and forest fragmentation interact to increase the prevalence of fires in tropical forests ( ''robust evidence'' , ''high agreement'' ). Warming increases water stress in trees ( [[#Corlett--2016|Corlett, 2016]] ) and, together with forest fragmentation, dramatically increases the desiccation of forest canopies—resulting in deforestation that then leads to even hotter and drier regional climates ( [[#Malhi--2014|Malhi et al., 2014]] ; [[#Lewis--2015|Lewis et al., 2015]] ). Warming and drought increase the invasion of grasses into forest edges and increase fire risk ( ''robust evidence'' , ''high agreement'' ) ( [[#Brando--2014|Brando et al., 2014]] ; [[#Balch--2015|Balch et al., 2015]] ; [[#Lewis--2015|Lewis et al., 2015]] ). Droughts and fires additively increase mortality and, consequently, reduce canopy cover and above-ground biomass (Cross-Chapter Paper 7) ( [[#Brando--2014|Brando et al., 2014]] , 2020; [[#Balch--2015|Balch et al., 2015]] ; [[#Lewis--2015|Lewis et al., 2015]] ).

<div id="2.4.3.7" class="h3-container"></div>

<span id="observed-changes-in-boreal-and-temperate-forests"></span>
==== 2.4.3.7 Observed Changes in Boreal and Temperate Forests ====

<div id="h3-21-siblings" class="h3-siblings"></div>

The AR5 found increased tree mortality, wildfire and plant phenology changes in boreal and temperate forests ( [[#Settele--2014|Settele et al., 2014]] ). Expanding on these conclusions, this assessment, using analyses of causal factors, attributes the following observed changes in boreal and temperate forests in the 20th and 21st centuries to anthropogenic climate change: upslope and poleward biome shifts at sites in Asia, Europe and North America ( [[#2.4.3.2.1|Section 2.4.3.2.1]] ); range shifts of plants ( [[#2.4.2.1|Section 2.4.2.1]] ); earlier blooming and leafing of plants ( [[#2.4.2.4|Section 2.4.2.4]] ); poleward shifts in tree-feeding insects ( [[#2.4.2.1|Section 2.4.2.1]] ); increases in insect pest outbreaks ( [[#2.4.4.3.3|Section 2.4.4.3.3]] ); increases in the area burned by wildfire in western North America ( [[#2.4.4.2.1|Section 2.4.4.2.1]] ); increased drought-induced tree mortality in western North America ( [[#2.4.4.3.1|Section 2.4.4.3.1]] ); and thawing of the permafrost that underlies extensive areas of boreal forest ( [[#2.4.3.9|Section 2.4.3.9]] )( [[#2.3|Section 2.3.2.5]] in ( [[#Gulev--2021|Gulev et al., 2021]] )). Atmospheric CO 2 from anthropogenic sources has also increased net primary productivity (NPP) ( [[#2.4.4.5.1|Section 2.4.4.5.1]] ). In summary, anthropogenic climate change has caused substantial changes in temperate and boreal forest ecosystems, including biome shifts and increases in wildfire, insect pest outbreaks and tree mortality, at a global mean surface temperature (GMST) increase of 0.9°C above the pre-industrial period ( ''robust evidence'' , ''high agreement'' ).

Other changes detected in boreal forests and consistent with, but not formally attributed to, climate change, include increased wildfire in Siberia ( [[#2.4.4.2.3|Section 2.4.4.2.3]] ), long-lasting smouldering below-ground fires in Canada and the USA ( [[#Scholten--2021|Scholten et al., 2021]] ), tree mortality in Europe ( [[#2.4.4.3.3|Section 2.4.4.3.3]] ) and post-fire shifts of boreal conifer to deciduous broadleaf tree species in Alaska ( [[#Mack--2021|Mack et al., 2021]] ). From 1930 to 1960, boreal forest growth became limited more by precipitation than temperature in the Northern Hemisphere ( [[#Babst--2019|Babst et al., 2019]] ).

For some vegetation, changes in land use and management have exerted more influence than climate change. These include upslope and poleward forest shifts in Europe following the abandonment of timber harvesting or livestock grazing ( [[#2.4.3.2.2|Section 2.4.3.2.2]] ), changes in wildfire in Europe affected by fire suppression, fire prevention and agricultural abandonment ( [[#2.4.4.2.3|Section 2.4.4.2.3]] ), and forest species composition changes in Scotland due to nitrogen deposition from air pollution ( [[#Hester--2019|Hester et al., 2019]] ). Remote sensing suggests that the area of temperate and boreal forests increased in Asia and Europe between 1982 and 2016 ( [[#Song--2018|Song et al., 2018]] ) and in Canada between 1984 and 2015 ( [[#Guindon--2018|Guindon et al., 2018]] ), but forest plantations and regrowth are probable drivers ( [[#Song--2018|Song et al., 2018]] ).

<div id="2.4.3.8" class="h3-container"></div>

<span id="observed-changes-in-peatlands"></span>
==== 2.4.3.8 Observed Changes in Peatlands ====

<div id="h3-22-siblings" class="h3-siblings"></div>

Globally, peatland ecosystems store approximately 25% (600 ± 100 GtC) of the world’s soil organic carbon ( [[#Yu--2010|Yu et al., 2010]] ; [[#Page--2011|Page et al., 2011]] ; [[#Hugelius--2020|Hugelius et al., 2020]] ) and 10% of the world’s freshwater resources ( [[#Joosten--2002|Joosten and Clarke, 2002]] ), despite only occupying 3% of the global land area ( [[#Xu--2018a|Xu et al., 2018a]] ). The long-term role of northern peatlands in the carbon cycle was mentioned for the first time in IPCC AR4 ( [[#IPCC--2007|IPCC, 2007]] ), while SR1.5 briefly mentioned the combined effects of changes in climate and land use on peatlands ( [[#IPCC--2018b|IPCC, 2018b]] ). New evidence confirms that climate change, including extreme weather events (e.g., droughts; [[IPCC:Wg2:Chapter:Chapter-8#8.3.1|Section 8.3.1.6]] ), permafrost degradation ( [[#2.3|Section 2.3.2.5]] ), SLR ( [[#2.3.3.3|Section 2.3.3.3]] ) and fire ( [[IPCC:Wg2:Chapter:Chapter-5#5.4.3.2|Section 5.4.3.2]] ) ( [[#Henman--2008|Henman and Poulter, 2008]] ; [[#Kirwan--2012|Kirwan and Mudd, 2012]] ; [[#Turetsky--2015|Turetsky et al., 2015]] ; [[#Page--2016|Page and Hooijer, 2016]] ; [[#Swindles--2019|Swindles et al., 2019]] ; [[#Hoyt--2020|Hoyt et al., 2020]] ; [[#Hugelius--2020|Hugelius et al., 2020]] ; [[#Jovani-Sancho--2021|Jovani-Sancho et al., 2021]] ; [[#Veraverbeke--2021|Veraverbeke et al., 2021]] ), superimposed on anthropogenic disturbances (e.g., draining for agriculture or mining; [[IPCC:Wg2:Chapter:Chapter-5#5.2.1|Section 5.2.1.1]] ), has led to rapid losses of peatland carbon across the world ( ''robust evidence'' , ''high agreement'' ) ( [[#Page--2011|Page et al., 2011]] ; [[#Leifeld--2019|Leifeld et al., 2019]] ; [[#Hoyt--2020|Hoyt et al., 2020]] ; [[#Turetsky--2020|Turetsky et al., 2020]] ; [[#Loisel--2021|Loisel et al., 2021]] ). Other essential peatland ecosystem services, such as water storage and biodiversity, are also being lost worldwide ( ''robust evidence'' , ''high agreement'' ) ( [[#Bonn--2014|Bonn et al., 2014]] ; [[#Martin-Ortega--2014|Martin-Ortega et al., 2014]] ; [[#Tiemeyer--2017|Tiemeyer et al., 2017]] ).

The switch from carbon sink to carbon source in peatlands globally is mainly attributable to changes in the depth of the water table, regardless of management or status ( ''robust evidence'' , ''high agreement'' ) ( [[#Lafleur--2005|Lafleur et al., 2005]] ; [[#Dommain--2011|Dommain et al., 2011]] ; [[#Lund--2012|Lund et al., 2012]] ; [[#Cobb--2017|Cobb et al., 2017]] ; [[#Evans--2021|Evans et al., 2021]] ; [[#Novita--2021|Novita et al., 2021]] ). Across the temperate and tropical biomes, extensive drainage and deforestation have caused widespread water table draw-downs and/or peat subsidence, as well as high CO 2 emissions ( ''medium evidence'' , ''high agreement'' ). Climate change is compounding these impacts ( ''medium evidence'' , ''medium agreement'' ). For example, in Indonesia, the highest emissions from drained tropical peatlands were reported in the extremely dry year of the 1997 El Niño (810–2570 TgC yr -1 ) ( [[#Page--2002|Page et al., 2002]] ) and the 2015 fire season (380 TgC yr -1 ) ( [[#Field--2016|Field et al., 2016]] ). These prolonged dry seasons have also led to tree die-offs and fires, which are relatively new phenomena at these latitudes ( ''medium evidence'' , ''high agreement'' ) ( [[#Cole--2015|Cole et al., 2015]] ; [[#Mezbahuddin--2015|Mezbahuddin et al., 2015]] ; [[#Fanin--2017|Fanin and van der Werf, 2017]] ; [[#Taufik--2017|Taufik et al., 2017]] ; [[#Cole--2019|Cole et al., 2019]] ). Low soil moisture contributes to increased fire propagation ( [[IPCC:Wg2:Chapter:Chapter-12#12.4|Section 12.4.2.2]] ) ( [[#Dadap--2019|Dadap et al., 2019]] ; [[#Canadell--2021|Canadell et al., 2021]] ), causing long-lasting fires responsible for smoke and haze pollution ( ''robust evidence'' , ''high agreement'' ) ( [[#Ballhorn--2009|Ballhorn et al., 2009]] ; [[#Page--2009|Page et al., 2009]] ; [[#Gaveau--2014|Gaveau et al., 2014]] ; [[#Huijnen--2016|Huijnen et al., 2016]] ; [[#Page--2016|Page and Hooijer, 2016]] ; [[#Hu--2018|Hu et al., 2018]] ; [[#Vadrevu--2019|Vadrevu et al., 2019]] ; [[#Niwa--2021|Niwa et al., 2021]] ). Increases in fires and smoke lead to habitat loss and negatively impact regional faunal populations ( ''limited evidence'' , ''high agreement'' ) ( [[#Neoh--2015|Neoh et al., 2015]] ; [[#Erb--2018b|Erb et al., 2018b]] ; [[#Thornton--2018|Thornton et al., 2018]] ).

In large, lowland tropical peatland basins that are less impacted by anthropogenic activities (i.e., the Amazon and Congo river basins), the direct impact of climate change is that of a decreased carbon sink ( ''limited evidence'' , ''medium agreement'' ) ( [[#Roucoux--2013|Roucoux et al., 2013]] ; [[#Gallego-Sala--2018|Gallego-Sala et al., 2018]] ; [[#Wang--2018a|Wang et al., 2018a]] ; [[#Dargie--2019|Dargie et al., 2019]] ; [[#Ribeiro--2021|Ribeiro et al., 2021]] ). As for the temperate and boreal regions, climatic drying also tends to promote peat oxidation and carbon loss to the atmosphere ( ''medium evidence'' , ''medium agreement'' ) ( [[#2.3.1|Section 2.3.1.3.4]] ) ( [[#Helbig--2020|Helbig et al., 2020]] ; [[#Zhang--2020|Zhang et al., 2020]] ). In Europe, increasing mean annual temperatures in the Baltic, Scandinavia, and continental Europe ( [[IPCC:Wg2:Chapter:Chapter-12#12.4|Section 12.4.5.1]] ) have led to widespread lowering of peatland water tables at intact sites ( [[#Swindles--2019|Swindles et al., 2019]] ), desiccation and die-off of sphagnum moss ( [[#Bragazza--2008|Bragazza, 2008]] ; [[#Lees--2019|Lees et al., 2019]] ) and increased intensity and frequency of fires, resulting in a rapid carbon loss ( [[#Davies--2013|Davies et al., 2013]] ; [[#Veraverbeke--2021|Veraverbeke et al., 2021]] ). Nevertheless, longer growing seasons and warmer, wetter climates have increased carbon accumulation and promoted thick deposits regionally, as reported for some North American sites ( ''limited evidence'' , ''medium agreement'' ) ( [[#Cai--2011|Cai and Yu, 2011]] ; [[#Shiller--2014|Shiller et al., 2014]] ; [[#Ott--2016|Ott and Chimner, 2016]] ).

In high-latitude peatlands, the net effect of climate change on the permafrost peatland carbon sink capacity remains uncertain ( [[#Abbott--2016|Abbott et al., 2016]] ; [[#McGuire--2018b|McGuire et al., 2018b]] ; [[#Laamrani--2020|Laamrani et al., 2020]] ; [[#Loisel--2021|Loisel et al., 2021]] ; [[#Sim--2021|Sim et al., 2021]] ; [[#Väliranta--2021|Väliranta et al., 2021]] ). Increasing air temperatures have been linked to permafrost degradation and altered hydrological regimes (2.3.3.2; Figure 2.4a; 2.4.3.9; Box 5.1), which have led to rapid changes in plant communities and bio-geochemical cycling ( ''robust evidence'' , ''high agreement'' ) ( [[#Liljedahl--2016|Liljedahl et al., 2016]] ; [[#Swindles--2016|Swindles et al., 2016]] ; [[#Voigt--2017|Voigt et al., 2017]] ; [[#Zhang--2017b|Zhang et al., 2017b]] ; [[#Voigt--2020|Voigt et al., 2020]] ; [[#Sim--2021|Sim et al., 2021]] ). In many instances, permafrost degradation triggers thermokarst land subsidence associated with local wetting ( ''robust evidence'' , ''high agreement'' ) ( [[#Jones--2013|Jones et al., 2013]] ; [[#Borge--2017|Borge et al., 2017]] ; [[#Olvmo--2020|Olvmo et al., 2020]] ; [[#Olefeldt--2021|Olefeldt et al., 2021]] ). Permafrost thaw in peatland-rich landscapes can also cause local drying through increased hydrological connectivity and runoff ( [[#Connon--2014|Connon et al., 2014]] ). In the first decades following thaw, increases in methane, CO 2 and nitrous oxide emissions have been recorded from peatland sites, depending on surface moisture conditions ( [[#Schuur--2009|Schuur et al., 2009]] ; [[#O’Donnell--2012|O’Donnell et al., 2012]] ; [[#Elberling--2013|Elberling et al., 2013]] ; [[#Matveev--2016|Matveev et al., 2016]] ; [[#Euskirchen--2020|Euskirchen et al., 2020]] ; [[#Hugelius--2020|Hugelius et al., 2020]] ). Conversely, some evidence suggests increased peat accumulation after thaw ( [[#Jones--2013|Jones et al., 2013]] ; [[#Estop-Aragonés--2018|Estop-Aragonés et al., 2018]] ; [[#Väliranta--2021|Väliranta et al., 2021]] ). There is also a need to consider the impact of wildfire on permafrost thaw, due to its effect on soil temperature regime ( [[#Gibson--2018|Gibson et al., 2018]] ), as fire intensity and frequency have increased across the boreal and Arctic biomes ( ''limited evidence'' , ''high agreement'' ) ( [[#Kasischke--2010|Kasischke et al., 2010]] ; [[#Scholten--2021|Scholten et al., 2021]] ).

The CO 2 emissions from degrading peatlands is contributing to climate change in a positive feedback loop ( ''robust evidence'' , ''high agreement)'' . At mid-latitudes, widespread anthropogenic disturbance led to large historical GHG emissions and current legacy emissions of 0.15 PgC yr -1 between 1990 and 2000 ( ''limited evidence'' , ''high agreement'' ) ( [[#Maljanen--2010|Maljanen et al., 2010]] ; [[#Tiemeyer--2016|Tiemeyer et al., 2016]] ; [[#Drexler--2018|Drexler et al., 2018]] ; [[#Qiu--2021|Qiu et al., 2021]] ). About 80 million hectares of peatland have been converted to agriculture, equivalent to 72 PgC emissions in 850–2010 CE ( [[#Leifeld--2019|Leifeld et al., 2019]] ; [[#Qiu--2021|Qiu et al., 2021]] ). In Southeast Asia (SEA), an estimated 20–25 Mha of peatlands have been converted to agriculture with carbon currently being lost at a rate of ~155 ± 30 MtC yr −1 ( [[#Miettinen--2016|Miettinen et al., 2016]] ; [[#Leifeld--2019|Leifeld et al., 2019]] ; [[#Hoyt--2020|Hoyt et al., 2020]] ). Extensive deforestation and drainage have caused widespread peat subsidence and large CO 2 emissions at a current average of ~10 ± 2 tonnes ha -1 yr -1 , excluding fires ( [[#Hoyt--2020|Hoyt et al., 2020]] ), with values estimated from point subsidence measurements being as high as 30–90 tonnes CO 2 ha −1 yr −1 locally ( ''robust evidence'' , ''high agreement'' ) ( [[#Wösten--1997|Wösten et al., 1997]] ; [[#Matysek--2018|Matysek et al., 2018]] ; [[#Swails--2018|Swails et al., 2018]] ; [[#Evans--2019|Evans et al., 2019]] ; [[#Conchedda--2020|Conchedda and Tubiello, 2020]] ; [[#Anshari--2021|Anshari et al., 2021]] ). On average, at the global scale, increases in GHG emissions from peatlands have primarily come from the compounded effects of LUC, drought and fire, with additional emissions from some thawing-permafrost peatlands ( ''robust evidence'' , ''high agreement'' ).

<div id="2.4.3.9" class="h3-container"></div>

<span id="observed-changes-in-polar-tundra"></span>
==== 2.4.3.9 Observed Changes in Polar Tundra ====

<div id="h3-23-siblings" class="h3-siblings"></div>

Warming at high latitudes, documented in both AR4 and AR5, is leading to earlier snow and sea ice melt and longer growing seasons ( [[#IPCC--2021a|IPCC, 2021a]] ) which are continuing to alter tundra plant communities ( ''medium evidence'' , ''high agreement'' ) ( [[#Post--2009|Post et al., 2009]] ; [[#Gauthier--2013|Gauthier et al., 2013]] ). Woody encroachment and increases in vegetation productivity, observed in both AR4 and AR5, are widespread and continuing. Both experiments and monitoring indicate that climate warming is causing increases in shrub, grass and sedge abundance, density, frequency, and height, with decreases in mosses and/or lichens ( ''robust evidence'' , ''high agreement'' ) ( [[#Myers-Smith--2011|Myers-Smith et al., 2011]] ; [[#Bjorkman--2018|Bjorkman et al., 2018]] ; [[#Bjorkman--2019|Bjorkman et al., 2019]] ) ''.'' Shrub growth is climate-sensitive and is greater in years with warmer growing seasons ( [[#Myers-Smith--2015|Myers-]] [[#Smith--2015|Smith et al., 2015]] ). Plant species that prefer warmer conditions are increasing ( [[#Elmendorf--2015|Elmendorf et al., 2015]] ; [[#Bjorkman--2018|Bjorkman et al., 2018]] ), plant cover is increasing and bare ground is decreasing in long-term monitoring plots ( [[#Bjorkman--2019|Bjorkman et al., 2019]] ; [[#Myers-Smith--2019|Myers-Smith et al., 2019]] ). Animals such as moose, beavers and songbirds may already be responding to these vegetation changes by expanding their ranges northward or upslope into shrub tundra ( [[#Boelman--2015|Boelman et al., 2015]] ; [[#Tape--2016a|Tape et al., 2016a]] ; [[#Tape--2016b|Tape et al., 2016b]] ; [[#Tape--2018|Tape et al., 2018]] ).

In addition to direct warming, indirect effects of climate change, first found in AR4 and AR5, continue, such as thawed permafrost, altered hydrology and enhanced nutrient cycling, and these processes are causing pronounced vegetation changes ( ''medium evidence'' , ''medium agreement'' ) ( [[#Schuur--2009|Schuur et al., 2009]] ; [[#Natali--2012|Natali et al., 2012]] ). Soil moisture status influences temperature sensitivity of plant growth and canopy heights ( [[#Myers-Smith--2015|Myers-]] [[#Smith--2015|Smith et al., 2015]] ; [[#Ackerman--2017|Ackerman et al., 2017]] ; [[#Bjorkman--2018|Bjorkman et al., 2018]] ). In tundra ecosystems, permafrost thawing can decouple below-ground plant growth dynamics from above-ground dynamics, with below-ground root growth continuing until soils re-freeze in autumn (Cross-Chapter Paper 6) ( [[#Iversen--2015|Iversen et al., 2015]] ; [[#Blume-Werry--2016|Blume-Werry et al., 2016]] ; [[#Radville--2016|Radville et al., 2016]] ).

<div id="2.4.4" class="h2-container"></div>

<span id="observed-changes-in-ecosystem-processes-and-services"></span>